diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,70021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 11666,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.714285714285715e-08,
+      "loss": 0.5529,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.142857142857143e-07,
+      "loss": 0.5725,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7142857142857146e-07,
+      "loss": 0.4912,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.285714285714286e-07,
+      "loss": 0.5575,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428575e-07,
+      "loss": 0.5187,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.428571428571429e-07,
+      "loss": 0.6355,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.5984,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.571428571428572e-07,
+      "loss": 0.6348,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.142857142857143e-07,
+      "loss": 0.5369,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 0.5762,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.285714285714287e-07,
+      "loss": 0.5363,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.857142857142858e-07,
+      "loss": 0.554,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.428571428571429e-07,
+      "loss": 0.6045,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.6339,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.571428571428572e-07,
+      "loss": 0.5054,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.142857142857144e-07,
+      "loss": 0.5938,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.714285714285715e-07,
+      "loss": 0.5323,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0285714285714286e-06,
+      "loss": 0.5791,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0857142857142858e-06,
+      "loss": 0.5463,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 0.5763,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.5176,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2571428571428573e-06,
+      "loss": 0.5574,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3142857142857143e-06,
+      "loss": 0.5089,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3714285714285717e-06,
+      "loss": 0.5031,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 0.5227,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4857142857142858e-06,
+      "loss": 0.4615,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.542857142857143e-06,
+      "loss": 0.4647,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.4697,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.657142857142857e-06,
+      "loss": 0.4701,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 0.4543,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7714285714285714e-06,
+      "loss": 0.4435,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8285714285714288e-06,
+      "loss": 0.447,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.885714285714286e-06,
+      "loss": 0.4642,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.942857142857143e-06,
+      "loss": 0.4464,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.5004,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0571428571428573e-06,
+      "loss": 0.442,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1142857142857147e-06,
+      "loss": 0.4708,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1714285714285716e-06,
+      "loss": 0.4746,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.228571428571429e-06,
+      "loss": 0.4735,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 0.469,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.342857142857143e-06,
+      "loss": 0.6212,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.4497,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4571428571428573e-06,
+      "loss": 0.4344,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5142857142857147e-06,
+      "loss": 0.3882,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.571428571428571e-06,
+      "loss": 0.4233,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6285714285714286e-06,
+      "loss": 0.4608,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.685714285714286e-06,
+      "loss": 0.4189,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7428571428571433e-06,
+      "loss": 0.4578,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.3942,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.3894,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9142857142857146e-06,
+      "loss": 0.4807,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9714285714285716e-06,
+      "loss": 0.5984,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.028571428571429e-06,
+      "loss": 0.4355,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.085714285714286e-06,
+      "loss": 0.4977,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.142857142857143e-06,
+      "loss": 0.4062,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.4528,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2571428571428577e-06,
+      "loss": 0.4639,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.314285714285714e-06,
+      "loss": 0.3911,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3714285714285716e-06,
+      "loss": 0.4202,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 0.3906,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4857142857142863e-06,
+      "loss": 0.4504,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.542857142857143e-06,
+      "loss": 0.4298,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.3845,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6571428571428576e-06,
+      "loss": 0.418,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7142857142857146e-06,
+      "loss": 0.4547,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.771428571428572e-06,
+      "loss": 0.4591,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.828571428571429e-06,
+      "loss": 0.3833,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.885714285714286e-06,
+      "loss": 0.421,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.942857142857143e-06,
+      "loss": 0.4047,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.448,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.057142857142858e-06,
+      "loss": 0.4734,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.114285714285715e-06,
+      "loss": 0.4569,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1714285714285715e-06,
+      "loss": 0.4249,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.228571428571429e-06,
+      "loss": 0.3678,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.437,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.342857142857143e-06,
+      "loss": 0.3923,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4e-06,
+      "loss": 0.4132,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.457142857142858e-06,
+      "loss": 0.4335,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.514285714285714e-06,
+      "loss": 0.4343,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 0.423,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.628571428571429e-06,
+      "loss": 0.4707,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.685714285714286e-06,
+      "loss": 0.4405,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.742857142857144e-06,
+      "loss": 0.3959,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.4052,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.857142857142858e-06,
+      "loss": 0.4122,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9142857142857145e-06,
+      "loss": 0.4609,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.971428571428572e-06,
+      "loss": 0.3727,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.028571428571429e-06,
+      "loss": 0.366,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.085714285714286e-06,
+      "loss": 0.3898,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.142857142857142e-06,
+      "loss": 0.4215,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2e-06,
+      "loss": 0.3989,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.257142857142857e-06,
+      "loss": 0.4144,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.314285714285715e-06,
+      "loss": 0.4241,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.371428571428572e-06,
+      "loss": 0.3887,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.428571428571429e-06,
+      "loss": 0.3746,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.485714285714287e-06,
+      "loss": 0.4066,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.542857142857143e-06,
+      "loss": 0.4607,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.3784,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6571428571428576e-06,
+      "loss": 0.4544,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.4081,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.771428571428572e-06,
+      "loss": 0.3824,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.828571428571429e-06,
+      "loss": 0.3564,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.885714285714285e-06,
+      "loss": 0.4146,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.942857142857143e-06,
+      "loss": 0.4246,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 0.3865,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.057142857142858e-06,
+      "loss": 0.4225,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.114285714285715e-06,
+      "loss": 0.3798,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.171428571428572e-06,
+      "loss": 0.4398,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.22857142857143e-06,
+      "loss": 0.4353,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.285714285714286e-06,
+      "loss": 0.4125,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.342857142857143e-06,
+      "loss": 0.3684,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.3872,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4571428571428575e-06,
+      "loss": 0.4212,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.514285714285715e-06,
+      "loss": 0.3522,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.571428571428572e-06,
+      "loss": 0.351,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.628571428571428e-06,
+      "loss": 0.3942,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.685714285714286e-06,
+      "loss": 0.3751,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.742857142857143e-06,
+      "loss": 0.3584,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.4096,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 0.4347,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.914285714285715e-06,
+      "loss": 0.3983,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.971428571428573e-06,
+      "loss": 0.397,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.028571428571429e-06,
+      "loss": 0.376,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.085714285714286e-06,
+      "loss": 0.3793,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.3593,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.399,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.257142857142858e-06,
+      "loss": 0.4334,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.314285714285715e-06,
+      "loss": 0.3678,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.371428571428571e-06,
+      "loss": 0.389,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 0.4114,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.485714285714286e-06,
+      "loss": 0.3561,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.542857142857144e-06,
+      "loss": 0.379,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.41,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.657142857142858e-06,
+      "loss": 0.3839,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.714285714285716e-06,
+      "loss": 0.4163,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.771428571428572e-06,
+      "loss": 0.3829,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.828571428571428e-06,
+      "loss": 0.3513,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.885714285714286e-06,
+      "loss": 0.4048,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.942857142857144e-06,
+      "loss": 0.3633,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.5511,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.057142857142857e-06,
+      "loss": 0.4023,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.114285714285715e-06,
+      "loss": 0.4149,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.171428571428573e-06,
+      "loss": 0.4139,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.22857142857143e-06,
+      "loss": 0.3245,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.285714285714287e-06,
+      "loss": 0.3558,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.342857142857143e-06,
+      "loss": 0.384,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.3575,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.457142857142859e-06,
+      "loss": 0.3829,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.514285714285715e-06,
+      "loss": 0.3997,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.3687,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.628571428571429e-06,
+      "loss": 0.3605,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.685714285714287e-06,
+      "loss": 0.4476,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.742857142857144e-06,
+      "loss": 0.3475,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-06,
+      "loss": 0.3788,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.857142857142858e-06,
+      "loss": 0.4537,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.914285714285716e-06,
+      "loss": 0.4049,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.971428571428572e-06,
+      "loss": 0.4028,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.028571428571428e-06,
+      "loss": 0.3348,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.085714285714286e-06,
+      "loss": 0.3737,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 0.381,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.411,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.257142857142858e-06,
+      "loss": 0.3879,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.314285714285714e-06,
+      "loss": 0.3915,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.371428571428572e-06,
+      "loss": 0.3442,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.42857142857143e-06,
+      "loss": 0.4283,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.485714285714287e-06,
+      "loss": 0.3656,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.542857142857143e-06,
+      "loss": 0.3682,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.3803,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.657142857142859e-06,
+      "loss": 0.3868,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 0.3719,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.771428571428571e-06,
+      "loss": 0.3582,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.828571428571429e-06,
+      "loss": 0.3861,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.885714285714287e-06,
+      "loss": 0.3943,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.942857142857145e-06,
+      "loss": 0.4069,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.3989,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0057142857142859e-05,
+      "loss": 0.3892,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0114285714285715e-05,
+      "loss": 0.3906,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0171428571428573e-05,
+      "loss": 0.3853,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.022857142857143e-05,
+      "loss": 0.4231,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0285714285714285e-05,
+      "loss": 0.3798,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0342857142857143e-05,
+      "loss": 0.3433,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.04e-05,
+      "loss": 0.3929,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.045714285714286e-05,
+      "loss": 0.3617,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0514285714285714e-05,
+      "loss": 0.3925,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0571428571428572e-05,
+      "loss": 0.3739,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.062857142857143e-05,
+      "loss": 0.3937,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0685714285714286e-05,
+      "loss": 0.3399,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0742857142857144e-05,
+      "loss": 0.3892,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 0.3416,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 0.3815,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0914285714285716e-05,
+      "loss": 0.3743,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0971428571428573e-05,
+      "loss": 0.3636,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1028571428571428e-05,
+      "loss": 0.4591,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1085714285714286e-05,
+      "loss": 0.3463,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1142857142857143e-05,
+      "loss": 0.3828,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.3506,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1257142857142857e-05,
+      "loss": 0.4203,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1314285714285715e-05,
+      "loss": 0.3699,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1371428571428573e-05,
+      "loss": 0.618,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.3959,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1485714285714287e-05,
+      "loss": 0.3685,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1542857142857145e-05,
+      "loss": 0.3714,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.16e-05,
+      "loss": 0.3918,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1657142857142859e-05,
+      "loss": 0.3181,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1714285714285716e-05,
+      "loss": 0.3566,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.177142857142857e-05,
+      "loss": 0.33,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1828571428571429e-05,
+      "loss": 0.3928,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1885714285714286e-05,
+      "loss": 0.3738,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1942857142857144e-05,
+      "loss": 0.3224,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 0.3654,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2057142857142858e-05,
+      "loss": 0.3367,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2114285714285716e-05,
+      "loss": 0.3665,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2171428571428572e-05,
+      "loss": 0.3373,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.222857142857143e-05,
+      "loss": 0.3722,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2285714285714288e-05,
+      "loss": 0.5845,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2342857142857144e-05,
+      "loss": 0.3638,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 0.3444,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.245714285714286e-05,
+      "loss": 0.4137,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2514285714285714e-05,
+      "loss": 0.3753,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2571428571428572e-05,
+      "loss": 0.3772,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.262857142857143e-05,
+      "loss": 0.4005,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2685714285714286e-05,
+      "loss": 0.405,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2742857142857143e-05,
+      "loss": 0.3678,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.3552,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 0.3738,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2914285714285715e-05,
+      "loss": 0.388,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2971428571428573e-05,
+      "loss": 0.3687,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.302857142857143e-05,
+      "loss": 0.368,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3085714285714287e-05,
+      "loss": 0.3291,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3142857142857145e-05,
+      "loss": 0.4016,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 0.3512,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3257142857142857e-05,
+      "loss": 0.3287,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3314285714285715e-05,
+      "loss": 0.3435,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3371428571428572e-05,
+      "loss": 0.3379,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3428571428571429e-05,
+      "loss": 0.3394,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3485714285714286e-05,
+      "loss": 0.3784,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3542857142857144e-05,
+      "loss": 0.3846,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.3494,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3657142857142858e-05,
+      "loss": 0.3472,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 0.3765,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3771428571428574e-05,
+      "loss": 0.431,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.382857142857143e-05,
+      "loss": 0.3636,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3885714285714288e-05,
+      "loss": 0.3846,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3942857142857145e-05,
+      "loss": 0.3311,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4e-05,
+      "loss": 0.6819,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4057142857142858e-05,
+      "loss": 0.3956,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4114285714285715e-05,
+      "loss": 0.4042,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4171428571428572e-05,
+      "loss": 0.5765,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.422857142857143e-05,
+      "loss": 0.3557,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.3547,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4342857142857145e-05,
+      "loss": 0.3671,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.3599,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4457142857142859e-05,
+      "loss": 0.3845,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4514285714285717e-05,
+      "loss": 0.3499,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4571428571428573e-05,
+      "loss": 0.3463,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.462857142857143e-05,
+      "loss": 0.4027,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4685714285714288e-05,
+      "loss": 0.3233,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4742857142857143e-05,
+      "loss": 0.3505,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.48e-05,
+      "loss": 0.3649,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 0.3671,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4914285714285715e-05,
+      "loss": 0.3244,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4971428571428572e-05,
+      "loss": 0.3764,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.502857142857143e-05,
+      "loss": 0.4233,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5085714285714288e-05,
+      "loss": 0.3888,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5142857142857144e-05,
+      "loss": 0.3811,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.3345,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.525714285714286e-05,
+      "loss": 0.6316,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5314285714285716e-05,
+      "loss": 0.3324,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5371428571428572e-05,
+      "loss": 0.3956,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.542857142857143e-05,
+      "loss": 0.4175,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5485714285714287e-05,
+      "loss": 0.3496,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5542857142857144e-05,
+      "loss": 0.3684,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 0.3895,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5657142857142856e-05,
+      "loss": 0.3804,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.3207,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.577142857142857e-05,
+      "loss": 0.3481,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.582857142857143e-05,
+      "loss": 0.3585,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5885714285714287e-05,
+      "loss": 0.382,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5942857142857143e-05,
+      "loss": 0.3727,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.4006,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.605714285714286e-05,
+      "loss": 0.3492,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6114285714285715e-05,
+      "loss": 0.38,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6171428571428574e-05,
+      "loss": 0.3618,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.622857142857143e-05,
+      "loss": 0.3693,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6285714285714287e-05,
+      "loss": 0.6064,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6342857142857146e-05,
+      "loss": 0.3769,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.64e-05,
+      "loss": 0.4042,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.645714285714286e-05,
+      "loss": 0.3466,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6514285714285714e-05,
+      "loss": 0.3666,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6571428571428574e-05,
+      "loss": 0.3208,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.662857142857143e-05,
+      "loss": 0.3625,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6685714285714286e-05,
+      "loss": 0.3184,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6742857142857146e-05,
+      "loss": 0.3839,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.3312,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6857142857142858e-05,
+      "loss": 0.3323,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6914285714285717e-05,
+      "loss": 0.3436,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6971428571428574e-05,
+      "loss": 0.345,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.702857142857143e-05,
+      "loss": 0.3371,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.708571428571429e-05,
+      "loss": 0.3867,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.3937,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.72e-05,
+      "loss": 0.3862,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7257142857142857e-05,
+      "loss": 0.366,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7314285714285717e-05,
+      "loss": 0.397,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7371428571428573e-05,
+      "loss": 0.3698,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.742857142857143e-05,
+      "loss": 0.3694,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.748571428571429e-05,
+      "loss": 0.5981,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7542857142857145e-05,
+      "loss": 0.3619,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.76e-05,
+      "loss": 0.326,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.765714285714286e-05,
+      "loss": 0.3652,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7714285714285717e-05,
+      "loss": 0.3784,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7771428571428573e-05,
+      "loss": 0.3403,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7828571428571432e-05,
+      "loss": 0.3816,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7885714285714285e-05,
+      "loss": 0.3508,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7942857142857144e-05,
+      "loss": 0.389,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8e-05,
+      "loss": 0.3719,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8057142857142857e-05,
+      "loss": 0.3411,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8114285714285716e-05,
+      "loss": 0.3459,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8171428571428572e-05,
+      "loss": 0.606,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.822857142857143e-05,
+      "loss": 0.3605,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 0.3584,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8342857142857144e-05,
+      "loss": 0.3464,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.3539,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.845714285714286e-05,
+      "loss": 0.3964,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8514285714285716e-05,
+      "loss": 0.3508,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.3266,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8628571428571428e-05,
+      "loss": 0.3758,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8685714285714287e-05,
+      "loss": 0.3062,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8742857142857143e-05,
+      "loss": 0.4012,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.88e-05,
+      "loss": 0.3679,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 0.337,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8914285714285715e-05,
+      "loss": 0.4494,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8971428571428575e-05,
+      "loss": 0.3542,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.902857142857143e-05,
+      "loss": 0.3492,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9085714285714287e-05,
+      "loss": 0.3257,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9142857142857146e-05,
+      "loss": 0.3489,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.317,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.925714285714286e-05,
+      "loss": 0.3724,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9314285714285718e-05,
+      "loss": 0.3696,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.937142857142857e-05,
+      "loss": 0.3848,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 0.3625,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9485714285714286e-05,
+      "loss": 0.2758,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9542857142857143e-05,
+      "loss": 0.4115,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.3625,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9657142857142858e-05,
+      "loss": 0.3627,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9714285714285718e-05,
+      "loss": 0.3516,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9771428571428574e-05,
+      "loss": 0.2944,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.982857142857143e-05,
+      "loss": 0.41,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.988571428571429e-05,
+      "loss": 0.376,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9942857142857142e-05,
+      "loss": 0.3127,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.3829,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999614624707e-05,
+      "loss": 0.3218,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998458498852e-05,
+      "loss": 0.3351,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999653162253e-05,
+      "loss": 0.3874,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999993833995886e-05,
+      "loss": 0.361,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999036561913e-05,
+      "loss": 0.3157,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999986126492526e-05,
+      "loss": 0.3591,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999981116616402e-05,
+      "loss": 0.3319,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997533599115e-05,
+      "loss": 0.3223,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999968784617204e-05,
+      "loss": 0.323,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999961462495078e-05,
+      "loss": 0.3703,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999953369625334e-05,
+      "loss": 0.3814,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999944506008594e-05,
+      "loss": 0.3477,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999934871645544e-05,
+      "loss": 0.3254,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999924466536925e-05,
+      "loss": 0.4109,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999991329068354e-05,
+      "loss": 0.3309,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999990134408625e-05,
+      "loss": 0.3484,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999888626745975e-05,
+      "loss": 0.3314,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999875138663694e-05,
+      "loss": 0.3262,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999986087984045e-05,
+      "loss": 0.3336,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999984585027734e-05,
+      "loss": 0.3501,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999830049975523e-05,
+      "loss": 0.3049,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999813478936213e-05,
+      "loss": 0.3162,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999796137160693e-05,
+      "loss": 0.359,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999778024650296e-05,
+      "loss": 0.3531,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999975914140642e-05,
+      "loss": 0.6349,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999973948743052e-05,
+      "loss": 0.3192,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999971906272411e-05,
+      "loss": 0.3427,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999697867288764e-05,
+      "loss": 0.3779,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999675901126117e-05,
+      "loss": 0.326,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999965316423786e-05,
+      "loss": 0.3265,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999629656625748e-05,
+      "loss": 0.3572,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999605378291593e-05,
+      "loss": 0.4268,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999580329237264e-05,
+      "loss": 0.3033,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999554509464695e-05,
+      "loss": 0.3462,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999952791897587e-05,
+      "loss": 0.3683,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999500557772843e-05,
+      "loss": 0.3035,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999947242585772e-05,
+      "loss": 0.3604,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999443523232676e-05,
+      "loss": 0.3183,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999413849899933e-05,
+      "loss": 0.3355,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999938340586178e-05,
+      "loss": 0.3467,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999352191120556e-05,
+      "loss": 0.3902,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999932020567868e-05,
+      "loss": 0.3678,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999287449538608e-05,
+      "loss": 0.3354,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999253922702868e-05,
+      "loss": 0.3428,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999921962517404e-05,
+      "loss": 0.3271,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999184556954777e-05,
+      "loss": 0.3287,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999914871804777e-05,
+      "loss": 0.3804,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999911210845579e-05,
+      "loss": 0.3475,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999074728181657e-05,
+      "loss": 0.3671,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999036577228245e-05,
+      "loss": 0.2861,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998997655598505e-05,
+      "loss": 0.3315,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998957963295434e-05,
+      "loss": 0.3436,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998917500322086e-05,
+      "loss": 0.4235,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998876266681585e-05,
+      "loss": 0.3639,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998834262377107e-05,
+      "loss": 0.3844,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998791487411887e-05,
+      "loss": 0.3854,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999874794178923e-05,
+      "loss": 0.3645,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999870362551248e-05,
+      "loss": 0.3372,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998658538585067e-05,
+      "loss": 0.3628,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998612681010452e-05,
+      "loss": 0.3358,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998566052792178e-05,
+      "loss": 0.38,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999851865393384e-05,
+      "loss": 0.3282,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998470484439084e-05,
+      "loss": 0.3574,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999842154431163e-05,
+      "loss": 0.381,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999837183355525e-05,
+      "loss": 0.3918,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998321352173767e-05,
+      "loss": 0.343,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999827010017108e-05,
+      "loss": 0.3431,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998218077551135e-05,
+      "loss": 0.3518,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998165284317944e-05,
+      "loss": 0.3424,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998111720475574e-05,
+      "loss": 0.3839,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998057386028157e-05,
+      "loss": 0.3334,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999800228097988e-05,
+      "loss": 0.3163,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997946405334986e-05,
+      "loss": 0.3484,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997889759097785e-05,
+      "loss": 0.3439,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997832342272642e-05,
+      "loss": 0.3445,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999777415486398e-05,
+      "loss": 0.3461,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999771519687629e-05,
+      "loss": 0.3525,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997655468314115e-05,
+      "loss": 0.3342,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997594969182054e-05,
+      "loss": 0.3387,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999753369948477e-05,
+      "loss": 0.6073,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999747165922699e-05,
+      "loss": 0.3946,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997408848413494e-05,
+      "loss": 0.3114,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999734526704912e-05,
+      "loss": 0.3873,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999728091513877e-05,
+      "loss": 0.3724,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999721579268741e-05,
+      "loss": 0.3638,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997149899700056e-05,
+      "loss": 0.3469,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999708323618178e-05,
+      "loss": 0.3553,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997015802137727e-05,
+      "loss": 0.3351,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999694759757309e-05,
+      "loss": 0.3518,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996878622493134e-05,
+      "loss": 0.3533,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996808876903168e-05,
+      "loss": 0.3652,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996738360808566e-05,
+      "loss": 0.3635,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996667074214768e-05,
+      "loss": 0.309,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996595017127268e-05,
+      "loss": 0.3436,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999652218955162e-05,
+      "loss": 0.4006,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996448591493433e-05,
+      "loss": 0.3715,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996374222958383e-05,
+      "loss": 0.2994,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99962990839522e-05,
+      "loss": 0.3869,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999622317448068e-05,
+      "loss": 0.3598,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996146494549672e-05,
+      "loss": 0.3042,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996069044165082e-05,
+      "loss": 0.3992,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999599082333288e-05,
+      "loss": 0.3358,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99959118320591e-05,
+      "loss": 0.3698,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995832070349827e-05,
+      "loss": 0.3474,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995751538211205e-05,
+      "loss": 0.3728,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999567023564945e-05,
+      "loss": 0.3209,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999558816267082e-05,
+      "loss": 0.6414,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995505319281645e-05,
+      "loss": 0.3498,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995421705488313e-05,
+      "loss": 0.3851,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999533732129726e-05,
+      "loss": 0.3364,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995252166714993e-05,
+      "loss": 0.355,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995166241748084e-05,
+      "loss": 0.3567,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995079546403143e-05,
+      "loss": 0.3937,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999499208068686e-05,
+      "loss": 0.3322,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994903844605973e-05,
+      "loss": 0.3702,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994814838167286e-05,
+      "loss": 0.3237,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994725061377653e-05,
+      "loss": 0.3216,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994634514244002e-05,
+      "loss": 0.3365,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994543196773307e-05,
+      "loss": 0.3491,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994451108972604e-05,
+      "loss": 0.3711,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999435825084899e-05,
+      "loss": 0.3419,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994264622409636e-05,
+      "loss": 0.3483,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999417022366174e-05,
+      "loss": 0.609,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999407505461259e-05,
+      "loss": 0.3381,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993979115269517e-05,
+      "loss": 0.3595,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993882405639914e-05,
+      "loss": 0.3449,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993784925731234e-05,
+      "loss": 0.353,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993686675550998e-05,
+      "loss": 0.3359,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993587655106766e-05,
+      "loss": 0.2803,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993487864406185e-05,
+      "loss": 0.3397,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993387303456938e-05,
+      "loss": 0.3652,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999328597226677e-05,
+      "loss": 0.2666,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99931838708435e-05,
+      "loss": 0.4054,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993080999195e-05,
+      "loss": 0.3444,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999297735732919e-05,
+      "loss": 0.3139,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992872945254064e-05,
+      "loss": 0.3209,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992767762977662e-05,
+      "loss": 0.3729,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99926618105081e-05,
+      "loss": 0.3699,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999255508785354e-05,
+      "loss": 0.3394,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992447595022214e-05,
+      "loss": 0.3563,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992339332022396e-05,
+      "loss": 0.2916,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992230298862436e-05,
+      "loss": 0.3495,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999212049555074e-05,
+      "loss": 0.3378,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992009922095766e-05,
+      "loss": 0.3357,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991898578506043e-05,
+      "loss": 0.3236,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991786464790145e-05,
+      "loss": 0.3782,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999167358095672e-05,
+      "loss": 0.366,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991559927014465e-05,
+      "loss": 0.3559,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999144550297214e-05,
+      "loss": 0.3563,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991330308838565e-05,
+      "loss": 0.5999,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991214344622616e-05,
+      "loss": 0.3724,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999109761033324e-05,
+      "loss": 0.3094,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999098010597942e-05,
+      "loss": 0.3472,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990861831570224e-05,
+      "loss": 0.3942,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990742787114765e-05,
+      "loss": 0.333,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990622972622216e-05,
+      "loss": 0.332,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990502388101813e-05,
+      "loss": 0.4093,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990381033562853e-05,
+      "loss": 0.4125,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990258909014684e-05,
+      "loss": 0.3181,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990136014466722e-05,
+      "loss": 0.3375,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999001234992844e-05,
+      "loss": 0.3879,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989887915409368e-05,
+      "loss": 0.3282,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99897627109191e-05,
+      "loss": 0.3341,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989636736467278e-05,
+      "loss": 0.3152,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998950999206362e-05,
+      "loss": 0.3618,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989382477717888e-05,
+      "loss": 0.5778,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989254193439915e-05,
+      "loss": 0.4129,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998912513923959e-05,
+      "loss": 0.3804,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988995315126852e-05,
+      "loss": 0.3449,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988864721111714e-05,
+      "loss": 0.3252,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998873335720424e-05,
+      "loss": 0.3777,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988601223414555e-05,
+      "loss": 0.3055,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988468319752846e-05,
+      "loss": 0.3441,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998833464622935e-05,
+      "loss": 0.3198,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998820020285437e-05,
+      "loss": 0.3326,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998806498963828e-05,
+      "loss": 0.5486,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987929006591487e-05,
+      "loss": 0.3384,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987792253724477e-05,
+      "loss": 0.3862,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987654731047793e-05,
+      "loss": 0.3315,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987516438572035e-05,
+      "loss": 0.303,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987377376307856e-05,
+      "loss": 0.363,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987237544265982e-05,
+      "loss": 0.3525,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998709694245718e-05,
+      "loss": 0.3682,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986955570892302e-05,
+      "loss": 0.3137,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998681342958223e-05,
+      "loss": 0.3818,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986670518537928e-05,
+      "loss": 0.3285,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986526837770405e-05,
+      "loss": 0.3517,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986382387290738e-05,
+      "loss": 0.3886,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986237167110066e-05,
+      "loss": 0.3514,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998609117723957e-05,
+      "loss": 0.3494,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998594441769051e-05,
+      "loss": 0.6116,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99857968884742e-05,
+      "loss": 0.3469,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985648589602005e-05,
+      "loss": 0.3199,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998549952108536e-05,
+      "loss": 0.3323,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985349682935747e-05,
+      "loss": 0.3555,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998519907516472e-05,
+      "loss": 0.38,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985047697783886e-05,
+      "loss": 0.3052,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984895550804918e-05,
+      "loss": 0.2993,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984742634239535e-05,
+      "loss": 0.3571,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984588948099528e-05,
+      "loss": 0.3547,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984434492396736e-05,
+      "loss": 0.3859,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984279267143072e-05,
+      "loss": 0.3196,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998412327235049e-05,
+      "loss": 0.323,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983966508031026e-05,
+      "loss": 0.3586,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983808974196752e-05,
+      "loss": 0.3491,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983650670859814e-05,
+      "loss": 0.3204,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998349159803241e-05,
+      "loss": 0.3123,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998333175572681e-05,
+      "loss": 0.3387,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983171143955326e-05,
+      "loss": 0.3529,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983009762730336e-05,
+      "loss": 0.3638,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998284761206428e-05,
+      "loss": 0.3256,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998268469196966e-05,
+      "loss": 0.3117,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982521002459026e-05,
+      "loss": 0.3577,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982356543545003e-05,
+      "loss": 0.3594,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982191315240257e-05,
+      "loss": 0.3307,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998202531755753e-05,
+      "loss": 0.3172,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981858550509617e-05,
+      "loss": 0.3244,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981691014109364e-05,
+      "loss": 0.355,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998152270836969e-05,
+      "loss": 0.3032,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998135363330357e-05,
+      "loss": 0.3289,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981183788924025e-05,
+      "loss": 0.303,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981013175244154e-05,
+      "loss": 0.3134,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980841792277104e-05,
+      "loss": 0.3936,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998066964003609e-05,
+      "loss": 0.4129,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980496718534375e-05,
+      "loss": 0.3412,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980323027785285e-05,
+      "loss": 0.3352,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998014856780221e-05,
+      "loss": 0.3317,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979973338598603e-05,
+      "loss": 0.3583,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979797340187957e-05,
+      "loss": 0.3521,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979620572583846e-05,
+      "loss": 0.3589,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979443035799893e-05,
+      "loss": 0.3391,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979264729849776e-05,
+      "loss": 0.3552,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979085654747248e-05,
+      "loss": 0.3248,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978905810506105e-05,
+      "loss": 0.3448,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978725197140204e-05,
+      "loss": 0.3793,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978543814663478e-05,
+      "loss": 0.2988,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99783616630899e-05,
+      "loss": 0.3255,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978178742433504e-05,
+      "loss": 0.3887,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977995052708398e-05,
+      "loss": 0.3858,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977810593928736e-05,
+      "loss": 0.326,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977625366108733e-05,
+      "loss": 0.3425,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977439369262668e-05,
+      "loss": 0.3419,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997725260340488e-05,
+      "loss": 0.3846,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977065068549756e-05,
+      "loss": 0.3597,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976876764711756e-05,
+      "loss": 0.3577,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976687691905394e-05,
+      "loss": 0.4155,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976497850145237e-05,
+      "loss": 0.337,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976307239445924e-05,
+      "loss": 0.3729,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976115859822146e-05,
+      "loss": 0.3073,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975923711288646e-05,
+      "loss": 0.3835,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975730793860242e-05,
+      "loss": 0.343,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99755371075518e-05,
+      "loss": 0.3936,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975342652378247e-05,
+      "loss": 0.3302,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997514742835457e-05,
+      "loss": 0.3183,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997495143549582e-05,
+      "loss": 0.3001,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99747546738171e-05,
+      "loss": 0.3329,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997455714333358e-05,
+      "loss": 0.6206,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974358844060476e-05,
+      "loss": 0.3369,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997415977601308e-05,
+      "loss": 0.3627,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973959939206734e-05,
+      "loss": 0.276,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973759333656835e-05,
+      "loss": 0.3221,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997355795937885e-05,
+      "loss": 0.3703,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99733558163883e-05,
+      "loss": 0.3236,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973152904700762e-05,
+      "loss": 0.3277,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972949224331876e-05,
+      "loss": 0.3821,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997274477529734e-05,
+      "loss": 0.3165,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972539557612918e-05,
+      "loss": 0.351,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972333571294418e-05,
+      "loss": 0.333,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972126816357723e-05,
+      "loss": 0.3143,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971919292818768e-05,
+      "loss": 0.3568,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971711000693544e-05,
+      "loss": 0.3401,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997150193999811e-05,
+      "loss": 0.3361,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997129211074858e-05,
+      "loss": 0.3162,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971081512961117e-05,
+      "loss": 0.3279,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970870146651964e-05,
+      "loss": 0.3678,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970658011837404e-05,
+      "loss": 0.3188,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970445108533795e-05,
+      "loss": 0.3673,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970231436757542e-05,
+      "loss": 0.3157,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970016996525112e-05,
+      "loss": 0.3331,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969801787853035e-05,
+      "loss": 0.3539,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969585810757902e-05,
+      "loss": 0.3256,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996936906525635e-05,
+      "loss": 0.3576,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969151551365097e-05,
+      "loss": 0.3077,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99689332691009e-05,
+      "loss": 0.3336,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996871421848058e-05,
+      "loss": 0.3175,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996849439952103e-05,
+      "loss": 0.3411,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968273812239185e-05,
+      "loss": 0.3721,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968052456652048e-05,
+      "loss": 0.3252,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967830332776684e-05,
+      "loss": 0.2779,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996760744063021e-05,
+      "loss": 0.2823,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967383780229805e-05,
+      "loss": 0.3191,
+      "step": 641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967159351592706e-05,
+      "loss": 0.3174,
+      "step": 642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966934154736216e-05,
+      "loss": 0.3321,
+      "step": 643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996670818967769e-05,
+      "loss": 0.302,
+      "step": 644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966481456434543e-05,
+      "loss": 0.5845,
+      "step": 645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996625395502425e-05,
+      "loss": 0.3726,
+      "step": 646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966025685464353e-05,
+      "loss": 0.3862,
+      "step": 647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965796647772434e-05,
+      "loss": 0.3524,
+      "step": 648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996556684196615e-05,
+      "loss": 0.3419,
+      "step": 649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996533626806322e-05,
+      "loss": 0.3483,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996510492608141e-05,
+      "loss": 0.3784,
+      "step": 651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964872816038547e-05,
+      "loss": 0.3185,
+      "step": 652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964639937952527e-05,
+      "loss": 0.3842,
+      "step": 653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99644062918413e-05,
+      "loss": 0.3243,
+      "step": 654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996417187772287e-05,
+      "loss": 0.3759,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963936695615307e-05,
+      "loss": 0.3246,
+      "step": 656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963700745536733e-05,
+      "loss": 0.3589,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963464027505343e-05,
+      "loss": 0.3464,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996322654153937e-05,
+      "loss": 0.3057,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996298828765713e-05,
+      "loss": 0.3293,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962749265876983e-05,
+      "loss": 0.3138,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962509476217348e-05,
+      "loss": 0.3162,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962268918696708e-05,
+      "loss": 0.3716,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962027593333603e-05,
+      "loss": 0.363,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961785500146638e-05,
+      "loss": 0.3375,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961542639154467e-05,
+      "loss": 0.3518,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961299010375813e-05,
+      "loss": 0.3,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996105461382945e-05,
+      "loss": 0.3693,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960809449534214e-05,
+      "loss": 0.3063,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960563517509008e-05,
+      "loss": 0.304,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960316817772783e-05,
+      "loss": 0.3294,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960069350344547e-05,
+      "loss": 0.3421,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959821115243385e-05,
+      "loss": 0.3419,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959572112488423e-05,
+      "loss": 0.3864,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959322342098854e-05,
+      "loss": 0.3907,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995907180409393e-05,
+      "loss": 0.3643,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958820498492958e-05,
+      "loss": 0.3456,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958568425315316e-05,
+      "loss": 0.279,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995831558458042e-05,
+      "loss": 0.3097,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958061976307767e-05,
+      "loss": 0.3608,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99578076005169e-05,
+      "loss": 0.3563,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957552457227428e-05,
+      "loss": 0.3356,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995729654645901e-05,
+      "loss": 0.3179,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957039868231382e-05,
+      "loss": 0.3613,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956782422564313e-05,
+      "loss": 0.3726,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956524209477658e-05,
+      "loss": 0.296,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995626522899131e-05,
+      "loss": 0.3303,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956005481125235e-05,
+      "loss": 0.3466,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955744965899452e-05,
+      "loss": 0.3399,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995548368333404e-05,
+      "loss": 0.3265,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955221633449137e-05,
+      "loss": 0.361,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995495881626494e-05,
+      "loss": 0.3459,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954695231801706e-05,
+      "loss": 0.3618,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995443088007975e-05,
+      "loss": 0.3339,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995416576111945e-05,
+      "loss": 0.3064,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995389987494124e-05,
+      "loss": 0.3008,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995363322156561e-05,
+      "loss": 0.3214,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995336580101311e-05,
+      "loss": 0.2971,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995309761330436e-05,
+      "loss": 0.6136,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995282865846002e-05,
+      "loss": 0.3083,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995255893650083e-05,
+      "loss": 0.3008,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952288447447573e-05,
+      "loss": 0.3506,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952017191321098e-05,
+      "loss": 0.3716,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951745168142312e-05,
+      "loss": 0.357,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951472377932183e-05,
+      "loss": 0.3171,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951198820711735e-05,
+      "loss": 0.356,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950924496502048e-05,
+      "loss": 0.3558,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950649405324275e-05,
+      "loss": 0.3903,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950373547199612e-05,
+      "loss": 0.417,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995009692214932e-05,
+      "loss": 0.314,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949819530194722e-05,
+      "loss": 0.3521,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99495413713572e-05,
+      "loss": 0.3274,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994926244565819e-05,
+      "loss": 0.619,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994898275311919e-05,
+      "loss": 0.3841,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948702293761763e-05,
+      "loss": 0.3521,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948421067607518e-05,
+      "loss": 0.3113,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948139074678138e-05,
+      "loss": 0.3591,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994785631499535e-05,
+      "loss": 0.3658,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994757278858095e-05,
+      "loss": 0.4016,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947288495456793e-05,
+      "loss": 0.3768,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947003435644788e-05,
+      "loss": 0.4028,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994671760916691e-05,
+      "loss": 0.3552,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946431016045187e-05,
+      "loss": 0.348,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994614365630171e-05,
+      "loss": 0.3878,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994585552995862e-05,
+      "loss": 0.3807,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945566637038133e-05,
+      "loss": 0.3594,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945276977562515e-05,
+      "loss": 0.3416,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994498655155408e-05,
+      "loss": 0.31,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994469535903523e-05,
+      "loss": 0.3784,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944403400028392e-05,
+      "loss": 0.3105,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944110674556082e-05,
+      "loss": 0.5735,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943817182640856e-05,
+      "loss": 0.3445,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943522924305337e-05,
+      "loss": 0.347,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943227899572198e-05,
+      "loss": 0.2965,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994293210846419e-05,
+      "loss": 0.3818,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99426355510041e-05,
+      "loss": 0.3463,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994233822721479e-05,
+      "loss": 0.3334,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994204013711918e-05,
+      "loss": 0.3172,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941741280740235e-05,
+      "loss": 0.3109,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941441658101e-05,
+      "loss": 0.3264,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941141269224564e-05,
+      "loss": 0.3964,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940840114134078e-05,
+      "loss": 0.348,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940538192852753e-05,
+      "loss": 0.3367,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940235505403867e-05,
+      "loss": 0.3307,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993993205181074e-05,
+      "loss": 0.2921,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993962783209677e-05,
+      "loss": 0.3784,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939322846285397e-05,
+      "loss": 0.3198,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939017094400128e-05,
+      "loss": 0.3177,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938710576464535e-05,
+      "loss": 0.6008,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993840329250224e-05,
+      "loss": 0.3267,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938095242536925e-05,
+      "loss": 0.3933,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993778642659233e-05,
+      "loss": 0.3732,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937476844692268e-05,
+      "loss": 0.306,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993716649686059e-05,
+      "loss": 0.3562,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936855383121217e-05,
+      "loss": 0.2776,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936543503498135e-05,
+      "loss": 0.3516,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936230858015376e-05,
+      "loss": 0.3804,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9935917446697038e-05,
+      "loss": 0.3049,
+      "step": 758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993560326956728e-05,
+      "loss": 0.3068,
+      "step": 759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935288326650314e-05,
+      "loss": 0.3506,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934972617970415e-05,
+      "loss": 0.3657,
+      "step": 761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993465614355192e-05,
+      "loss": 0.311,
+      "step": 762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934338903419213e-05,
+      "loss": 0.3981,
+      "step": 763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934020897596752e-05,
+      "loss": 0.5864,
+      "step": 764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933702126109048e-05,
+      "loss": 0.3585,
+      "step": 765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933382588980665e-05,
+      "loss": 0.5985,
+      "step": 766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933062286236235e-05,
+      "loss": 0.3079,
+      "step": 767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932741217900444e-05,
+      "loss": 0.3301,
+      "step": 768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993241938399804e-05,
+      "loss": 0.372,
+      "step": 769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993209678455383e-05,
+      "loss": 0.3864,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931773419592675e-05,
+      "loss": 0.3069,
+      "step": 771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931449289139495e-05,
+      "loss": 0.2963,
+      "step": 772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993112439321928e-05,
+      "loss": 0.3364,
+      "step": 773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993079873185707e-05,
+      "loss": 0.3281,
+      "step": 774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993047230507796e-05,
+      "loss": 0.3059,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993014511290711e-05,
+      "loss": 0.3986,
+      "step": 776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929817155369746e-05,
+      "loss": 0.3258,
+      "step": 777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929488432491137e-05,
+      "loss": 0.3033,
+      "step": 778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929158944296627e-05,
+      "loss": 0.3777,
+      "step": 779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928828690811603e-05,
+      "loss": 0.3599,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928497672061523e-05,
+      "loss": 0.3278,
+      "step": 781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99281658880719e-05,
+      "loss": 0.3645,
+      "step": 782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992783333886831e-05,
+      "loss": 0.3481,
+      "step": 783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927500024476378e-05,
+      "loss": 0.3496,
+      "step": 784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927165944921803e-05,
+      "loss": 0.2755,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926831100230322e-05,
+      "loss": 0.3297,
+      "step": 786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926495490427753e-05,
+      "loss": 0.3339,
+      "step": 787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992615911553996e-05,
+      "loss": 0.2912,
+      "step": 788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925821975592866e-05,
+      "loss": 0.3555,
+      "step": 789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925484070612465e-05,
+      "loss": 0.3691,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925145400624788e-05,
+      "loss": 0.316,
+      "step": 791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992480596565595e-05,
+      "loss": 0.3272,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924465765732106e-05,
+      "loss": 0.3127,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992412480087948e-05,
+      "loss": 0.3422,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992378307112435e-05,
+      "loss": 0.35,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923440576493056e-05,
+      "loss": 0.3304,
+      "step": 796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923097317011995e-05,
+      "loss": 0.3016,
+      "step": 797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922753292707627e-05,
+      "loss": 0.3372,
+      "step": 798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992240850360646e-05,
+      "loss": 0.324,
+      "step": 799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992206294973508e-05,
+      "loss": 0.3347,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992171663112011e-05,
+      "loss": 0.3676,
+      "step": 801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921369547788246e-05,
+      "loss": 0.261,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921021699766243e-05,
+      "loss": 0.2889,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920673087080903e-05,
+      "loss": 0.3527,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920323709759108e-05,
+      "loss": 0.3105,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919973567827776e-05,
+      "loss": 0.2941,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919622661313897e-05,
+      "loss": 0.6091,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991927099024452e-05,
+      "loss": 0.3502,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918918554646745e-05,
+      "loss": 0.3297,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918565354547738e-05,
+      "loss": 0.3663,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918211389974726e-05,
+      "loss": 0.3846,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917856660954985e-05,
+      "loss": 0.3278,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991750116751586e-05,
+      "loss": 0.2994,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917144909684745e-05,
+      "loss": 0.3167,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916787887489108e-05,
+      "loss": 0.3237,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916430100956458e-05,
+      "loss": 0.2934,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991607155011437e-05,
+      "loss": 0.3155,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915712234990486e-05,
+      "loss": 0.5964,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915352155612503e-05,
+      "loss": 0.3163,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914991312008164e-05,
+      "loss": 0.3191,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991462970420529e-05,
+      "loss": 0.3099,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914267332231746e-05,
+      "loss": 0.3243,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991390419611546e-05,
+      "loss": 0.3366,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991354029588443e-05,
+      "loss": 0.3292,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913175631566698e-05,
+      "loss": 0.3362,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912810203190367e-05,
+      "loss": 0.3617,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991244401078361e-05,
+      "loss": 0.3519,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912077054374646e-05,
+      "loss": 0.3312,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911709333991758e-05,
+      "loss": 0.4035,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911340849663293e-05,
+      "loss": 0.3193,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910971601417645e-05,
+      "loss": 0.3417,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991060158928328e-05,
+      "loss": 0.3412,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910230813288713e-05,
+      "loss": 0.3275,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909859273462525e-05,
+      "loss": 0.295,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909486969833346e-05,
+      "loss": 0.3278,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990911390242988e-05,
+      "loss": 0.3846,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908740071280873e-05,
+      "loss": 0.3689,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908365476415146e-05,
+      "loss": 0.3255,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907990117861564e-05,
+      "loss": 0.3314,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907613995649063e-05,
+      "loss": 0.3344,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907237109806627e-05,
+      "loss": 0.3314,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906859460363307e-05,
+      "loss": 0.3077,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906481047348215e-05,
+      "loss": 0.33,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906101870790512e-05,
+      "loss": 0.3832,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905721930719425e-05,
+      "loss": 0.3558,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990534122716423e-05,
+      "loss": 0.3477,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904959760154287e-05,
+      "loss": 0.335,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904577529718982e-05,
+      "loss": 0.3232,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904194535887783e-05,
+      "loss": 0.3464,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903810778690204e-05,
+      "loss": 0.3001,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903426258155833e-05,
+      "loss": 0.3398,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903040974314293e-05,
+      "loss": 0.3693,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990265492719529e-05,
+      "loss": 0.3096,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902268116828578e-05,
+      "loss": 0.3546,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901880543243966e-05,
+      "loss": 0.3275,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901492206471325e-05,
+      "loss": 0.3226,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901103106540597e-05,
+      "loss": 0.2867,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900713243481758e-05,
+      "loss": 0.3155,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900322617324863e-05,
+      "loss": 0.3881,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899931228100024e-05,
+      "loss": 0.3416,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98995390758374e-05,
+      "loss": 0.3362,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989914616056722e-05,
+      "loss": 0.3507,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898752482319766e-05,
+      "loss": 0.3228,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898358041125382e-05,
+      "loss": 0.3431,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989796283701447e-05,
+      "loss": 0.3011,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989756687001749e-05,
+      "loss": 0.3501,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989717014016496e-05,
+      "loss": 0.5685,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989677264748746e-05,
+      "loss": 0.3452,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896374392015624e-05,
+      "loss": 0.3271,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989597537378015e-05,
+      "loss": 0.3281,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895575592811795e-05,
+      "loss": 0.3235,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895175049141366e-05,
+      "loss": 0.5752,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894773742799737e-05,
+      "loss": 0.3206,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989437167381784e-05,
+      "loss": 0.3124,
+      "step": 874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893968842226664e-05,
+      "loss": 0.3116,
+      "step": 875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893565248057257e-05,
+      "loss": 0.3371,
+      "step": 876
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893160891340728e-05,
+      "loss": 0.2988,
+      "step": 877
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989275577210824e-05,
+      "loss": 0.3743,
+      "step": 878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892349890391015e-05,
+      "loss": 0.3319,
+      "step": 879
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891943246220344e-05,
+      "loss": 0.2971,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891535839627565e-05,
+      "loss": 0.3267,
+      "step": 881
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891127670644076e-05,
+      "loss": 0.37,
+      "step": 882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890718739301346e-05,
+      "loss": 0.3203,
+      "step": 883
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989030904563088e-05,
+      "loss": 0.3338,
+      "step": 884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988989858966427e-05,
+      "loss": 0.2997,
+      "step": 885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889487371433134e-05,
+      "loss": 0.3385,
+      "step": 886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889075390969182e-05,
+      "loss": 0.3393,
+      "step": 887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888662648304162e-05,
+      "loss": 0.3691,
+      "step": 888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988824914346989e-05,
+      "loss": 0.2976,
+      "step": 889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887834876498228e-05,
+      "loss": 0.3438,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887419847421113e-05,
+      "loss": 0.3074,
+      "step": 891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887004056270532e-05,
+      "loss": 0.3702,
+      "step": 892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988658750307853e-05,
+      "loss": 0.3333,
+      "step": 893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886170187877214e-05,
+      "loss": 0.3124,
+      "step": 894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988575211069875e-05,
+      "loss": 0.3116,
+      "step": 895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885333271575362e-05,
+      "loss": 0.3234,
+      "step": 896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884913670539327e-05,
+      "loss": 0.2993,
+      "step": 897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884493307622993e-05,
+      "loss": 0.3008,
+      "step": 898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988407218285875e-05,
+      "loss": 0.6298,
+      "step": 899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883650296279068e-05,
+      "loss": 0.2564,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883227647916454e-05,
+      "loss": 0.3343,
+      "step": 901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882804237803487e-05,
+      "loss": 0.316,
+      "step": 902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98823800659728e-05,
+      "loss": 0.3342,
+      "step": 903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881955132457095e-05,
+      "loss": 0.3773,
+      "step": 904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988152943728911e-05,
+      "loss": 0.3944,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881102980501664e-05,
+      "loss": 0.3303,
+      "step": 906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880675762127624e-05,
+      "loss": 0.3203,
+      "step": 907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988024778219992e-05,
+      "loss": 0.3059,
+      "step": 908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879819040751532e-05,
+      "loss": 0.3581,
+      "step": 909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879389537815514e-05,
+      "loss": 0.3987,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878959273424968e-05,
+      "loss": 0.3455,
+      "step": 911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987852824761305e-05,
+      "loss": 0.3065,
+      "step": 912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987809646041299e-05,
+      "loss": 0.2888,
+      "step": 913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987766391185806e-05,
+      "loss": 0.3319,
+      "step": 914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98772306019816e-05,
+      "loss": 0.3539,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876796530817017e-05,
+      "loss": 0.3716,
+      "step": 916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876361698397755e-05,
+      "loss": 0.3692,
+      "step": 917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875926104757337e-05,
+      "loss": 0.3367,
+      "step": 918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875489749929334e-05,
+      "loss": 0.5887,
+      "step": 919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875052633947373e-05,
+      "loss": 0.3486,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987461475684515e-05,
+      "loss": 0.3786,
+      "step": 921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874176118656415e-05,
+      "loss": 0.3452,
+      "step": 922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873736719414977e-05,
+      "loss": 0.364,
+      "step": 923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98732965591547e-05,
+      "loss": 0.308,
+      "step": 924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872855637909506e-05,
+      "loss": 0.3063,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872413955713382e-05,
+      "loss": 0.3671,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871971512600375e-05,
+      "loss": 0.3909,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987152830860458e-05,
+      "loss": 0.36,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987108434376016e-05,
+      "loss": 0.3239,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870639618101333e-05,
+      "loss": 0.3331,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987019413166238e-05,
+      "loss": 0.35,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986974788447763e-05,
+      "loss": 0.3496,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986930087658148e-05,
+      "loss": 0.3754,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868853108008387e-05,
+      "loss": 0.3561,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868404578792858e-05,
+      "loss": 0.3477,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867955288969468e-05,
+      "loss": 0.3076,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986750523857284e-05,
+      "loss": 0.3112,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867054427637667e-05,
+      "loss": 0.3149,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986660285619869e-05,
+      "loss": 0.3094,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986615052429072e-05,
+      "loss": 0.3588,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986569743194862e-05,
+      "loss": 0.3519,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865243579207304e-05,
+      "loss": 0.3169,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986478896610176e-05,
+      "loss": 0.3291,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986433359266703e-05,
+      "loss": 0.3159,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863877458938204e-05,
+      "loss": 0.3229,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863420564950445e-05,
+      "loss": 0.3204,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862962910738965e-05,
+      "loss": 0.2886,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862504496339036e-05,
+      "loss": 0.3209,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862045321785994e-05,
+      "loss": 0.3069,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861585387115228e-05,
+      "loss": 0.3577,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861124692362188e-05,
+      "loss": 0.3427,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986066323756238e-05,
+      "loss": 0.3471,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860201022751376e-05,
+      "loss": 0.3679,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859738047964795e-05,
+      "loss": 0.3129,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859274313238327e-05,
+      "loss": 0.338,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985880981860771e-05,
+      "loss": 0.3145,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858344564108743e-05,
+      "loss": 0.3157,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985787854977729e-05,
+      "loss": 0.3309,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985741177564927e-05,
+      "loss": 0.3246,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856944241760655e-05,
+      "loss": 0.3215,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985647594814748e-05,
+      "loss": 0.3479,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856006894845844e-05,
+      "loss": 0.3265,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98555370818919e-05,
+      "loss": 0.3215,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985506650932185e-05,
+      "loss": 0.3504,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854595177171968e-05,
+      "loss": 0.3011,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854123085478587e-05,
+      "loss": 0.3987,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853650234278088e-05,
+      "loss": 0.3414,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853176623606916e-05,
+      "loss": 0.2996,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852702253501578e-05,
+      "loss": 0.3721,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985222712399863e-05,
+      "loss": 0.2773,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98517512351347e-05,
+      "loss": 0.3486,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851274586946463e-05,
+      "loss": 0.3467,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850797179470657e-05,
+      "loss": 0.3059,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985031901274408e-05,
+      "loss": 0.2716,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849840086803584e-05,
+      "loss": 0.3488,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849360401686084e-05,
+      "loss": 0.371,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848879957428552e-05,
+      "loss": 0.5894,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848398754068018e-05,
+      "loss": 0.6475,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847916791641567e-05,
+      "loss": 0.3305,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847434070186355e-05,
+      "loss": 0.2983,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846950589739576e-05,
+      "loss": 0.3073,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846466350338506e-05,
+      "loss": 0.2871,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984598135202046e-05,
+      "loss": 0.347,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845495594822824e-05,
+      "loss": 0.3453,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984500907878303e-05,
+      "loss": 0.3809,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844521803938588e-05,
+      "loss": 0.3064,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844033770327048e-05,
+      "loss": 0.3229,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984354497798602e-05,
+      "loss": 0.3281,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984305542695319e-05,
+      "loss": 0.3356,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984256511726628e-05,
+      "loss": 0.3574,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984207404896309e-05,
+      "loss": 0.3558,
+      "step": 991
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984158222208146e-05,
+      "loss": 0.3175,
+      "step": 992
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841089636659296e-05,
+      "loss": 0.3337,
+      "step": 993
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840596292734573e-05,
+      "loss": 0.3245,
+      "step": 994
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984010219034531e-05,
+      "loss": 0.3109,
+      "step": 995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839607329529594e-05,
+      "loss": 0.3693,
+      "step": 996
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983911171032556e-05,
+      "loss": 0.3055,
+      "step": 997
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983861533277142e-05,
+      "loss": 0.3206,
+      "step": 998
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838118196905417e-05,
+      "loss": 0.3652,
+      "step": 999
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983762030276588e-05,
+      "loss": 0.3318,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837121650391173e-05,
+      "loss": 0.3504,
+      "step": 1001
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836622239819743e-05,
+      "loss": 0.3115,
+      "step": 1002
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983612207109007e-05,
+      "loss": 0.36,
+      "step": 1003
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983562114424071e-05,
+      "loss": 0.332,
+      "step": 1004
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983511945931027e-05,
+      "loss": 0.3432,
+      "step": 1005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834617016337424e-05,
+      "loss": 0.36,
+      "step": 1006
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983411381536089e-05,
+      "loss": 0.3352,
+      "step": 1007
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833609856419452e-05,
+      "loss": 0.3488,
+      "step": 1008
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983310513955196e-05,
+      "loss": 0.3171,
+      "step": 1009
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832599664797306e-05,
+      "loss": 0.3448,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983209343219446e-05,
+      "loss": 0.3558,
+      "step": 1011
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831586441782427e-05,
+      "loss": 0.3386,
+      "step": 1012
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831078693600295e-05,
+      "loss": 0.3693,
+      "step": 1013
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983057018768719e-05,
+      "loss": 0.3177,
+      "step": 1014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830060924082316e-05,
+      "loss": 0.3682,
+      "step": 1015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829550902824914e-05,
+      "loss": 0.3482,
+      "step": 1016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98290401239543e-05,
+      "loss": 0.3864,
+      "step": 1017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828528587509836e-05,
+      "loss": 0.299,
+      "step": 1018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828016293530954e-05,
+      "loss": 0.3115,
+      "step": 1019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982750324205714e-05,
+      "loss": 0.329,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982698943312793e-05,
+      "loss": 0.3254,
+      "step": 1021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826474866782933e-05,
+      "loss": 0.3503,
+      "step": 1022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825959543061812e-05,
+      "loss": 0.3382,
+      "step": 1023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825443462004278e-05,
+      "loss": 0.3307,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982492662365011e-05,
+      "loss": 0.2749,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824409028039143e-05,
+      "loss": 0.3265,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823890675211275e-05,
+      "loss": 0.3478,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823371565206452e-05,
+      "loss": 0.3239,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822851698064692e-05,
+      "loss": 0.2896,
+      "step": 1029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822331073826056e-05,
+      "loss": 0.2916,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821809692530673e-05,
+      "loss": 0.623,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821287554218733e-05,
+      "loss": 0.3153,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820764658930477e-05,
+      "loss": 0.3569,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820241006706203e-05,
+      "loss": 0.3831,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819716597586277e-05,
+      "loss": 0.6156,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981919143161112e-05,
+      "loss": 0.301,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98186655088212e-05,
+      "loss": 0.3411,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818138829257063e-05,
+      "loss": 0.3174,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817611392959294e-05,
+      "loss": 0.3257,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817083199968552e-05,
+      "loss": 0.3298,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981655425032554e-05,
+      "loss": 0.3502,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816024544071038e-05,
+      "loss": 0.3313,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981549408124586e-05,
+      "loss": 0.3438,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814962861890903e-05,
+      "loss": 0.3137,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814430886047105e-05,
+      "loss": 0.3065,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813898153755465e-05,
+      "loss": 0.3173,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981336466505705e-05,
+      "loss": 0.2928,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812830419992976e-05,
+      "loss": 0.5908,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981229541860442e-05,
+      "loss": 0.2985,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981175966093262e-05,
+      "loss": 0.3712,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811223147018862e-05,
+      "loss": 0.3634,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98106858769045e-05,
+      "loss": 0.3686,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981014785063095e-05,
+      "loss": 0.3173,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980960906823968e-05,
+      "loss": 0.355,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809069529772215e-05,
+      "loss": 0.3669,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808529235270134e-05,
+      "loss": 0.2851,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807988184775085e-05,
+      "loss": 0.3068,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980744637832877e-05,
+      "loss": 0.3116,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980690381597295e-05,
+      "loss": 0.3416,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806360497749436e-05,
+      "loss": 0.3684,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805816423700114e-05,
+      "loss": 0.2925,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805271593866914e-05,
+      "loss": 0.3238,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804726008291827e-05,
+      "loss": 0.3391,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804179667016906e-05,
+      "loss": 0.3081,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803632570084265e-05,
+      "loss": 0.2841,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980308471753606e-05,
+      "loss": 0.3983,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802536109414526e-05,
+      "loss": 0.2763,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980198674576194e-05,
+      "loss": 0.3582,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801436626620658e-05,
+      "loss": 0.3123,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800885752033067e-05,
+      "loss": 0.3397,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800334122041626e-05,
+      "loss": 0.6584,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799781736688862e-05,
+      "loss": 0.3849,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979922859601734e-05,
+      "loss": 0.2826,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798674700069698e-05,
+      "loss": 0.3252,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798120048888628e-05,
+      "loss": 0.3536,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797564642516876e-05,
+      "loss": 0.2996,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797008480997253e-05,
+      "loss": 0.3048,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796451564372624e-05,
+      "loss": 0.3671,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795893892685918e-05,
+      "loss": 0.3156,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979533546598011e-05,
+      "loss": 0.3223,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794776284298247e-05,
+      "loss": 0.2913,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794216347683425e-05,
+      "loss": 0.3216,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97936556561788e-05,
+      "loss": 0.3663,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979309420982759e-05,
+      "loss": 0.3406,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792532008673067e-05,
+      "loss": 0.3624,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791969052758563e-05,
+      "loss": 0.3109,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979140534212747e-05,
+      "loss": 0.3622,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979084087682323e-05,
+      "loss": 0.2924,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790275656889356e-05,
+      "loss": 0.3307,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978970968236941e-05,
+      "loss": 0.3334,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978914295330701e-05,
+      "loss": 0.312,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788575469745845e-05,
+      "loss": 0.3537,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788007231729647e-05,
+      "loss": 0.324,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787438239302217e-05,
+      "loss": 0.3004,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978686849250741e-05,
+      "loss": 0.3788,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786297991389136e-05,
+      "loss": 0.2927,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978572673599137e-05,
+      "loss": 0.3024,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785154726358134e-05,
+      "loss": 0.363,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784581962533528e-05,
+      "loss": 0.3209,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784008444561692e-05,
+      "loss": 0.5852,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783434172486825e-05,
+      "loss": 0.3198,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782859146353196e-05,
+      "loss": 0.357,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782283366205122e-05,
+      "loss": 0.3658,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781706832086984e-05,
+      "loss": 0.3181,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978112954404321e-05,
+      "loss": 0.3444,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780551502118306e-05,
+      "loss": 0.3747,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779972706356818e-05,
+      "loss": 0.3378,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779393156803356e-05,
+      "loss": 0.2866,
+      "step": 1108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778812853502592e-05,
+      "loss": 0.3322,
+      "step": 1109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778231796499254e-05,
+      "loss": 0.3243,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777649985838123e-05,
+      "loss": 0.3409,
+      "step": 1111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777067421564046e-05,
+      "loss": 0.3838,
+      "step": 1112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776484103721918e-05,
+      "loss": 0.3428,
+      "step": 1113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775900032356704e-05,
+      "loss": 0.2975,
+      "step": 1114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977531520751342e-05,
+      "loss": 0.332,
+      "step": 1115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774729629237143e-05,
+      "loss": 0.3278,
+      "step": 1116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774143297573003e-05,
+      "loss": 0.3423,
+      "step": 1117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977355621256619e-05,
+      "loss": 0.3135,
+      "step": 1118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977296837426196e-05,
+      "loss": 0.3113,
+      "step": 1119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772379782705616e-05,
+      "loss": 0.2944,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771790437942528e-05,
+      "loss": 0.3194,
+      "step": 1121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771200340018115e-05,
+      "loss": 0.3669,
+      "step": 1122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977060948897786e-05,
+      "loss": 0.3427,
+      "step": 1123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770017884867306e-05,
+      "loss": 0.2729,
+      "step": 1124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769425527732046e-05,
+      "loss": 0.3396,
+      "step": 1125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768832417617737e-05,
+      "loss": 0.3331,
+      "step": 1126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768238554570098e-05,
+      "loss": 0.2742,
+      "step": 1127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767643938634896e-05,
+      "loss": 0.3248,
+      "step": 1128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767048569857963e-05,
+      "loss": 0.2985,
+      "step": 1129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766452448285184e-05,
+      "loss": 0.3088,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765855573962512e-05,
+      "loss": 0.3185,
+      "step": 1131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765257946935944e-05,
+      "loss": 0.3276,
+      "step": 1132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764659567251546e-05,
+      "loss": 0.3474,
+      "step": 1133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764060434955437e-05,
+      "loss": 0.3538,
+      "step": 1134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97634605500938e-05,
+      "loss": 0.301,
+      "step": 1135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976285991271286e-05,
+      "loss": 0.3145,
+      "step": 1136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762258522858917e-05,
+      "loss": 0.3041,
+      "step": 1137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761656380578328e-05,
+      "loss": 0.3067,
+      "step": 1138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761053485917497e-05,
+      "loss": 0.5691,
+      "step": 1139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760449838922894e-05,
+      "loss": 0.3526,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759845439641047e-05,
+      "loss": 0.3307,
+      "step": 1141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759240288118536e-05,
+      "loss": 0.3066,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758634384402007e-05,
+      "loss": 0.3082,
+      "step": 1143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758027728538157e-05,
+      "loss": 0.3555,
+      "step": 1144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757420320573747e-05,
+      "loss": 0.3967,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756812160555586e-05,
+      "loss": 0.3606,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975620324853056e-05,
+      "loss": 0.269,
+      "step": 1147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755593584545594e-05,
+      "loss": 0.343,
+      "step": 1148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975498316864768e-05,
+      "loss": 0.3651,
+      "step": 1149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975437200088386e-05,
+      "loss": 0.3627,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753760081301242e-05,
+      "loss": 0.3104,
+      "step": 1151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753147409946997e-05,
+      "loss": 0.2998,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752533986868337e-05,
+      "loss": 0.3517,
+      "step": 1153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975191981211255e-05,
+      "loss": 0.3054,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751304885726967e-05,
+      "loss": 0.3893,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750689207758985e-05,
+      "loss": 0.3115,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975007277825606e-05,
+      "loss": 0.2866,
+      "step": 1157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749455597265704e-05,
+      "loss": 0.3831,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974883766483548e-05,
+      "loss": 0.3689,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974821898101302e-05,
+      "loss": 0.3565,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974759954584601e-05,
+      "loss": 0.3069,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746979359382193e-05,
+      "loss": 0.3671,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746358421669365e-05,
+      "loss": 0.3289,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974573673275539e-05,
+      "loss": 0.2828,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745114292688183e-05,
+      "loss": 0.3294,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744491101515715e-05,
+      "loss": 0.3215,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743867159286022e-05,
+      "loss": 0.2885,
+      "step": 1167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743242466047196e-05,
+      "loss": 0.2812,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742617021847385e-05,
+      "loss": 0.3553,
+      "step": 1169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741990826734793e-05,
+      "loss": 0.3026,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741363880757682e-05,
+      "loss": 0.3358,
+      "step": 1171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974073618396438e-05,
+      "loss": 0.2916,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974010773640326e-05,
+      "loss": 0.3527,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739478538122765e-05,
+      "loss": 0.6641,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738848589171388e-05,
+      "loss": 0.3412,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738217889597684e-05,
+      "loss": 0.3173,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737586439450262e-05,
+      "loss": 0.319,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736954238777793e-05,
+      "loss": 0.2921,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736321287629e-05,
+      "loss": 0.3076,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735687586052673e-05,
+      "loss": 0.3096,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735053134097653e-05,
+      "loss": 0.356,
+      "step": 1181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973441793181284e-05,
+      "loss": 0.2906,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973378197924719e-05,
+      "loss": 0.3333,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973314527644972e-05,
+      "loss": 0.3184,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732507823469507e-05,
+      "loss": 0.3072,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973186962035568e-05,
+      "loss": 0.316,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973123066715743e-05,
+      "loss": 0.2676,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730590963924005e-05,
+      "loss": 0.3167,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729950510704708e-05,
+      "loss": 0.3391,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97293093075489e-05,
+      "loss": 0.3276,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972866735450601e-05,
+      "loss": 0.326,
+      "step": 1191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728024651625506e-05,
+      "loss": 0.3426,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972738119895693e-05,
+      "loss": 0.2963,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972673699654988e-05,
+      "loss": 0.3212,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726092044453996e-05,
+      "loss": 0.3074,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725446342719e-05,
+      "loss": 0.3315,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724799891394653e-05,
+      "loss": 0.3264,
+      "step": 1197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724152690530785e-05,
+      "loss": 0.3649,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723504740177274e-05,
+      "loss": 0.3009,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972285604038406e-05,
+      "loss": 0.3538,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972220659120115e-05,
+      "loss": 0.306,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972155639267859e-05,
+      "loss": 0.2983,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97209054448665e-05,
+      "loss": 0.3271,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720253747815055e-05,
+      "loss": 0.3235,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719601301574476e-05,
+      "loss": 0.3336,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718948106195055e-05,
+      "loss": 0.3091,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971829416172714e-05,
+      "loss": 0.3513,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717639468221127e-05,
+      "loss": 0.3291,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716984025727478e-05,
+      "loss": 0.3432,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971632783429672e-05,
+      "loss": 0.3279,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715670893979416e-05,
+      "loss": 0.3201,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971501320482621e-05,
+      "loss": 0.3038,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714354766887787e-05,
+      "loss": 0.3313,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97136955802149e-05,
+      "loss": 0.3126,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713035644858354e-05,
+      "loss": 0.3312,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712374960869015e-05,
+      "loss": 0.6537,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711713528297802e-05,
+      "loss": 0.2822,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97110513471957e-05,
+      "loss": 0.3393,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710388417613742e-05,
+      "loss": 0.2998,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709724739603024e-05,
+      "loss": 0.3191,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709060313214703e-05,
+      "loss": 0.3303,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708395138499986e-05,
+      "loss": 0.3647,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707729215510143e-05,
+      "loss": 0.285,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707062544296497e-05,
+      "loss": 0.3701,
+      "step": 1224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970639512491044e-05,
+      "loss": 0.3284,
+      "step": 1225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705726957403398e-05,
+      "loss": 0.2914,
+      "step": 1226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705058041826887e-05,
+      "loss": 0.284,
+      "step": 1227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704388378232454e-05,
+      "loss": 0.3489,
+      "step": 1228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703717966671715e-05,
+      "loss": 0.2979,
+      "step": 1229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970304680719634e-05,
+      "loss": 0.3108,
+      "step": 1230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702374899858067e-05,
+      "loss": 0.3232,
+      "step": 1231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701702244708673e-05,
+      "loss": 0.287,
+      "step": 1232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970102884180001e-05,
+      "loss": 0.2937,
+      "step": 1233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700354691183977e-05,
+      "loss": 0.3445,
+      "step": 1234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969967979291254e-05,
+      "loss": 0.3489,
+      "step": 1235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699004147037706e-05,
+      "loss": 0.3497,
+      "step": 1236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698327753611557e-05,
+      "loss": 0.3231,
+      "step": 1237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697650612686228e-05,
+      "loss": 0.3339,
+      "step": 1238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696972724313904e-05,
+      "loss": 0.3009,
+      "step": 1239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696294088546842e-05,
+      "loss": 0.3696,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969561470543734e-05,
+      "loss": 0.3711,
+      "step": 1241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694934575037762e-05,
+      "loss": 0.3234,
+      "step": 1242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694253697400532e-05,
+      "loss": 0.3335,
+      "step": 1243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693572072578127e-05,
+      "loss": 0.3385,
+      "step": 1244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692889700623084e-05,
+      "loss": 0.3211,
+      "step": 1245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692206581588e-05,
+      "loss": 0.3268,
+      "step": 1246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969152271552552e-05,
+      "loss": 0.3088,
+      "step": 1247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690838102488356e-05,
+      "loss": 0.316,
+      "step": 1248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690152742529277e-05,
+      "loss": 0.3044,
+      "step": 1249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689466635701106e-05,
+      "loss": 0.2941,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968877978205672e-05,
+      "loss": 0.2966,
+      "step": 1251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688092181649065e-05,
+      "loss": 0.3041,
+      "step": 1252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687403834531133e-05,
+      "loss": 0.3092,
+      "step": 1253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968671474075598e-05,
+      "loss": 0.3373,
+      "step": 1254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968602490037672e-05,
+      "loss": 0.2792,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685334313446523e-05,
+      "loss": 0.3065,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968464298001861e-05,
+      "loss": 0.3174,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968395090014627e-05,
+      "loss": 0.3482,
+      "step": 1258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683258073882845e-05,
+      "loss": 0.2797,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682564501281733e-05,
+      "loss": 0.3176,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681870182396395e-05,
+      "loss": 0.632,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681175117280343e-05,
+      "loss": 0.3442,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680479305987147e-05,
+      "loss": 0.3046,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679782748570434e-05,
+      "loss": 0.3264,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679085445083903e-05,
+      "loss": 0.321,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678387395581292e-05,
+      "loss": 0.2701,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96776886001164e-05,
+      "loss": 0.2859,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676989058743088e-05,
+      "loss": 0.3779,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676288771515275e-05,
+      "loss": 0.335,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675587738486935e-05,
+      "loss": 0.3172,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674885959712106e-05,
+      "loss": 0.3051,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674183435244867e-05,
+      "loss": 0.3927,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673480165139372e-05,
+      "loss": 0.3196,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672776149449826e-05,
+      "loss": 0.3372,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672071388230485e-05,
+      "loss": 0.3205,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967136588153568e-05,
+      "loss": 0.604,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670659629419774e-05,
+      "loss": 0.316,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669952631937206e-05,
+      "loss": 0.3635,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966924488914247e-05,
+      "loss": 0.3298,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668536401090123e-05,
+      "loss": 0.3095,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667827167834756e-05,
+      "loss": 0.3401,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667117189431045e-05,
+      "loss": 0.3059,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666406465933703e-05,
+      "loss": 0.2832,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966569499739752e-05,
+      "loss": 0.3093,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664982783877323e-05,
+      "loss": 0.3404,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966426982542801e-05,
+      "loss": 0.3428,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966355612210453e-05,
+      "loss": 0.3384,
+      "step": 1287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662841673961893e-05,
+      "loss": 0.2938,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662126481055165e-05,
+      "loss": 0.3517,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966141054343947e-05,
+      "loss": 0.386,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660693861169992e-05,
+      "loss": 0.3474,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659976434301967e-05,
+      "loss": 0.2773,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.3221,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658539346991504e-05,
+      "loss": 0.3381,
+      "step": 1294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657819686659838e-05,
+      "loss": 0.3088,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657099281951148e-05,
+      "loss": 0.2853,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656378132920964e-05,
+      "loss": 0.3268,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655656239624864e-05,
+      "loss": 0.3094,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654933602118494e-05,
+      "loss": 0.3052,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965421022045755e-05,
+      "loss": 0.6509,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653486094697785e-05,
+      "loss": 0.3095,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652761224895006e-05,
+      "loss": 0.3526,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652035611105093e-05,
+      "loss": 0.3631,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651309253383964e-05,
+      "loss": 0.3007,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650582151787608e-05,
+      "loss": 0.3162,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649854306372065e-05,
+      "loss": 0.2925,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964912571719343e-05,
+      "loss": 0.3151,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648396384307866e-05,
+      "loss": 0.3362,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964766630777158e-05,
+      "loss": 0.3009,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646935487640848e-05,
+      "loss": 0.3253,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646203923971992e-05,
+      "loss": 0.2739,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645471616821404e-05,
+      "loss": 0.2913,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644738566245526e-05,
+      "loss": 0.3408,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964400477230085e-05,
+      "loss": 0.348,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643270235043942e-05,
+      "loss": 0.3062,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964253495453141e-05,
+      "loss": 0.358,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641798930819932e-05,
+      "loss": 0.3432,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641062163966232e-05,
+      "loss": 0.3134,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96403246540271e-05,
+      "loss": 0.3087,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639586401059376e-05,
+      "loss": 0.3083,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638847405119966e-05,
+      "loss": 0.3452,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963810766626582e-05,
+      "loss": 0.2977,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963736718455396e-05,
+      "loss": 0.3087,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963662596004146e-05,
+      "loss": 0.2944,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635883992785443e-05,
+      "loss": 0.2821,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635141282843105e-05,
+      "loss": 0.3586,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634397830271685e-05,
+      "loss": 0.3307,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963365363512848e-05,
+      "loss": 0.2733,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632908697470857e-05,
+      "loss": 0.3167,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963216301735623e-05,
+      "loss": 0.3754,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963141659484207e-05,
+      "loss": 0.3229,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630669429985908e-05,
+      "loss": 0.2964,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629921522845338e-05,
+      "loss": 0.2958,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629172873477995e-05,
+      "loss": 0.3058,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628423481941587e-05,
+      "loss": 0.3494,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627673348293874e-05,
+      "loss": 0.3159,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962692247259267e-05,
+      "loss": 0.3448,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962617085489585e-05,
+      "loss": 0.3494,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625418495261343e-05,
+      "loss": 0.3315,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962466539374714e-05,
+      "loss": 0.3564,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9623911550411286e-05,
+      "loss": 0.3583,
+      "step": 1341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623156965311884e-05,
+      "loss": 0.2705,
+      "step": 1342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622401638507093e-05,
+      "loss": 0.2719,
+      "step": 1343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621645570055127e-05,
+      "loss": 0.3119,
+      "step": 1344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620888760014262e-05,
+      "loss": 0.3333,
+      "step": 1345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962013120844283e-05,
+      "loss": 0.3217,
+      "step": 1346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961937291539922e-05,
+      "loss": 0.3124,
+      "step": 1347
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618613880941876e-05,
+      "loss": 0.3288,
+      "step": 1348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617854105129303e-05,
+      "loss": 0.3229,
+      "step": 1349
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617093588020057e-05,
+      "loss": 0.3679,
+      "step": 1350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616332329672756e-05,
+      "loss": 0.353,
+      "step": 1351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615570330146077e-05,
+      "loss": 0.3467,
+      "step": 1352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614807589498746e-05,
+      "loss": 0.3987,
+      "step": 1353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614044107789553e-05,
+      "loss": 0.3444,
+      "step": 1354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613279885077347e-05,
+      "loss": 0.2725,
+      "step": 1355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612514921421028e-05,
+      "loss": 0.3038,
+      "step": 1356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611749216879555e-05,
+      "loss": 0.3329,
+      "step": 1357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610982771511947e-05,
+      "loss": 0.2932,
+      "step": 1358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610215585377275e-05,
+      "loss": 0.3118,
+      "step": 1359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609447658534673e-05,
+      "loss": 0.3145,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608678991043325e-05,
+      "loss": 0.3352,
+      "step": 1361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607909582962478e-05,
+      "loss": 0.2866,
+      "step": 1362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607139434351435e-05,
+      "loss": 0.2622,
+      "step": 1363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606368545269557e-05,
+      "loss": 0.3335,
+      "step": 1364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605596915776254e-05,
+      "loss": 0.2919,
+      "step": 1365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604824545931005e-05,
+      "loss": 0.3305,
+      "step": 1366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604051435793342e-05,
+      "loss": 0.3397,
+      "step": 1367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603277585422847e-05,
+      "loss": 0.3036,
+      "step": 1368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602502994879163e-05,
+      "loss": 0.3799,
+      "step": 1369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601727664222e-05,
+      "loss": 0.2985,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960095159351111e-05,
+      "loss": 0.3703,
+      "step": 1371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600174782806313e-05,
+      "loss": 0.3688,
+      "step": 1372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599397232167478e-05,
+      "loss": 0.2938,
+      "step": 1373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598618941654535e-05,
+      "loss": 0.3423,
+      "step": 1374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597839911327475e-05,
+      "loss": 0.3302,
+      "step": 1375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959706014124634e-05,
+      "loss": 0.3286,
+      "step": 1376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596279631471228e-05,
+      "loss": 0.3239,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595498382062295e-05,
+      "loss": 0.2922,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594716393079765e-05,
+      "loss": 0.3518,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593933664583903e-05,
+      "loss": 0.3268,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593150196635037e-05,
+      "loss": 0.3065,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592365989293557e-05,
+      "loss": 0.2964,
+      "step": 1382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95915810426199e-05,
+      "loss": 0.3573,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590795356674578e-05,
+      "loss": 0.34,
+      "step": 1384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590008931518133e-05,
+      "loss": 0.3181,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589221767211188e-05,
+      "loss": 0.3452,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588433863814405e-05,
+      "loss": 0.2997,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587645221388522e-05,
+      "loss": 0.3014,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958685583999432e-05,
+      "loss": 0.3198,
+      "step": 1389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586065719692636e-05,
+      "loss": 0.3335,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958527486054438e-05,
+      "loss": 0.3108,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584483262610492e-05,
+      "loss": 0.3513,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583690925951996e-05,
+      "loss": 0.2822,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582897850629958e-05,
+      "loss": 0.3318,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582104036705506e-05,
+      "loss": 0.2783,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581309484239818e-05,
+      "loss": 0.3094,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580514193294137e-05,
+      "loss": 0.3391,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579718163929767e-05,
+      "loss": 0.2934,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957892139620805e-05,
+      "loss": 0.3187,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578123890190405e-05,
+      "loss": 0.3441,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577325645938294e-05,
+      "loss": 0.3539,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957652666351325e-05,
+      "loss": 0.2948,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575726942976844e-05,
+      "loss": 0.3293,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574926484390725e-05,
+      "loss": 0.3058,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574125287816582e-05,
+      "loss": 0.3119,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957332335331617e-05,
+      "loss": 0.3138,
+      "step": 1406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95725206809513e-05,
+      "loss": 0.3298,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571717270783827e-05,
+      "loss": 0.2903,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570913122875686e-05,
+      "loss": 0.3236,
+      "step": 1409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570108237288853e-05,
+      "loss": 0.316,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569302614085366e-05,
+      "loss": 0.3088,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568496253327317e-05,
+      "loss": 0.3454,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956768915507685e-05,
+      "loss": 0.3477,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566881319396184e-05,
+      "loss": 0.3123,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566072746347576e-05,
+      "loss": 0.3279,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956526343599335e-05,
+      "loss": 0.3585,
+      "step": 1416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956445338839588e-05,
+      "loss": 0.316,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95636426036176e-05,
+      "loss": 0.2994,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562831081721007e-05,
+      "loss": 0.318,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956201882276864e-05,
+      "loss": 0.2904,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561205826823116e-05,
+      "loss": 0.3336,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956039209394709e-05,
+      "loss": 0.3124,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559577624203277e-05,
+      "loss": 0.3179,
+      "step": 1423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558762417654456e-05,
+      "loss": 0.313,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557946474363462e-05,
+      "loss": 0.3146,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955712979439318e-05,
+      "loss": 0.3507,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556312377806555e-05,
+      "loss": 0.3499,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955549422466659e-05,
+      "loss": 0.3262,
+      "step": 1428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955467533503635e-05,
+      "loss": 0.282,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553855708978943e-05,
+      "loss": 0.3256,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553035346557543e-05,
+      "loss": 0.3435,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552214247835387e-05,
+      "loss": 0.3371,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955139241287575e-05,
+      "loss": 0.3386,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550569841741984e-05,
+      "loss": 0.3165,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549746534497484e-05,
+      "loss": 0.3429,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548922491205708e-05,
+      "loss": 0.3297,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954809771193017e-05,
+      "loss": 0.3159,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547272196734436e-05,
+      "loss": 0.2679,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954644594568214e-05,
+      "loss": 0.28,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545618958836957e-05,
+      "loss": 0.2993,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544791236262634e-05,
+      "loss": 0.295,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954396277802296e-05,
+      "loss": 0.295,
+      "step": 1442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543133584181798e-05,
+      "loss": 0.345,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954230365480305e-05,
+      "loss": 0.3275,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954147298995069e-05,
+      "loss": 0.2953,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540641589688735e-05,
+      "loss": 0.4037,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953980945408127e-05,
+      "loss": 0.3205,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538976583192428e-05,
+      "loss": 0.3191,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538142977086406e-05,
+      "loss": 0.3475,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953730863582745e-05,
+      "loss": 0.3641,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536473559479873e-05,
+      "loss": 0.6105,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953563774810803e-05,
+      "loss": 0.2547,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953480120177635e-05,
+      "loss": 0.353,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533963920549307e-05,
+      "loss": 0.6849,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533125904491433e-05,
+      "loss": 0.384,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953228715366732e-05,
+      "loss": 0.3024,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953144766814161e-05,
+      "loss": 0.2779,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953060744797901e-05,
+      "loss": 0.2972,
+      "step": 1458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529766493244284e-05,
+      "loss": 0.3231,
+      "step": 1459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952892480400224e-05,
+      "loss": 0.5813,
+      "step": 1460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952808238031776e-05,
+      "loss": 0.3079,
+      "step": 1461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952723922225577e-05,
+      "loss": 0.2817,
+      "step": 1462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526395329881253e-05,
+      "loss": 0.3154,
+      "step": 1463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525550703259257e-05,
+      "loss": 0.32,
+      "step": 1464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952470534245488e-05,
+      "loss": 0.3211,
+      "step": 1465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952385924753328e-05,
+      "loss": 0.2892,
+      "step": 1466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523012418559663e-05,
+      "loss": 0.3455,
+      "step": 1467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522164855599306e-05,
+      "loss": 0.3442,
+      "step": 1468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521316558717534e-05,
+      "loss": 0.2601,
+      "step": 1469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520467527979726e-05,
+      "loss": 0.2995,
+      "step": 1470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519617763451322e-05,
+      "loss": 0.2793,
+      "step": 1471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518767265197823e-05,
+      "loss": 0.2986,
+      "step": 1472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951791603328477e-05,
+      "loss": 0.3049,
+      "step": 1473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517064067777786e-05,
+      "loss": 0.28,
+      "step": 1474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516211368742524e-05,
+      "loss": 0.3165,
+      "step": 1475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951535793624471e-05,
+      "loss": 0.3018,
+      "step": 1476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514503770350125e-05,
+      "loss": 0.3166,
+      "step": 1477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513648871124604e-05,
+      "loss": 0.3506,
+      "step": 1478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512793238634035e-05,
+      "loss": 0.309,
+      "step": 1479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511936872944367e-05,
+      "loss": 0.3057,
+      "step": 1480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95110797741216e-05,
+      "loss": 0.3157,
+      "step": 1481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510221942231803e-05,
+      "loss": 0.2953,
+      "step": 1482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950936337734109e-05,
+      "loss": 0.3173,
+      "step": 1483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508504079515637e-05,
+      "loss": 0.3226,
+      "step": 1484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507644048821674e-05,
+      "loss": 0.2922,
+      "step": 1485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506783285325482e-05,
+      "loss": 0.3123,
+      "step": 1486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950592178909341e-05,
+      "loss": 0.3209,
+      "step": 1487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950505956019186e-05,
+      "loss": 0.2998,
+      "step": 1488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950419659868728e-05,
+      "loss": 0.3032,
+      "step": 1489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503332904646188e-05,
+      "loss": 0.2921,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502468478135158e-05,
+      "loss": 0.3018,
+      "step": 1491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501603319220807e-05,
+      "loss": 0.3274,
+      "step": 1492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950073742796982e-05,
+      "loss": 0.3657,
+      "step": 1493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499870804448936e-05,
+      "loss": 0.3357,
+      "step": 1494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499003448724956e-05,
+      "loss": 0.3713,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498135360864724e-05,
+      "loss": 0.3123,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497266540935146e-05,
+      "loss": 0.324,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496396989003195e-05,
+      "loss": 0.2944,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495526705135885e-05,
+      "loss": 0.3209,
+      "step": 1499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494655689400294e-05,
+      "loss": 0.6093,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949378394186356e-05,
+      "loss": 0.3105,
+      "step": 1501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949291146259287e-05,
+      "loss": 0.2971,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949203825165547e-05,
+      "loss": 0.3077,
+      "step": 1503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949116430911866e-05,
+      "loss": 0.3217,
+      "step": 1504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490289635049805e-05,
+      "loss": 0.3035,
+      "step": 1505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489414229516318e-05,
+      "loss": 0.3545,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948853809258567e-05,
+      "loss": 0.3203,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948766122432539e-05,
+      "loss": 0.3126,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948678362480306e-05,
+      "loss": 0.3262,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948590529408633e-05,
+      "loss": 0.3683,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485026232242887e-05,
+      "loss": 0.2927,
+      "step": 1511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948414643934049e-05,
+      "loss": 0.3082,
+      "step": 1512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483265915446946e-05,
+      "loss": 0.3093,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482384660630125e-05,
+      "loss": 0.2876,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948150267495795e-05,
+      "loss": 0.3708,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480619958498393e-05,
+      "loss": 0.3249,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94797365113195e-05,
+      "loss": 0.3624,
+      "step": 1517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478852333489356e-05,
+      "loss": 0.2885,
+      "step": 1518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947796742507611e-05,
+      "loss": 0.2864,
+      "step": 1519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947708178614797e-05,
+      "loss": 0.2759,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476195416773188e-05,
+      "loss": 0.2808,
+      "step": 1521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947530831702009e-05,
+      "loss": 0.3254,
+      "step": 1522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474420486957045e-05,
+      "loss": 0.3075,
+      "step": 1523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473531926652483e-05,
+      "loss": 0.2863,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947264263617489e-05,
+      "loss": 0.3137,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947175261559281e-05,
+      "loss": 0.3484,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470861864974837e-05,
+      "loss": 0.3353,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946997038438963e-05,
+      "loss": 0.2685,
+      "step": 1528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94690781739059e-05,
+      "loss": 0.2676,
+      "step": 1529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946818523359241e-05,
+      "loss": 0.3207,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467291563517987e-05,
+      "loss": 0.3172,
+      "step": 1531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946639716375151e-05,
+      "loss": 0.3126,
+      "step": 1532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465502034361908e-05,
+      "loss": 0.3037,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946460617541819e-05,
+      "loss": 0.3279,
+      "step": 1534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463709586989387e-05,
+      "loss": 0.2889,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462812269144613e-05,
+      "loss": 0.275,
+      "step": 1536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461914221953026e-05,
+      "loss": 0.3031,
+      "step": 1537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461015445483843e-05,
+      "loss": 0.2833,
+      "step": 1538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946011593980634e-05,
+      "loss": 0.3129,
+      "step": 1539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945921570498984e-05,
+      "loss": 0.3472,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458314741103737e-05,
+      "loss": 0.3273,
+      "step": 1541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457413048217466e-05,
+      "loss": 0.32,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945651062640053e-05,
+      "loss": 0.3675,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455607475722478e-05,
+      "loss": 0.3088,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454703596252926e-05,
+      "loss": 0.3207,
+      "step": 1545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453798988061535e-05,
+      "loss": 0.3292,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945289365121803e-05,
+      "loss": 0.309,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451987585792195e-05,
+      "loss": 0.2941,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945108079185386e-05,
+      "loss": 0.3597,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450173269472915e-05,
+      "loss": 0.2799,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449265018719307e-05,
+      "loss": 0.3589,
+      "step": 1551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448356039663044e-05,
+      "loss": 0.3272,
+      "step": 1552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447446332374182e-05,
+      "loss": 0.3453,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944653589692284e-05,
+      "loss": 0.3012,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445624733379186e-05,
+      "loss": 0.3046,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944471284181345e-05,
+      "loss": 0.316,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443800222295918e-05,
+      "loss": 0.291,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442886874896924e-05,
+      "loss": 0.3208,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944197279968687e-05,
+      "loss": 0.3055,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441057996736207e-05,
+      "loss": 0.3518,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440142466115443e-05,
+      "loss": 0.2662,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439226207895143e-05,
+      "loss": 0.2976,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943830922214593e-05,
+      "loss": 0.2922,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437391508938476e-05,
+      "loss": 0.3578,
+      "step": 1564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436473068343516e-05,
+      "loss": 0.31,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435553900431838e-05,
+      "loss": 0.3405,
+      "step": 1566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943463400527429e-05,
+      "loss": 0.306,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433713382941768e-05,
+      "loss": 0.3032,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943279203350523e-05,
+      "loss": 0.288,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431869957035698e-05,
+      "loss": 0.3475,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943094715360423e-05,
+      "loss": 0.3038,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430023623281955e-05,
+      "loss": 0.3118,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429099366140055e-05,
+      "loss": 0.3512,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428174382249764e-05,
+      "loss": 0.3785,
+      "step": 1574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942724867168238e-05,
+      "loss": 0.3773,
+      "step": 1575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426322234509248e-05,
+      "loss": 0.3151,
+      "step": 1576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425395070801775e-05,
+      "loss": 0.3328,
+      "step": 1577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424467180631422e-05,
+      "loss": 0.3016,
+      "step": 1578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942353856406971e-05,
+      "loss": 0.3376,
+      "step": 1579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422609221188208e-05,
+      "loss": 0.2816,
+      "step": 1580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421679152058545e-05,
+      "loss": 0.3253,
+      "step": 1581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420748356752405e-05,
+      "loss": 0.3448,
+      "step": 1582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419816835341528e-05,
+      "loss": 0.2772,
+      "step": 1583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941888458789772e-05,
+      "loss": 0.3293,
+      "step": 1584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417951614492824e-05,
+      "loss": 0.2995,
+      "step": 1585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417017915198758e-05,
+      "loss": 0.3094,
+      "step": 1586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416083490087475e-05,
+      "loss": 0.3469,
+      "step": 1587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415148339231008e-05,
+      "loss": 0.3573,
+      "step": 1588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414212462701428e-05,
+      "loss": 0.3596,
+      "step": 1589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941327586057087e-05,
+      "loss": 0.3703,
+      "step": 1590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412338532911515e-05,
+      "loss": 0.3645,
+      "step": 1591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411400479795618e-05,
+      "loss": 0.2745,
+      "step": 1592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410461701295474e-05,
+      "loss": 0.2949,
+      "step": 1593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940952219748344e-05,
+      "loss": 0.6079,
+      "step": 1594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940858196843193e-05,
+      "loss": 0.3245,
+      "step": 1595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940764101421341e-05,
+      "loss": 0.3019,
+      "step": 1596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406699334900407e-05,
+      "loss": 0.3044,
+      "step": 1597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405756930565496e-05,
+      "loss": 0.3335,
+      "step": 1598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940481380128132e-05,
+      "loss": 0.2662,
+      "step": 1599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403869947120563e-05,
+      "loss": 0.3498,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402925368155978e-05,
+      "loss": 0.3083,
+      "step": 1601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940198006446037e-05,
+      "loss": 0.313,
+      "step": 1602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940103403610659e-05,
+      "loss": 0.2927,
+      "step": 1603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940008728316756e-05,
+      "loss": 0.2975,
+      "step": 1604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939913980571625e-05,
+      "loss": 0.3222,
+      "step": 1605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398191603825687e-05,
+      "loss": 0.3059,
+      "step": 1606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939724267756895e-05,
+      "loss": 0.3022,
+      "step": 1607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396293027019186e-05,
+      "loss": 0.2991,
+      "step": 1608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395342652249578e-05,
+      "loss": 0.3347,
+      "step": 1609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394391553333384e-05,
+      "loss": 0.3745,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939343973034391e-05,
+      "loss": 0.2777,
+      "step": 1611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392487183354514e-05,
+      "loss": 0.2809,
+      "step": 1612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391533912438615e-05,
+      "loss": 0.3855,
+      "step": 1613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939057991766969e-05,
+      "loss": 0.3334,
+      "step": 1614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389625199121264e-05,
+      "loss": 0.271,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938866975686692e-05,
+      "loss": 0.2993,
+      "step": 1616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387713590980305e-05,
+      "loss": 0.3319,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386756701535115e-05,
+      "loss": 0.3168,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385799088605095e-05,
+      "loss": 0.2659,
+      "step": 1619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938484075226406e-05,
+      "loss": 0.3383,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938388169258587e-05,
+      "loss": 0.299,
+      "step": 1621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382921909644448e-05,
+      "loss": 0.2908,
+      "step": 1622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938196140351377e-05,
+      "loss": 0.3201,
+      "step": 1623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938100017426786e-05,
+      "loss": 0.3209,
+      "step": 1624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938003822198081e-05,
+      "loss": 0.2896,
+      "step": 1625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379075546726764e-05,
+      "loss": 0.2878,
+      "step": 1626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378112148579916e-05,
+      "loss": 0.3328,
+      "step": 1627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377148027614523e-05,
+      "loss": 0.3584,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376183183904896e-05,
+      "loss": 0.3284,
+      "step": 1629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375217617525396e-05,
+      "loss": 0.3144,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374251328550448e-05,
+      "loss": 0.2917,
+      "step": 1631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373284317054525e-05,
+      "loss": 0.3212,
+      "step": 1632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372316583112163e-05,
+      "loss": 0.3152,
+      "step": 1633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937134812679795e-05,
+      "loss": 0.3564,
+      "step": 1634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370378948186526e-05,
+      "loss": 0.3588,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369409047352593e-05,
+      "loss": 0.3024,
+      "step": 1636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936843842437091e-05,
+      "loss": 0.3436,
+      "step": 1637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936746707931628e-05,
+      "loss": 0.3406,
+      "step": 1638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366495012263575e-05,
+      "loss": 0.3167,
+      "step": 1639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365522223287717e-05,
+      "loss": 0.3439,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936454871246368e-05,
+      "loss": 0.364,
+      "step": 1641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363574479866504e-05,
+      "loss": 0.3007,
+      "step": 1642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936259952557127e-05,
+      "loss": 0.3105,
+      "step": 1643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361623849653126e-05,
+      "loss": 0.3159,
+      "step": 1644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360647452187273e-05,
+      "loss": 0.3434,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359670333248967e-05,
+      "loss": 0.35,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935869249291352e-05,
+      "loss": 0.3565,
+      "step": 1647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357713931256298e-05,
+      "loss": 0.2778,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356734648352727e-05,
+      "loss": 0.3659,
+      "step": 1649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935575464427828e-05,
+      "loss": 0.2652,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935477391910849e-05,
+      "loss": 0.3193,
+      "step": 1651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353792472918954e-05,
+      "loss": 0.2858,
+      "step": 1652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352810305785314e-05,
+      "loss": 0.2687,
+      "step": 1653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935182741778326e-05,
+      "loss": 0.2885,
+      "step": 1654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350843808988566e-05,
+      "loss": 0.3188,
+      "step": 1655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349859479477034e-05,
+      "loss": 0.3995,
+      "step": 1656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934887442932453e-05,
+      "loss": 0.308,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934788865860698e-05,
+      "loss": 0.3033,
+      "step": 1658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346902167400363e-05,
+      "loss": 0.33,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345914955780708e-05,
+      "loss": 0.2574,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344927023824112e-05,
+      "loss": 0.6281,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343938371606714e-05,
+      "loss": 0.332,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342948999204712e-05,
+      "loss": 0.2936,
+      "step": 1663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934195890669437e-05,
+      "loss": 0.4017,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340968094151997e-05,
+      "loss": 0.3555,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339976561653956e-05,
+      "loss": 0.317,
+      "step": 1666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933898430927667e-05,
+      "loss": 0.3245,
+      "step": 1667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933799133709662e-05,
+      "loss": 0.2429,
+      "step": 1668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336997645190342e-05,
+      "loss": 0.3087,
+      "step": 1669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933600323363442e-05,
+      "loss": 0.3079,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335008102505494e-05,
+      "loss": 0.314,
+      "step": 1671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334012251880274e-05,
+      "loss": 0.3101,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933301568183551e-05,
+      "loss": 0.3582,
+      "step": 1673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933201839244801e-05,
+      "loss": 0.3174,
+      "step": 1674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331020383794647e-05,
+      "loss": 0.3832,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330021655952333e-05,
+      "loss": 0.3176,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329022208998056e-05,
+      "loss": 0.35,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328022043008842e-05,
+      "loss": 0.3019,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327021158061776e-05,
+      "loss": 0.3632,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932601955423401e-05,
+      "loss": 0.2994,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325017231602737e-05,
+      "loss": 0.3564,
+      "step": 1681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932401419024521e-05,
+      "loss": 0.3143,
+      "step": 1682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323010430238746e-05,
+      "loss": 0.3065,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93220059516607e-05,
+      "loss": 0.3391,
+      "step": 1684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321000754588497e-05,
+      "loss": 0.3037,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931999483909961e-05,
+      "loss": 0.3147,
+      "step": 1686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318988205271577e-05,
+      "loss": 0.3432,
+      "step": 1687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317980853181975e-05,
+      "loss": 0.3123,
+      "step": 1688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316972782908455e-05,
+      "loss": 0.2975,
+      "step": 1689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315963994528707e-05,
+      "loss": 0.3085,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9314954488120484e-05,
+      "loss": 0.3236,
+      "step": 1691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93139442637616e-05,
+      "loss": 0.306,
+      "step": 1692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312933321529912e-05,
+      "loss": 0.3127,
+      "step": 1693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311921661503338e-05,
+      "loss": 0.2866,
+      "step": 1694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310909283759855e-05,
+      "loss": 0.3159,
+      "step": 1695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930989618837749e-05,
+      "loss": 0.3041,
+      "step": 1696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930888237543433e-05,
+      "loss": 0.3199,
+      "step": 1697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307867845008513e-05,
+      "loss": 0.3293,
+      "step": 1698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306852597178233e-05,
+      "loss": 0.3046,
+      "step": 1699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305836632021744e-05,
+      "loss": 0.6284,
+      "step": 1700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930481994961735e-05,
+      "loss": 0.3041,
+      "step": 1701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303802550043404e-05,
+      "loss": 0.3094,
+      "step": 1702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302784433378333e-05,
+      "loss": 0.32,
+      "step": 1703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301765599700604e-05,
+      "loss": 0.2931,
+      "step": 1704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300746049088746e-05,
+      "loss": 0.3551,
+      "step": 1705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299725781621335e-05,
+      "loss": 0.2919,
+      "step": 1706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298704797377013e-05,
+      "loss": 0.3267,
+      "step": 1707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929768309643447e-05,
+      "loss": 0.2783,
+      "step": 1708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929666067887246e-05,
+      "loss": 0.2867,
+      "step": 1709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929563754476978e-05,
+      "loss": 0.3345,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294613694205285e-05,
+      "loss": 0.3818,
+      "step": 1711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293589127257896e-05,
+      "loss": 0.3317,
+      "step": 1712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292563844006578e-05,
+      "loss": 0.3288,
+      "step": 1713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291537844530352e-05,
+      "loss": 0.316,
+      "step": 1714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92905111289083e-05,
+      "loss": 0.3204,
+      "step": 1715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928948369721956e-05,
+      "loss": 0.2514,
+      "step": 1716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288455549543315e-05,
+      "loss": 0.2993,
+      "step": 1717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928742668595881e-05,
+      "loss": 0.3181,
+      "step": 1718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286397106545348e-05,
+      "loss": 0.3017,
+      "step": 1719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285366811382283e-05,
+      "loss": 0.2833,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284335800549026e-05,
+      "loss": 0.2975,
+      "step": 1721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928330407412504e-05,
+      "loss": 0.3507,
+      "step": 1722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282271632189844e-05,
+      "loss": 0.373,
+      "step": 1723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281238474823016e-05,
+      "loss": 0.3446,
+      "step": 1724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280204602104185e-05,
+      "loss": 0.2996,
+      "step": 1725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927917001411304e-05,
+      "loss": 0.3392,
+      "step": 1726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278134710929322e-05,
+      "loss": 0.3228,
+      "step": 1727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277098692632824e-05,
+      "loss": 0.3207,
+      "step": 1728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276061959303397e-05,
+      "loss": 0.2594,
+      "step": 1729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927502451102095e-05,
+      "loss": 0.3571,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927398634786544e-05,
+      "loss": 0.3116,
+      "step": 1731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927294746991689e-05,
+      "loss": 0.3369,
+      "step": 1732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271907877255364e-05,
+      "loss": 0.2906,
+      "step": 1733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270867569960994e-05,
+      "loss": 0.325,
+      "step": 1734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269826548113964e-05,
+      "loss": 0.3072,
+      "step": 1735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268784811794507e-05,
+      "loss": 0.3358,
+      "step": 1736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926774236108291e-05,
+      "loss": 0.3163,
+      "step": 1737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926669919605953e-05,
+      "loss": 0.288,
+      "step": 1738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926565531680476e-05,
+      "loss": 0.3553,
+      "step": 1739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264610723399065e-05,
+      "loss": 0.2657,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926356541592295e-05,
+      "loss": 0.2704,
+      "step": 1741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262519394456985e-05,
+      "loss": 0.3201,
+      "step": 1742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261472659081793e-05,
+      "loss": 0.3238,
+      "step": 1743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260425209878052e-05,
+      "loss": 0.3533,
+      "step": 1744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259377046926494e-05,
+      "loss": 0.2781,
+      "step": 1745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258328170307905e-05,
+      "loss": 0.3461,
+      "step": 1746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257278580103124e-05,
+      "loss": 0.3197,
+      "step": 1747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256228276393055e-05,
+      "loss": 0.3057,
+      "step": 1748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255177259258647e-05,
+      "loss": 0.3086,
+      "step": 1749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254125528780908e-05,
+      "loss": 0.3104,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253073085040895e-05,
+      "loss": 0.3045,
+      "step": 1751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252019928119733e-05,
+      "loss": 0.2928,
+      "step": 1752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925096605809859e-05,
+      "loss": 0.3046,
+      "step": 1753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924991147505869e-05,
+      "loss": 0.3116,
+      "step": 1754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924885617908132e-05,
+      "loss": 0.3054,
+      "step": 1755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247800170247817e-05,
+      "loss": 0.3149,
+      "step": 1756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924674344863957e-05,
+      "loss": 0.2634,
+      "step": 1757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924568601433803e-05,
+      "loss": 0.3314,
+      "step": 1758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244627867424695e-05,
+      "loss": 0.3168,
+      "step": 1759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243569007981126e-05,
+      "loss": 0.2955,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242509436088928e-05,
+      "loss": 0.34,
+      "step": 1761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924144915182977e-05,
+      "loss": 0.3589,
+      "step": 1762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240388155285378e-05,
+      "loss": 0.5908,
+      "step": 1763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239326446537526e-05,
+      "loss": 0.2861,
+      "step": 1764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238264025668043e-05,
+      "loss": 0.3053,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237200892758814e-05,
+      "loss": 0.2995,
+      "step": 1766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236137047891783e-05,
+      "loss": 0.3101,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235072491148946e-05,
+      "loss": 0.295,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234007222612356e-05,
+      "loss": 0.3499,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232941242364114e-05,
+      "loss": 0.6174,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923187455048638e-05,
+      "loss": 0.3262,
+      "step": 1771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230807147061374e-05,
+      "loss": 0.3286,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229739032171365e-05,
+      "loss": 0.3104,
+      "step": 1773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228670205898675e-05,
+      "loss": 0.3065,
+      "step": 1774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227600668325687e-05,
+      "loss": 0.2874,
+      "step": 1775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226530419534834e-05,
+      "loss": 0.2909,
+      "step": 1776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225459459608604e-05,
+      "loss": 0.2921,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224387788629547e-05,
+      "loss": 0.2999,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223315406680254e-05,
+      "loss": 0.2816,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222242313843385e-05,
+      "loss": 0.3271,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922116851020165e-05,
+      "loss": 0.2969,
+      "step": 1781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220093995837805e-05,
+      "loss": 0.3345,
+      "step": 1782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219018770834676e-05,
+      "loss": 0.3407,
+      "step": 1783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921794283527513e-05,
+      "loss": 0.3454,
+      "step": 1784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216866189242095e-05,
+      "loss": 0.3488,
+      "step": 1785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921578883281856e-05,
+      "loss": 0.3583,
+      "step": 1786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921471076608756e-05,
+      "loss": 0.3239,
+      "step": 1787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213631989132184e-05,
+      "loss": 0.3118,
+      "step": 1788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212552502035576e-05,
+      "loss": 0.3254,
+      "step": 1789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211472304880945e-05,
+      "loss": 0.316,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210391397751544e-05,
+      "loss": 0.3038,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920930978073068e-05,
+      "loss": 0.3088,
+      "step": 1792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920822745390173e-05,
+      "loss": 0.3186,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207144417348103e-05,
+      "loss": 0.3231,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920606067115328e-05,
+      "loss": 0.3181,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204976215400788e-05,
+      "loss": 0.6426,
+      "step": 1796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920389105017421e-05,
+      "loss": 0.2999,
+      "step": 1797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920280517555719e-05,
+      "loss": 0.3567,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201718591633417e-05,
+      "loss": 0.3383,
+      "step": 1799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200631298486647e-05,
+      "loss": 0.2611,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199543296200675e-05,
+      "loss": 0.3422,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919845458485936e-05,
+      "loss": 0.3254,
+      "step": 1802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919736516454662e-05,
+      "loss": 0.3239,
+      "step": 1803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919627503534642e-05,
+      "loss": 0.3033,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195184197342775e-05,
+      "loss": 0.3134,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194092650619767e-05,
+      "loss": 0.3568,
+      "step": 1806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193000395261532e-05,
+      "loss": 0.3514,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9191907431352248e-05,
+      "loss": 0.3237,
+      "step": 1808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190813758976156e-05,
+      "loss": 0.2794,
+      "step": 1809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189719378217554e-05,
+      "loss": 0.3182,
+      "step": 1810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188624289160792e-05,
+      "loss": 0.2975,
+      "step": 1811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918752849189027e-05,
+      "loss": 0.3179,
+      "step": 1812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918643198649045e-05,
+      "loss": 0.3536,
+      "step": 1813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918533477304584e-05,
+      "loss": 0.3471,
+      "step": 1814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184236851641017e-05,
+      "loss": 0.3458,
+      "step": 1815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183138222360596e-05,
+      "loss": 0.3276,
+      "step": 1816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182038885289256e-05,
+      "loss": 0.2862,
+      "step": 1817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180938840511727e-05,
+      "loss": 0.289,
+      "step": 1818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91798380881128e-05,
+      "loss": 0.3012,
+      "step": 1819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178736628177308e-05,
+      "loss": 0.2905,
+      "step": 1820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177634460790152e-05,
+      "loss": 0.2877,
+      "step": 1821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176531586036282e-05,
+      "loss": 0.3076,
+      "step": 1822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175428004000695e-05,
+      "loss": 0.3479,
+      "step": 1823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917432371476846e-05,
+      "loss": 0.3928,
+      "step": 1824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173218718424683e-05,
+      "loss": 0.3532,
+      "step": 1825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917211301505453e-05,
+      "loss": 0.2908,
+      "step": 1826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917100660474323e-05,
+      "loss": 0.3114,
+      "step": 1827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169899487576056e-05,
+      "loss": 0.3756,
+      "step": 1828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168791663638338e-05,
+      "loss": 0.2842,
+      "step": 1829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167683133015465e-05,
+      "loss": 0.3354,
+      "step": 1830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166573895792873e-05,
+      "loss": 0.5896,
+      "step": 1831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916546395205606e-05,
+      "loss": 0.3014,
+      "step": 1832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916435330189057e-05,
+      "loss": 0.2882,
+      "step": 1833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163241945382012e-05,
+      "loss": 0.3216,
+      "step": 1834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162129882616046e-05,
+      "loss": 0.3249,
+      "step": 1835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161017113678376e-05,
+      "loss": 0.3039,
+      "step": 1836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159903638654773e-05,
+      "loss": 0.3002,
+      "step": 1837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158789457631054e-05,
+      "loss": 0.2936,
+      "step": 1838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157674570693104e-05,
+      "loss": 0.3344,
+      "step": 1839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156558977926847e-05,
+      "loss": 0.3528,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155442679418267e-05,
+      "loss": 0.3334,
+      "step": 1841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91543256752534e-05,
+      "loss": 0.3035,
+      "step": 1842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915320796551835e-05,
+      "loss": 0.3264,
+      "step": 1843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152089550299253e-05,
+      "loss": 0.2941,
+      "step": 1844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150970429682316e-05,
+      "loss": 0.3088,
+      "step": 1845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149850603753793e-05,
+      "loss": 0.3586,
+      "step": 1846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91487300726e-05,
+      "loss": 0.3423,
+      "step": 1847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147608836307296e-05,
+      "loss": 0.2903,
+      "step": 1848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91464868949621e-05,
+      "loss": 0.3341,
+      "step": 1849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145364248650892e-05,
+      "loss": 0.3488,
+      "step": 1850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144240897460195e-05,
+      "loss": 0.2908,
+      "step": 1851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914311684147659e-05,
+      "loss": 0.3142,
+      "step": 1852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141992080786718e-05,
+      "loss": 0.3373,
+      "step": 1853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140866615477272e-05,
+      "loss": 0.2561,
+      "step": 1854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913974044563499e-05,
+      "loss": 0.3109,
+      "step": 1855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138613571346676e-05,
+      "loss": 0.272,
+      "step": 1856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137485992699182e-05,
+      "loss": 0.289,
+      "step": 1857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136357709779418e-05,
+      "loss": 0.3573,
+      "step": 1858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135228722674347e-05,
+      "loss": 0.3286,
+      "step": 1859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913409903147098e-05,
+      "loss": 0.27,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91329686362564e-05,
+      "loss": 0.2725,
+      "step": 1861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131837537117724e-05,
+      "loss": 0.3669,
+      "step": 1862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130705734142127e-05,
+      "loss": 0.3441,
+      "step": 1863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912957322741685e-05,
+      "loss": 0.2839,
+      "step": 1864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912844001702918e-05,
+      "loss": 0.3696,
+      "step": 1865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912730610306646e-05,
+      "loss": 0.2759,
+      "step": 1866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912617148561608e-05,
+      "loss": 0.3264,
+      "step": 1867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125036164765502e-05,
+      "loss": 0.2915,
+      "step": 1868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912390014060222e-05,
+      "loss": 0.3107,
+      "step": 1869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91227634132138e-05,
+      "loss": 0.3126,
+      "step": 1870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121625982687854e-05,
+      "loss": 0.3183,
+      "step": 1871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120487849112046e-05,
+      "loss": 0.3176,
+      "step": 1872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91193490125741e-05,
+      "loss": 0.3382,
+      "step": 1873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118209473161794e-05,
+      "loss": 0.2983,
+      "step": 1874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117069230962956e-05,
+      "loss": 0.3129,
+      "step": 1875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115928286065467e-05,
+      "loss": 0.2848,
+      "step": 1876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114786638557272e-05,
+      "loss": 0.2885,
+      "step": 1877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911364428852636e-05,
+      "loss": 0.2938,
+      "step": 1878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112501236060777e-05,
+      "loss": 0.2717,
+      "step": 1879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111357481248627e-05,
+      "loss": 0.3409,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110213024178062e-05,
+      "loss": 0.3215,
+      "step": 1881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109067864937292e-05,
+      "loss": 0.3062,
+      "step": 1882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910792200361458e-05,
+      "loss": 0.3286,
+      "step": 1883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9106775440298242e-05,
+      "loss": 0.2964,
+      "step": 1884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105628175076654e-05,
+      "loss": 0.2551,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104480208038236e-05,
+      "loss": 0.3083,
+      "step": 1886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103331539271473e-05,
+      "loss": 0.3112,
+      "step": 1887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102182168864894e-05,
+      "loss": 0.279,
+      "step": 1888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101032096907088e-05,
+      "loss": 0.2707,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90998813234867e-05,
+      "loss": 0.2793,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909872984869242e-05,
+      "loss": 0.3558,
+      "step": 1891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097577672613002e-05,
+      "loss": 0.2997,
+      "step": 1892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909642479533725e-05,
+      "loss": 0.288,
+      "step": 1893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9095271216954022e-05,
+      "loss": 0.2843,
+      "step": 1894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909411693755223e-05,
+      "loss": 0.327,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909296195722084e-05,
+      "loss": 0.3427,
+      "step": 1896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909180627604887e-05,
+      "loss": 0.2953,
+      "step": 1897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090649894125395e-05,
+      "loss": 0.2892,
+      "step": 1898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908949281153955e-05,
+      "loss": 0.3901,
+      "step": 1899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088335028380504e-05,
+      "loss": 0.3151,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087176544737507e-05,
+      "loss": 0.3711,
+      "step": 1901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086017360699843e-05,
+      "loss": 0.2921,
+      "step": 1902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084857476356852e-05,
+      "loss": 0.3477,
+      "step": 1903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908369689179794e-05,
+      "loss": 0.2933,
+      "step": 1904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082535607112554e-05,
+      "loss": 0.2968,
+      "step": 1905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081373622390204e-05,
+      "loss": 0.3433,
+      "step": 1906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080210937720443e-05,
+      "loss": 0.2717,
+      "step": 1907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907904755319289e-05,
+      "loss": 0.3653,
+      "step": 1908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077883468897215e-05,
+      "loss": 0.2932,
+      "step": 1909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076718684923136e-05,
+      "loss": 0.3005,
+      "step": 1910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075553201360432e-05,
+      "loss": 0.2938,
+      "step": 1911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907438701829893e-05,
+      "loss": 0.3301,
+      "step": 1912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073220135828513e-05,
+      "loss": 0.3149,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072052554039123e-05,
+      "loss": 0.3231,
+      "step": 1914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070884273020745e-05,
+      "loss": 0.3517,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906971529286343e-05,
+      "loss": 0.6,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068545613657273e-05,
+      "loss": 0.3317,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906737523549243e-05,
+      "loss": 0.3067,
+      "step": 1918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906620415845911e-05,
+      "loss": 0.2936,
+      "step": 1919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065032382647566e-05,
+      "loss": 0.3136,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063859908148123e-05,
+      "loss": 0.3419,
+      "step": 1921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906268673505114e-05,
+      "loss": 0.2871,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9061512863447046e-05,
+      "loss": 0.3053,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906033829342631e-05,
+      "loss": 0.2992,
+      "step": 1924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905916302507947e-05,
+      "loss": 0.3175,
+      "step": 1925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057987058497106e-05,
+      "loss": 0.3081,
+      "step": 1926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056810393769857e-05,
+      "loss": 0.3236,
+      "step": 1927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055633030988417e-05,
+      "loss": 0.2919,
+      "step": 1928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054454970243525e-05,
+      "loss": 0.3991,
+      "step": 1929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905327621162598e-05,
+      "loss": 0.338,
+      "step": 1930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052096755226643e-05,
+      "loss": 0.2745,
+      "step": 1931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050916601136418e-05,
+      "loss": 0.3105,
+      "step": 1932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904973574944626e-05,
+      "loss": 0.3568,
+      "step": 1933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048554200247184e-05,
+      "loss": 0.2861,
+      "step": 1934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047371953630262e-05,
+      "loss": 0.3148,
+      "step": 1935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904618900968662e-05,
+      "loss": 0.2726,
+      "step": 1936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045005368507418e-05,
+      "loss": 0.3517,
+      "step": 1937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90438210301839e-05,
+      "loss": 0.3078,
+      "step": 1938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042635994807344e-05,
+      "loss": 0.2863,
+      "step": 1939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041450262469087e-05,
+      "loss": 0.4042,
+      "step": 1940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040263833260513e-05,
+      "loss": 0.3239,
+      "step": 1941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903907670727308e-05,
+      "loss": 0.3154,
+      "step": 1942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037888884598272e-05,
+      "loss": 0.3273,
+      "step": 1943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036700365327648e-05,
+      "loss": 0.3491,
+      "step": 1944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035511149552816e-05,
+      "loss": 0.3012,
+      "step": 1945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034321237365424e-05,
+      "loss": 0.3345,
+      "step": 1946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033130628857194e-05,
+      "loss": 0.6304,
+      "step": 1947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903193932411989e-05,
+      "loss": 0.3242,
+      "step": 1948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903074732324533e-05,
+      "loss": 0.3118,
+      "step": 1949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029554626325386e-05,
+      "loss": 0.2962,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902836123345199e-05,
+      "loss": 0.2737,
+      "step": 1951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902716714471712e-05,
+      "loss": 0.3114,
+      "step": 1952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025972360212813e-05,
+      "loss": 0.3246,
+      "step": 1953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024776880031154e-05,
+      "loss": 0.3302,
+      "step": 1954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023580704264284e-05,
+      "loss": 0.2795,
+      "step": 1955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022383833004404e-05,
+      "loss": 0.3148,
+      "step": 1956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021186266343756e-05,
+      "loss": 0.2803,
+      "step": 1957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019988004374645e-05,
+      "loss": 0.291,
+      "step": 1958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901878904718943e-05,
+      "loss": 0.3284,
+      "step": 1959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017589394880515e-05,
+      "loss": 0.2957,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016389047540368e-05,
+      "loss": 0.2958,
+      "step": 1961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015188005261505e-05,
+      "loss": 0.3433,
+      "step": 1962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90139862681365e-05,
+      "loss": 0.2927,
+      "step": 1963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901278383625797e-05,
+      "loss": 0.2879,
+      "step": 1964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011580709718594e-05,
+      "loss": 0.283,
+      "step": 1965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010376888611106e-05,
+      "loss": 0.3064,
+      "step": 1966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009172373028286e-05,
+      "loss": 0.3329,
+      "step": 1967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007967163062978e-05,
+      "loss": 0.2832,
+      "step": 1968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900676125880807e-05,
+      "loss": 0.2922,
+      "step": 1969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005554660356505e-05,
+      "loss": 0.3642,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004347367801288e-05,
+      "loss": 0.2974,
+      "step": 1971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003139381235467e-05,
+      "loss": 0.2886,
+      "step": 1972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900193070075215e-05,
+      "loss": 0.2899,
+      "step": 1973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000721326444492e-05,
+      "loss": 0.6342,
+      "step": 1974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899951125840571e-05,
+      "loss": 0.2646,
+      "step": 1975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998300496729066e-05,
+      "loss": 0.2878,
+      "step": 1976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997089041507882e-05,
+      "loss": 0.2545,
+      "step": 1977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899587689283553e-05,
+      "loss": 0.2688,
+      "step": 1978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994664050805437e-05,
+      "loss": 0.3099,
+      "step": 1979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993450515511086e-05,
+      "loss": 0.3057,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992236287046008e-05,
+      "loss": 0.6141,
+      "step": 1981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991021365503782e-05,
+      "loss": 0.3133,
+      "step": 1982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989805750978062e-05,
+      "loss": 0.3036,
+      "step": 1983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898858944356253e-05,
+      "loss": 0.2885,
+      "step": 1984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898737244335094e-05,
+      "loss": 0.2827,
+      "step": 1985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898615475043709e-05,
+      "loss": 0.2995,
+      "step": 1986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984936364914835e-05,
+      "loss": 0.3311,
+      "step": 1987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983717286878078e-05,
+      "loss": 0.2932,
+      "step": 1988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898249751642078e-05,
+      "loss": 0.3197,
+      "step": 1989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981277053636963e-05,
+      "loss": 0.3235,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980055898620688e-05,
+      "loss": 0.3088,
+      "step": 1991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978834051466073e-05,
+      "loss": 0.3177,
+      "step": 1992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977611512267294e-05,
+      "loss": 0.2531,
+      "step": 1993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976388281118584e-05,
+      "loss": 0.3004,
+      "step": 1994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975164358114216e-05,
+      "loss": 0.3114,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973939743348527e-05,
+      "loss": 0.3235,
+      "step": 1996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972714436915905e-05,
+      "loss": 0.3309,
+      "step": 1997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897148843891079e-05,
+      "loss": 0.3483,
+      "step": 1998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970261749427674e-05,
+      "loss": 0.3162,
+      "step": 1999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969034368561105e-05,
+      "loss": 0.3036,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967806296405686e-05,
+      "loss": 0.2573,
+      "step": 2001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896657753305607e-05,
+      "loss": 0.2863,
+      "step": 2002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965348078606962e-05,
+      "loss": 0.317,
+      "step": 2003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896411793315312e-05,
+      "loss": 0.3018,
+      "step": 2004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962887096789363e-05,
+      "loss": 0.3557,
+      "step": 2005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961655569610557e-05,
+      "loss": 0.3107,
+      "step": 2006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960423351711622e-05,
+      "loss": 0.3047,
+      "step": 2007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959190443187525e-05,
+      "loss": 0.3517,
+      "step": 2008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89579568441333e-05,
+      "loss": 0.3871,
+      "step": 2009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956722554644026e-05,
+      "loss": 0.305,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895548757481483e-05,
+      "loss": 0.2921,
+      "step": 2011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954251904740904e-05,
+      "loss": 0.2544,
+      "step": 2012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953015544517482e-05,
+      "loss": 0.2961,
+      "step": 2013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951778494239862e-05,
+      "loss": 0.2822,
+      "step": 2014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895054075400339e-05,
+      "loss": 0.2968,
+      "step": 2015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894930232390346e-05,
+      "loss": 0.3514,
+      "step": 2016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894806320403553e-05,
+      "loss": 0.3347,
+      "step": 2017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89468233944951e-05,
+      "loss": 0.2739,
+      "step": 2018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894558289537773e-05,
+      "loss": 0.2888,
+      "step": 2019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944341706779033e-05,
+      "loss": 0.3325,
+      "step": 2020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894309982879467e-05,
+      "loss": 0.313,
+      "step": 2021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941857261520363e-05,
+      "loss": 0.2653,
+      "step": 2022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940614005051882e-05,
+      "loss": 0.3132,
+      "step": 2023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893937005948505e-05,
+      "loss": 0.3065,
+      "step": 2024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938125424915744e-05,
+      "loss": 0.3253,
+      "step": 2025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936880101439893e-05,
+      "loss": 0.3522,
+      "step": 2026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935634089153486e-05,
+      "loss": 0.3019,
+      "step": 2027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934387388152554e-05,
+      "loss": 0.3073,
+      "step": 2028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893313999853319e-05,
+      "loss": 0.2911,
+      "step": 2029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931891920391533e-05,
+      "loss": 0.2897,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930643153823777e-05,
+      "loss": 0.3073,
+      "step": 2031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929393698926177e-05,
+      "loss": 0.2939,
+      "step": 2032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928143555795034e-05,
+      "loss": 0.3099,
+      "step": 2033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89268927245267e-05,
+      "loss": 0.3145,
+      "step": 2034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925641205217583e-05,
+      "loss": 0.2874,
+      "step": 2035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924388997964147e-05,
+      "loss": 0.3191,
+      "step": 2036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923136102862902e-05,
+      "loss": 0.3145,
+      "step": 2037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921882520010416e-05,
+      "loss": 0.3859,
+      "step": 2038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892062824950331e-05,
+      "loss": 0.312,
+      "step": 2039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919373291438257e-05,
+      "loss": 0.2589,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918117645911985e-05,
+      "loss": 0.3284,
+      "step": 2041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916861313021268e-05,
+      "loss": 0.3251,
+      "step": 2042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891560429286294e-05,
+      "loss": 0.3199,
+      "step": 2043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891434658553389e-05,
+      "loss": 0.3227,
+      "step": 2044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913088191131047e-05,
+      "loss": 0.2993,
+      "step": 2045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891182910975141e-05,
+      "loss": 0.3227,
+      "step": 2046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891056934149202e-05,
+      "loss": 0.3022,
+      "step": 2047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890930888644997e-05,
+      "loss": 0.3165,
+      "step": 2048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908047744722417e-05,
+      "loss": 0.3163,
+      "step": 2049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890678591640656e-05,
+      "loss": 0.3312,
+      "step": 2050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905523401599655e-05,
+      "loss": 0.3138,
+      "step": 2051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890426020039901e-05,
+      "loss": 0.3002,
+      "step": 2052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890299631290198e-05,
+      "loss": 0.3511,
+      "step": 2053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901731739205992e-05,
+      "loss": 0.3078,
+      "step": 2054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900466479408505e-05,
+      "loss": 0.3206,
+      "step": 2055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899200533607037e-05,
+      "loss": 0.3278,
+      "step": 2056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897933901899165e-05,
+      "loss": 0.3536,
+      "step": 2057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896666584382516e-05,
+      "loss": 0.318,
+      "step": 2058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895398581154763e-05,
+      "loss": 0.2911,
+      "step": 2059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894129892313643e-05,
+      "loss": 0.2946,
+      "step": 2060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892860517956938e-05,
+      "loss": 0.287,
+      "step": 2061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891590458182486e-05,
+      "loss": 0.6182,
+      "step": 2062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890319713088178e-05,
+      "loss": 0.3275,
+      "step": 2063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888904828277195e-05,
+      "loss": 0.2731,
+      "step": 2064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887776167331803e-05,
+      "loss": 0.3074,
+      "step": 2065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886503366865786e-05,
+      "loss": 0.2817,
+      "step": 2066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885229881472002e-05,
+      "loss": 0.3303,
+      "step": 2067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88839557112486e-05,
+      "loss": 0.3032,
+      "step": 2068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882680856293785e-05,
+      "loss": 0.3172,
+      "step": 2069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881405316705824e-05,
+      "loss": 0.2908,
+      "step": 2070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880129092583027e-05,
+      "loss": 0.3937,
+      "step": 2071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878852184023754e-05,
+      "loss": 0.2861,
+      "step": 2072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877574591126427e-05,
+      "loss": 0.3243,
+      "step": 2073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876296313989516e-05,
+      "loss": 0.2978,
+      "step": 2074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875017352711547e-05,
+      "loss": 0.3172,
+      "step": 2075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887373770739109e-05,
+      "loss": 0.3356,
+      "step": 2076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872457378126778e-05,
+      "loss": 0.3309,
+      "step": 2077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8871176365017293e-05,
+      "loss": 0.3307,
+      "step": 2078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869894668161365e-05,
+      "loss": 0.2736,
+      "step": 2079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868612287657783e-05,
+      "loss": 0.2706,
+      "step": 2080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886732922360539e-05,
+      "loss": 0.3145,
+      "step": 2081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866045476103073e-05,
+      "loss": 0.3381,
+      "step": 2082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864761045249777e-05,
+      "loss": 0.3324,
+      "step": 2083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863475931144506e-05,
+      "loss": 0.3838,
+      "step": 2084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862190133886303e-05,
+      "loss": 0.3732,
+      "step": 2085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860903653574277e-05,
+      "loss": 0.2498,
+      "step": 2086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859616490307578e-05,
+      "loss": 0.3399,
+      "step": 2087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858328644185414e-05,
+      "loss": 0.3198,
+      "step": 2088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857040115307055e-05,
+      "loss": 0.3476,
+      "step": 2089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855750903771805e-05,
+      "loss": 0.3046,
+      "step": 2090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885446100967903e-05,
+      "loss": 0.3245,
+      "step": 2091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853170433128155e-05,
+      "loss": 0.3167,
+      "step": 2092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851879174218645e-05,
+      "loss": 0.2946,
+      "step": 2093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885058723305003e-05,
+      "loss": 0.3139,
+      "step": 2094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884929460972188e-05,
+      "loss": 0.3494,
+      "step": 2095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848001304333828e-05,
+      "loss": 0.2764,
+      "step": 2096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846707316985556e-05,
+      "loss": 0.3117,
+      "step": 2097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845412647776795e-05,
+      "loss": 0.2834,
+      "step": 2098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844117296807332e-05,
+      "loss": 0.2692,
+      "step": 2099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842821264177012e-05,
+      "loss": 0.2645,
+      "step": 2100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884152454998572e-05,
+      "loss": 0.3059,
+      "step": 2101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840227154333405e-05,
+      "loss": 0.3278,
+      "step": 2102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883892907732006e-05,
+      "loss": 0.3282,
+      "step": 2103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883763031904574e-05,
+      "loss": 0.6036,
+      "step": 2104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883633087961054e-05,
+      "loss": 0.2932,
+      "step": 2105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835030759114617e-05,
+      "loss": 0.2885,
+      "step": 2106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883372995765818e-05,
+      "loss": 0.2918,
+      "step": 2107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832428475341486e-05,
+      "loss": 0.2858,
+      "step": 2108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831126312264843e-05,
+      "loss": 0.2882,
+      "step": 2109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829823468528624e-05,
+      "loss": 0.331,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882851994423324e-05,
+      "loss": 0.3467,
+      "step": 2111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827215739479163e-05,
+      "loss": 0.2709,
+      "step": 2112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825910854366914e-05,
+      "loss": 0.319,
+      "step": 2113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824605288997064e-05,
+      "loss": 0.306,
+      "step": 2114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823299043470243e-05,
+      "loss": 0.3165,
+      "step": 2115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882199211788713e-05,
+      "loss": 0.3169,
+      "step": 2116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820684512348455e-05,
+      "loss": 0.3154,
+      "step": 2117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819376226955e-05,
+      "loss": 0.3041,
+      "step": 2118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818067261807606e-05,
+      "loss": 0.303,
+      "step": 2119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881675761700716e-05,
+      "loss": 0.2918,
+      "step": 2120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815447292654598e-05,
+      "loss": 0.3339,
+      "step": 2121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881413628885092e-05,
+      "loss": 0.2827,
+      "step": 2122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812824605697163e-05,
+      "loss": 0.3256,
+      "step": 2123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811512243294436e-05,
+      "loss": 0.3273,
+      "step": 2124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810199201743884e-05,
+      "loss": 0.2935,
+      "step": 2125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880888548114671e-05,
+      "loss": 0.2778,
+      "step": 2126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807571081604167e-05,
+      "loss": 0.3264,
+      "step": 2127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806256003217566e-05,
+      "loss": 0.3231,
+      "step": 2128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804940246088265e-05,
+      "loss": 0.3083,
+      "step": 2129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803623810317678e-05,
+      "loss": 0.3403,
+      "step": 2130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802306696007265e-05,
+      "loss": 0.2906,
+      "step": 2131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880098890325855e-05,
+      "loss": 0.3035,
+      "step": 2132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799670432173093e-05,
+      "loss": 0.2933,
+      "step": 2133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879835128285252e-05,
+      "loss": 0.3176,
+      "step": 2134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797031455398504e-05,
+      "loss": 0.2634,
+      "step": 2135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879571094991277e-05,
+      "loss": 0.3088,
+      "step": 2136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87943897664971e-05,
+      "loss": 0.2938,
+      "step": 2137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793067905253318e-05,
+      "loss": 0.3134,
+      "step": 2138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791745366283313e-05,
+      "loss": 0.2851,
+      "step": 2139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879042214968901e-05,
+      "loss": 0.317,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878909825557241e-05,
+      "loss": 0.2835,
+      "step": 2141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878777368403554e-05,
+      "loss": 0.3166,
+      "step": 2142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87864484351805e-05,
+      "loss": 0.2952,
+      "step": 2143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785122509109425e-05,
+      "loss": 0.2638,
+      "step": 2144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783795905924516e-05,
+      "loss": 0.2883,
+      "step": 2145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782468625728027e-05,
+      "loss": 0.3054,
+      "step": 2146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781140668622243e-05,
+      "loss": 0.3151,
+      "step": 2147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8779812034709532e-05,
+      "loss": 0.2928,
+      "step": 2148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877848272409229e-05,
+      "loss": 0.2657,
+      "step": 2149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877715273687297e-05,
+      "loss": 0.3471,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775822073154093e-05,
+      "loss": 0.3234,
+      "step": 2151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774490733038214e-05,
+      "loss": 0.3738,
+      "step": 2152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877315871662794e-05,
+      "loss": 0.3361,
+      "step": 2153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771826024025944e-05,
+      "loss": 0.302,
+      "step": 2154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770492655334938e-05,
+      "loss": 0.3044,
+      "step": 2155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87691586106577e-05,
+      "loss": 0.2828,
+      "step": 2156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767823890097044e-05,
+      "loss": 0.3239,
+      "step": 2157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8766488493755845e-05,
+      "loss": 0.3028,
+      "step": 2158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876515242173703e-05,
+      "loss": 0.3709,
+      "step": 2159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763815674143574e-05,
+      "loss": 0.2885,
+      "step": 2160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762478251078508e-05,
+      "loss": 0.2751,
+      "step": 2161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876114015264492e-05,
+      "loss": 0.2958,
+      "step": 2162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759801378945938e-05,
+      "loss": 0.3157,
+      "step": 2163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758461930084745e-05,
+      "loss": 0.2953,
+      "step": 2164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757121806164588e-05,
+      "loss": 0.2949,
+      "step": 2165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875578100728875e-05,
+      "loss": 0.3612,
+      "step": 2166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754439533560576e-05,
+      "loss": 0.3277,
+      "step": 2167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875309738508346e-05,
+      "loss": 0.319,
+      "step": 2168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751754561960847e-05,
+      "loss": 0.286,
+      "step": 2169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750411064296237e-05,
+      "loss": 0.2993,
+      "step": 2170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874906689219318e-05,
+      "loss": 0.3162,
+      "step": 2171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747722045755273e-05,
+      "loss": 0.2867,
+      "step": 2172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874637652508618e-05,
+      "loss": 0.3014,
+      "step": 2173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87450303302896e-05,
+      "loss": 0.2891,
+      "step": 2174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874368346146929e-05,
+      "loss": 0.2783,
+      "step": 2175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742335918729066e-05,
+      "loss": 0.3362,
+      "step": 2176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740987702172787e-05,
+      "loss": 0.3497,
+      "step": 2177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739638811904363e-05,
+      "loss": 0.3555,
+      "step": 2178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738289248027764e-05,
+      "loss": 0.2637,
+      "step": 2179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736939010647008e-05,
+      "loss": 0.3163,
+      "step": 2180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873558809986616e-05,
+      "loss": 0.3246,
+      "step": 2181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873423651578935e-05,
+      "loss": 0.3236,
+      "step": 2182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732884258520745e-05,
+      "loss": 0.3394,
+      "step": 2183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731531328164572e-05,
+      "loss": 0.3182,
+      "step": 2184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730177724825107e-05,
+      "loss": 0.2806,
+      "step": 2185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872882344860668e-05,
+      "loss": 0.3619,
+      "step": 2186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727468499613675e-05,
+      "loss": 0.2708,
+      "step": 2187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726112877950517e-05,
+      "loss": 0.3092,
+      "step": 2188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87247565837217e-05,
+      "loss": 0.2755,
+      "step": 2189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723399617031754e-05,
+      "loss": 0.3241,
+      "step": 2190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722041977985264e-05,
+      "loss": 0.2998,
+      "step": 2191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720683666686882e-05,
+      "loss": 0.311,
+      "step": 2192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871932468324129e-05,
+      "loss": 0.3071,
+      "step": 2193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717965027753235e-05,
+      "loss": 0.283,
+      "step": 2194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716604700327516e-05,
+      "loss": 0.3267,
+      "step": 2195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871524370106897e-05,
+      "loss": 0.3349,
+      "step": 2196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871388203008251e-05,
+      "loss": 0.3089,
+      "step": 2197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712519687473075e-05,
+      "loss": 0.3096,
+      "step": 2198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711156673345675e-05,
+      "loss": 0.3047,
+      "step": 2199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709792987805357e-05,
+      "loss": 0.3143,
+      "step": 2200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708428630957236e-05,
+      "loss": 0.3167,
+      "step": 2201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707063602906466e-05,
+      "loss": 0.2883,
+      "step": 2202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705697903758254e-05,
+      "loss": 0.2916,
+      "step": 2203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704331533617866e-05,
+      "loss": 0.2983,
+      "step": 2204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702964492590613e-05,
+      "loss": 0.3063,
+      "step": 2205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701596780781855e-05,
+      "loss": 0.2296,
+      "step": 2206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700228398297017e-05,
+      "loss": 0.3269,
+      "step": 2207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698859345241562e-05,
+      "loss": 0.3207,
+      "step": 2208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697489621721013e-05,
+      "loss": 0.3031,
+      "step": 2209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696119227840937e-05,
+      "loss": 0.2792,
+      "step": 2210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869474816370696e-05,
+      "loss": 0.5715,
+      "step": 2211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693376429424756e-05,
+      "loss": 0.274,
+      "step": 2212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692004025100054e-05,
+      "loss": 0.2643,
+      "step": 2213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869063095083863e-05,
+      "loss": 0.3497,
+      "step": 2214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689257206746312e-05,
+      "loss": 0.2688,
+      "step": 2215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687882792928987e-05,
+      "loss": 0.2903,
+      "step": 2216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868650770949258e-05,
+      "loss": 0.3087,
+      "step": 2217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685131956543082e-05,
+      "loss": 0.3093,
+      "step": 2218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683755534186528e-05,
+      "loss": 0.3425,
+      "step": 2219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682378442529005e-05,
+      "loss": 0.3351,
+      "step": 2220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681000681676652e-05,
+      "loss": 0.3079,
+      "step": 2221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867962225173566e-05,
+      "loss": 0.2826,
+      "step": 2222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678243152812273e-05,
+      "loss": 0.3386,
+      "step": 2223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676863385012785e-05,
+      "loss": 0.3022,
+      "step": 2224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867548294844354e-05,
+      "loss": 0.3186,
+      "step": 2225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674101843210935e-05,
+      "loss": 0.278,
+      "step": 2226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867272006942142e-05,
+      "loss": 0.6252,
+      "step": 2227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671337627181497e-05,
+      "loss": 0.3765,
+      "step": 2228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669954516597717e-05,
+      "loss": 0.3115,
+      "step": 2229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866857073777668e-05,
+      "loss": 0.3462,
+      "step": 2230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866718629082504e-05,
+      "loss": 0.3608,
+      "step": 2231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866580117584951e-05,
+      "loss": 0.3091,
+      "step": 2232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664415392956848e-05,
+      "loss": 0.3033,
+      "step": 2233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663028942253854e-05,
+      "loss": 0.326,
+      "step": 2234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86616418238474e-05,
+      "loss": 0.3051,
+      "step": 2235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.2784,
+      "step": 2236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658865584351787e-05,
+      "loss": 0.273,
+      "step": 2237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865747646347661e-05,
+      "loss": 0.338,
+      "step": 2238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865608667532593e-05,
+      "loss": 0.3135,
+      "step": 2239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865469622000686e-05,
+      "loss": 0.2948,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653305097626565e-05,
+      "loss": 0.3037,
+      "step": 2241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865191330829227e-05,
+      "loss": 0.3492,
+      "step": 2242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865052085211125e-05,
+      "loss": 0.6414,
+      "step": 2243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649127729190825e-05,
+      "loss": 0.3001,
+      "step": 2244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647733939638373e-05,
+      "loss": 0.3071,
+      "step": 2245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864633948356132e-05,
+      "loss": 0.2802,
+      "step": 2246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864494436106714e-05,
+      "loss": 0.3492,
+      "step": 2247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643548572263364e-05,
+      "loss": 0.2723,
+      "step": 2248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642152117257574e-05,
+      "loss": 0.3063,
+      "step": 2249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640754996157397e-05,
+      "loss": 0.3141,
+      "step": 2250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639357209070526e-05,
+      "loss": 0.2906,
+      "step": 2251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637958756104687e-05,
+      "loss": 0.2592,
+      "step": 2252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863655963736767e-05,
+      "loss": 0.308,
+      "step": 2253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863515985296731e-05,
+      "loss": 0.325,
+      "step": 2254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633759403011496e-05,
+      "loss": 0.2852,
+      "step": 2255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632358287608167e-05,
+      "loss": 0.3242,
+      "step": 2256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630956506865313e-05,
+      "loss": 0.2997,
+      "step": 2257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629554060890982e-05,
+      "loss": 0.3124,
+      "step": 2258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8628150949793262e-05,
+      "loss": 0.3447,
+      "step": 2259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626747173680302e-05,
+      "loss": 0.3174,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862534273266029e-05,
+      "loss": 0.3788,
+      "step": 2261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623937626841485e-05,
+      "loss": 0.3508,
+      "step": 2262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862253185633218e-05,
+      "loss": 0.2986,
+      "step": 2263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8621125421240722e-05,
+      "loss": 0.2567,
+      "step": 2264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619718321675514e-05,
+      "loss": 0.2867,
+      "step": 2265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861831055774501e-05,
+      "loss": 0.2954,
+      "step": 2266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616902129557718e-05,
+      "loss": 0.3412,
+      "step": 2267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861549303722218e-05,
+      "loss": 0.3212,
+      "step": 2268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614083280847013e-05,
+      "loss": 0.3067,
+      "step": 2269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612672860540865e-05,
+      "loss": 0.3444,
+      "step": 2270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611261776412455e-05,
+      "loss": 0.308,
+      "step": 2271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609850028570536e-05,
+      "loss": 0.279,
+      "step": 2272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608437617123917e-05,
+      "loss": 0.3302,
+      "step": 2273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8607024542181465e-05,
+      "loss": 0.3188,
+      "step": 2274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605610803852086e-05,
+      "loss": 0.3171,
+      "step": 2275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604196402244752e-05,
+      "loss": 0.2758,
+      "step": 2276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602781337468472e-05,
+      "loss": 0.2822,
+      "step": 2277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8601365609632315e-05,
+      "loss": 0.2995,
+      "step": 2278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599949218845394e-05,
+      "loss": 0.333,
+      "step": 2279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598532165216882e-05,
+      "loss": 0.2821,
+      "step": 2280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597114448856e-05,
+      "loss": 0.3109,
+      "step": 2281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595696069872013e-05,
+      "loss": 0.2556,
+      "step": 2282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594277028374245e-05,
+      "loss": 0.3195,
+      "step": 2283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592857324472073e-05,
+      "loss": 0.2954,
+      "step": 2284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591436958274914e-05,
+      "loss": 0.6501,
+      "step": 2285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590015929892245e-05,
+      "loss": 0.2958,
+      "step": 2286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85885942394336e-05,
+      "loss": 0.5961,
+      "step": 2287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858717188700854e-05,
+      "loss": 0.274,
+      "step": 2288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585748872726704e-05,
+      "loss": 0.3195,
+      "step": 2289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584325196697767e-05,
+      "loss": 0.2817,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858290085903146e-05,
+      "loss": 0.2811,
+      "step": 2291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858147585983757e-05,
+      "loss": 0.3346,
+      "step": 2292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858005019922592e-05,
+      "loss": 0.2809,
+      "step": 2293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578623877306394e-05,
+      "loss": 0.3316,
+      "step": 2294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577196894188926e-05,
+      "loss": 0.3046,
+      "step": 2295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857576924998351e-05,
+      "loss": 0.3535,
+      "step": 2296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574340944800165e-05,
+      "loss": 0.2656,
+      "step": 2297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572911978748993e-05,
+      "loss": 0.3254,
+      "step": 2298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8571482351940124e-05,
+      "loss": 0.296,
+      "step": 2299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857005206448375e-05,
+      "loss": 0.3737,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568621116490108e-05,
+      "loss": 0.3925,
+      "step": 2301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856718950806949e-05,
+      "loss": 0.2954,
+      "step": 2302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565757239332232e-05,
+      "loss": 0.3333,
+      "step": 2303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564324310388735e-05,
+      "loss": 0.3148,
+      "step": 2304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8562890721349434e-05,
+      "loss": 0.3179,
+      "step": 2305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856145647232483e-05,
+      "loss": 0.3184,
+      "step": 2306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560021563425462e-05,
+      "loss": 0.2841,
+      "step": 2307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558585994761932e-05,
+      "loss": 0.3315,
+      "step": 2308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855714976644488e-05,
+      "loss": 0.3156,
+      "step": 2309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555712878585005e-05,
+      "loss": 0.3032,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855427533129306e-05,
+      "loss": 0.2994,
+      "step": 2311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552837124679835e-05,
+      "loss": 0.334,
+      "step": 2312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855139825885619e-05,
+      "loss": 0.2925,
+      "step": 2313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854995873393302e-05,
+      "loss": 0.3335,
+      "step": 2314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548518550021274e-05,
+      "loss": 0.327,
+      "step": 2315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547077707231963e-05,
+      "loss": 0.3067,
+      "step": 2316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545636205676133e-05,
+      "loss": 0.2688,
+      "step": 2317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544194045464888e-05,
+      "loss": 0.2933,
+      "step": 2318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542751226709382e-05,
+      "loss": 0.3052,
+      "step": 2319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8541307749520828e-05,
+      "loss": 0.2762,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853986361401047e-05,
+      "loss": 0.2846,
+      "step": 2321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538418820289628e-05,
+      "loss": 0.3182,
+      "step": 2322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853697336846965e-05,
+      "loss": 0.3107,
+      "step": 2323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535527258661944e-05,
+      "loss": 0.3061,
+      "step": 2324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534080490977977e-05,
+      "loss": 0.3303,
+      "step": 2325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853263306552925e-05,
+      "loss": 0.2715,
+      "step": 2326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853118498242733e-05,
+      "loss": 0.2714,
+      "step": 2327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529736241783825e-05,
+      "loss": 0.3378,
+      "step": 2328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528286843710398e-05,
+      "loss": 0.2717,
+      "step": 2329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852683678831876e-05,
+      "loss": 0.3435,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525386075720675e-05,
+      "loss": 0.3502,
+      "step": 2331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523934706027952e-05,
+      "loss": 0.3094,
+      "step": 2332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522482679352464e-05,
+      "loss": 0.3027,
+      "step": 2333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521029995806123e-05,
+      "loss": 0.3154,
+      "step": 2334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851957665550089e-05,
+      "loss": 0.2781,
+      "step": 2335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851812265854879e-05,
+      "loss": 0.3049,
+      "step": 2336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851666800506188e-05,
+      "loss": 0.2771,
+      "step": 2337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515212695152284e-05,
+      "loss": 0.3619,
+      "step": 2338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851375672893217e-05,
+      "loss": 0.311,
+      "step": 2339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851230010651375e-05,
+      "loss": 0.6145,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510842828009303e-05,
+      "loss": 0.293,
+      "step": 2341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850938489353114e-05,
+      "loss": 0.279,
+      "step": 2342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850792630319164e-05,
+      "loss": 0.3413,
+      "step": 2343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506467057103217e-05,
+      "loss": 0.2908,
+      "step": 2344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505007155378347e-05,
+      "loss": 0.2761,
+      "step": 2345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503546598129547e-05,
+      "loss": 0.2833,
+      "step": 2346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502085385469396e-05,
+      "loss": 0.2955,
+      "step": 2347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850062351751051e-05,
+      "loss": 0.304,
+      "step": 2348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499160994365568e-05,
+      "loss": 0.2961,
+      "step": 2349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849769781614729e-05,
+      "loss": 0.2924,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849623398296846e-05,
+      "loss": 0.4272,
+      "step": 2351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8494769494941883e-05,
+      "loss": 0.3087,
+      "step": 2352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493304352180455e-05,
+      "loss": 0.3331,
+      "step": 2353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491838554797096e-05,
+      "loss": 0.3338,
+      "step": 2354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490372102904778e-05,
+      "loss": 0.2849,
+      "step": 2355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848890499661653e-05,
+      "loss": 0.2883,
+      "step": 2356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848743723604543e-05,
+      "loss": 0.3171,
+      "step": 2357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485968821304604e-05,
+      "loss": 0.2959,
+      "step": 2358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8484499752507234e-05,
+      "loss": 0.3009,
+      "step": 2359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483030029766548e-05,
+      "loss": 0.3013,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848155965319582e-05,
+      "loss": 0.3588,
+      "step": 2361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8480088622908382e-05,
+      "loss": 0.3114,
+      "step": 2362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478616939017615e-05,
+      "loss": 0.3317,
+      "step": 2363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847714460163695e-05,
+      "loss": 0.3439,
+      "step": 2364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475671610879864e-05,
+      "loss": 0.3274,
+      "step": 2365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847419796685989e-05,
+      "loss": 0.3441,
+      "step": 2366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847272366969061e-05,
+      "loss": 0.2842,
+      "step": 2367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471248719485654e-05,
+      "loss": 0.2921,
+      "step": 2368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469773116358705e-05,
+      "loss": 0.2912,
+      "step": 2369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468296860423494e-05,
+      "loss": 0.3018,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466819951793805e-05,
+      "loss": 0.2886,
+      "step": 2371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846534239058347e-05,
+      "loss": 0.3138,
+      "step": 2372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846386417690637e-05,
+      "loss": 0.3499,
+      "step": 2373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462385310876444e-05,
+      "loss": 0.3368,
+      "step": 2374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460905792607667e-05,
+      "loss": 0.2726,
+      "step": 2375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459425622214082e-05,
+      "loss": 0.298,
+      "step": 2376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457944799809765e-05,
+      "loss": 0.2847,
+      "step": 2377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845646332550886e-05,
+      "loss": 0.2843,
+      "step": 2378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454981199425542e-05,
+      "loss": 0.2782,
+      "step": 2379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453498421674055e-05,
+      "loss": 0.2857,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845201499236868e-05,
+      "loss": 0.3391,
+      "step": 2381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450530911623747e-05,
+      "loss": 0.3425,
+      "step": 2382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844904617955365e-05,
+      "loss": 0.264,
+      "step": 2383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447560796272817e-05,
+      "loss": 0.2936,
+      "step": 2384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446074761895746e-05,
+      "loss": 0.2835,
+      "step": 2385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844458807653696e-05,
+      "loss": 0.2734,
+      "step": 2386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844310074031105e-05,
+      "loss": 0.2825,
+      "step": 2387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441612753332658e-05,
+      "loss": 0.2701,
+      "step": 2388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440124115716463e-05,
+      "loss": 0.373,
+      "step": 2389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843863482757721e-05,
+      "loss": 0.3769,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8437144889029675e-05,
+      "loss": 0.351,
+      "step": 2391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435654300188705e-05,
+      "loss": 0.3513,
+      "step": 2392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434163061169178e-05,
+      "loss": 0.3262,
+      "step": 2393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432671172086044e-05,
+      "loss": 0.3374,
+      "step": 2394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431178633054275e-05,
+      "loss": 0.2755,
+      "step": 2395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429685444188922e-05,
+      "loss": 0.289,
+      "step": 2396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428191605605067e-05,
+      "loss": 0.307,
+      "step": 2397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426697117417848e-05,
+      "loss": 0.3223,
+      "step": 2398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425201979742455e-05,
+      "loss": 0.2998,
+      "step": 2399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423706192694118e-05,
+      "loss": 0.3238,
+      "step": 2400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8422209756388132e-05,
+      "loss": 0.3258,
+      "step": 2401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420712670939837e-05,
+      "loss": 0.3314,
+      "step": 2402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419214936464613e-05,
+      "loss": 0.2272,
+      "step": 2403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417716553077903e-05,
+      "loss": 0.2912,
+      "step": 2404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416217520895198e-05,
+      "loss": 0.259,
+      "step": 2405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841471784003203e-05,
+      "loss": 0.3311,
+      "step": 2406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413217510603988e-05,
+      "loss": 0.2832,
+      "step": 2407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411716532726707e-05,
+      "loss": 0.3456,
+      "step": 2408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410214906515887e-05,
+      "loss": 0.2956,
+      "step": 2409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408712632087256e-05,
+      "loss": 0.2783,
+      "step": 2410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407209709556603e-05,
+      "loss": 0.3012,
+      "step": 2411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405706139039766e-05,
+      "loss": 0.3292,
+      "step": 2412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404201920652635e-05,
+      "loss": 0.311,
+      "step": 2413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402697054511145e-05,
+      "loss": 0.2782,
+      "step": 2414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401191540731286e-05,
+      "loss": 0.3084,
+      "step": 2415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839968537942909e-05,
+      "loss": 0.251,
+      "step": 2416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839817857072066e-05,
+      "loss": 0.3433,
+      "step": 2417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396671114722112e-05,
+      "loss": 0.3035,
+      "step": 2418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839516301154965e-05,
+      "loss": 0.2882,
+      "step": 2419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393654261319504e-05,
+      "loss": 0.3397,
+      "step": 2420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839214486414796e-05,
+      "loss": 0.2822,
+      "step": 2421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8390634820151353e-05,
+      "loss": 0.3478,
+      "step": 2422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389124129446078e-05,
+      "loss": 0.3356,
+      "step": 2423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387612792148566e-05,
+      "loss": 0.2992,
+      "step": 2424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386100808375305e-05,
+      "loss": 0.3185,
+      "step": 2425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384588178242828e-05,
+      "loss": 0.2957,
+      "step": 2426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383074901867728e-05,
+      "loss": 0.3258,
+      "step": 2427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381560979366633e-05,
+      "loss": 0.3,
+      "step": 2428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380046410856234e-05,
+      "loss": 0.2985,
+      "step": 2429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8378531196453265e-05,
+      "loss": 0.3295,
+      "step": 2430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8377015336274507e-05,
+      "loss": 0.6105,
+      "step": 2431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375498830436805e-05,
+      "loss": 0.3095,
+      "step": 2432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8373981679057036e-05,
+      "loss": 0.3124,
+      "step": 2433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372463882252133e-05,
+      "loss": 0.3447,
+      "step": 2434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837094544013909e-05,
+      "loss": 0.3058,
+      "step": 2435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369426352834927e-05,
+      "loss": 0.3349,
+      "step": 2436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367906620456737e-05,
+      "loss": 0.2932,
+      "step": 2437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366386243121654e-05,
+      "loss": 0.2755,
+      "step": 2438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364865220946856e-05,
+      "loss": 0.3089,
+      "step": 2439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363343554049582e-05,
+      "loss": 0.3532,
+      "step": 2440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836182124254711e-05,
+      "loss": 0.2551,
+      "step": 2441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8360298286556774e-05,
+      "loss": 0.361,
+      "step": 2442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358774686195956e-05,
+      "loss": 0.2964,
+      "step": 2443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8357250441582085e-05,
+      "loss": 0.3021,
+      "step": 2444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355725552832648e-05,
+      "loss": 0.2558,
+      "step": 2445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354200020065168e-05,
+      "loss": 0.2985,
+      "step": 2446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352673843397232e-05,
+      "loss": 0.2946,
+      "step": 2447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351147022946468e-05,
+      "loss": 0.3085,
+      "step": 2448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349619558830553e-05,
+      "loss": 0.6111,
+      "step": 2449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348091451167224e-05,
+      "loss": 0.3421,
+      "step": 2450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346562700074253e-05,
+      "loss": 0.2653,
+      "step": 2451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834503330566947e-05,
+      "loss": 0.293,
+      "step": 2452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343503268070752e-05,
+      "loss": 0.315,
+      "step": 2453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341972587396032e-05,
+      "loss": 0.2822,
+      "step": 2454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340441263763282e-05,
+      "loss": 0.3389,
+      "step": 2455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833890929729053e-05,
+      "loss": 0.3104,
+      "step": 2456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337376688095854e-05,
+      "loss": 0.325,
+      "step": 2457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833584343629738e-05,
+      "loss": 0.3056,
+      "step": 2458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334309542013282e-05,
+      "loss": 0.2958,
+      "step": 2459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332775005361786e-05,
+      "loss": 0.3148,
+      "step": 2460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331239826461165e-05,
+      "loss": 0.3084,
+      "step": 2461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329704005429745e-05,
+      "loss": 0.3241,
+      "step": 2462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328167542385898e-05,
+      "loss": 0.3234,
+      "step": 2463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326630437448045e-05,
+      "loss": 0.2882,
+      "step": 2464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325092690734663e-05,
+      "loss": 0.3796,
+      "step": 2465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8323554302364273e-05,
+      "loss": 0.3371,
+      "step": 2466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322015272455445e-05,
+      "loss": 0.3239,
+      "step": 2467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83204756011268e-05,
+      "loss": 0.2818,
+      "step": 2468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318935288497007e-05,
+      "loss": 0.312,
+      "step": 2469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831739433468479e-05,
+      "loss": 0.2676,
+      "step": 2470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315852739808914e-05,
+      "loss": 0.3077,
+      "step": 2471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314310503988198e-05,
+      "loss": 0.2911,
+      "step": 2472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831276762734151e-05,
+      "loss": 0.3268,
+      "step": 2473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311224109987768e-05,
+      "loss": 0.2565,
+      "step": 2474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830967995204594e-05,
+      "loss": 0.3039,
+      "step": 2475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308135153635037e-05,
+      "loss": 0.2855,
+      "step": 2476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830658971487413e-05,
+      "loss": 0.6318,
+      "step": 2477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305043635882334e-05,
+      "loss": 0.293,
+      "step": 2478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830349691677881e-05,
+      "loss": 0.258,
+      "step": 2479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830194955768277e-05,
+      "loss": 0.2902,
+      "step": 2480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830040155871348e-05,
+      "loss": 0.3034,
+      "step": 2481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298852919990254e-05,
+      "loss": 0.3347,
+      "step": 2482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297303641632448e-05,
+      "loss": 0.2749,
+      "step": 2483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829575372375948e-05,
+      "loss": 0.293,
+      "step": 2484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294203166490797e-05,
+      "loss": 0.3049,
+      "step": 2485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8292651969945923e-05,
+      "loss": 0.2784,
+      "step": 2486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291100134244407e-05,
+      "loss": 0.2871,
+      "step": 2487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289547659505867e-05,
+      "loss": 0.3049,
+      "step": 2488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287994545849948e-05,
+      "loss": 0.295,
+      "step": 2489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828644079339636e-05,
+      "loss": 0.3129,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284886402264864e-05,
+      "loss": 0.2702,
+      "step": 2491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283331372575258e-05,
+      "loss": 0.3613,
+      "step": 2492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281775704447402e-05,
+      "loss": 0.2748,
+      "step": 2493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280219398001192e-05,
+      "loss": 0.2891,
+      "step": 2494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8278662453356588e-05,
+      "loss": 0.3527,
+      "step": 2495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277104870633588e-05,
+      "loss": 0.2952,
+      "step": 2496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827554664995224e-05,
+      "loss": 0.306,
+      "step": 2497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827398779143265e-05,
+      "loss": 0.2878,
+      "step": 2498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272428295194965e-05,
+      "loss": 0.3098,
+      "step": 2499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270868161359377e-05,
+      "loss": 0.327,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269307390046143e-05,
+      "loss": 0.3403,
+      "step": 2501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267745981375555e-05,
+      "loss": 0.3116,
+      "step": 2502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826618393546796e-05,
+      "loss": 0.6324,
+      "step": 2503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826462125244375e-05,
+      "loss": 0.5909,
+      "step": 2504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826305793242337e-05,
+      "loss": 0.2809,
+      "step": 2505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261493975527312e-05,
+      "loss": 0.2873,
+      "step": 2506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.825992938187612e-05,
+      "loss": 0.3083,
+      "step": 2507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8258364151590386e-05,
+      "loss": 0.2934,
+      "step": 2508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825679828479075e-05,
+      "loss": 0.3502,
+      "step": 2509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82552317815979e-05,
+      "loss": 0.2945,
+      "step": 2510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253664642132576e-05,
+      "loss": 0.2768,
+      "step": 2511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825209686651556e-05,
+      "loss": 0.3544,
+      "step": 2512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250528454867695e-05,
+      "loss": 0.3334,
+      "step": 2513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248959407309862e-05,
+      "loss": 0.2967,
+      "step": 2514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247389723962998e-05,
+      "loss": 0.2784,
+      "step": 2515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245819404948088e-05,
+      "loss": 0.2604,
+      "step": 2516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244248450386156e-05,
+      "loss": 0.265,
+      "step": 2517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242676860398295e-05,
+      "loss": 0.2751,
+      "step": 2518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241104635105627e-05,
+      "loss": 0.2861,
+      "step": 2519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823953177462934e-05,
+      "loss": 0.3376,
+      "step": 2520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823795827909065e-05,
+      "loss": 0.2474,
+      "step": 2521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236384148610843e-05,
+      "loss": 0.2532,
+      "step": 2522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823480938331124e-05,
+      "loss": 0.2949,
+      "step": 2523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233233983313224e-05,
+      "loss": 0.325,
+      "step": 2524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231657948738212e-05,
+      "loss": 0.3331,
+      "step": 2525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230081279707675e-05,
+      "loss": 0.3224,
+      "step": 2526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228503976343147e-05,
+      "loss": 0.3739,
+      "step": 2527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226926038766185e-05,
+      "loss": 0.2748,
+      "step": 2528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225347467098418e-05,
+      "loss": 0.3304,
+      "step": 2529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822376826146151e-05,
+      "loss": 0.2908,
+      "step": 2530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8222188421977178e-05,
+      "loss": 0.3076,
+      "step": 2531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220607948767187e-05,
+      "loss": 0.3163,
+      "step": 2532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821902684195336e-05,
+      "loss": 0.2876,
+      "step": 2533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217445101657553e-05,
+      "loss": 0.3032,
+      "step": 2534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821586272800168e-05,
+      "loss": 0.319,
+      "step": 2535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214279721107705e-05,
+      "loss": 0.3348,
+      "step": 2536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212696081097636e-05,
+      "loss": 0.301,
+      "step": 2537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211111808093534e-05,
+      "loss": 0.3395,
+      "step": 2538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209526902217506e-05,
+      "loss": 0.3055,
+      "step": 2539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820794136359171e-05,
+      "loss": 0.3011,
+      "step": 2540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820635519233835e-05,
+      "loss": 0.3148,
+      "step": 2541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820476838857968e-05,
+      "loss": 0.3375,
+      "step": 2542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8203180952438e-05,
+      "loss": 0.3148,
+      "step": 2543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820159288403567e-05,
+      "loss": 0.3963,
+      "step": 2544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200004183495085e-05,
+      "loss": 0.2939,
+      "step": 2545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198414850938694e-05,
+      "loss": 0.3329,
+      "step": 2546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196824886488996e-05,
+      "loss": 0.3519,
+      "step": 2547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195234290268536e-05,
+      "loss": 0.2755,
+      "step": 2548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193643062399913e-05,
+      "loss": 0.2719,
+      "step": 2549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192051203005768e-05,
+      "loss": 0.3317,
+      "step": 2550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190458712208795e-05,
+      "loss": 0.3207,
+      "step": 2551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8188865590131733e-05,
+      "loss": 0.2989,
+      "step": 2552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187271836897377e-05,
+      "loss": 0.295,
+      "step": 2553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185677452628557e-05,
+      "loss": 0.2838,
+      "step": 2554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818408243744817e-05,
+      "loss": 0.6395,
+      "step": 2555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182486791479145e-05,
+      "loss": 0.2875,
+      "step": 2556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818089051484447e-05,
+      "loss": 0.2908,
+      "step": 2557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179293607667177e-05,
+      "loss": 0.2991,
+      "step": 2558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817769607007035e-05,
+      "loss": 0.276,
+      "step": 2559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176097902177115e-05,
+      "loss": 0.2872,
+      "step": 2560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174499104110653e-05,
+      "loss": 0.3059,
+      "step": 2561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817289967599419e-05,
+      "loss": 0.3077,
+      "step": 2562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171299617951007e-05,
+      "loss": 0.299,
+      "step": 2563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816969893010442e-05,
+      "loss": 0.2706,
+      "step": 2564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816809761257781e-05,
+      "loss": 0.3182,
+      "step": 2565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81664956654946e-05,
+      "loss": 0.3344,
+      "step": 2566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816489308897825e-05,
+      "loss": 0.3079,
+      "step": 2567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163289883152285e-05,
+      "loss": 0.2955,
+      "step": 2568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161686048140275e-05,
+      "loss": 0.2844,
+      "step": 2569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160081584065833e-05,
+      "loss": 0.2552,
+      "step": 2570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158476491052616e-05,
+      "loss": 0.2571,
+      "step": 2571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815687076922435e-05,
+      "loss": 0.3284,
+      "step": 2572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155264418704785e-05,
+      "loss": 0.3126,
+      "step": 2573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153657439617738e-05,
+      "loss": 0.3306,
+      "step": 2574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152049832087063e-05,
+      "loss": 0.2721,
+      "step": 2575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150441596236667e-05,
+      "loss": 0.3096,
+      "step": 2576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148832732190508e-05,
+      "loss": 0.3085,
+      "step": 2577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814722324007258e-05,
+      "loss": 0.3082,
+      "step": 2578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145613120006947e-05,
+      "loss": 0.3274,
+      "step": 2579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144002372117705e-05,
+      "loss": 0.3143,
+      "step": 2580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142390996528994e-05,
+      "loss": 0.3163,
+      "step": 2581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814077899336502e-05,
+      "loss": 0.3377,
+      "step": 2582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813916636275003e-05,
+      "loss": 0.2891,
+      "step": 2583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137553104808308e-05,
+      "loss": 0.3035,
+      "step": 2584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135939219664205e-05,
+      "loss": 0.3094,
+      "step": 2585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813432470744211e-05,
+      "loss": 0.2516,
+      "step": 2586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132709568266457e-05,
+      "loss": 0.2946,
+      "step": 2587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131093802261738e-05,
+      "loss": 0.3115,
+      "step": 2588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812947740955248e-05,
+      "loss": 0.2675,
+      "step": 2589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127860390263275e-05,
+      "loss": 0.3096,
+      "step": 2590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8126242744518753e-05,
+      "loss": 0.2975,
+      "step": 2591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124624472443592e-05,
+      "loss": 0.2581,
+      "step": 2592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812300557416252e-05,
+      "loss": 0.2588,
+      "step": 2593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121386049800317e-05,
+      "loss": 0.3183,
+      "step": 2594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119765899481807e-05,
+      "loss": 0.2561,
+      "step": 2595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811814512333186e-05,
+      "loss": 0.3161,
+      "step": 2596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81165237214754e-05,
+      "loss": 0.2949,
+      "step": 2597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114901694037402e-05,
+      "loss": 0.3259,
+      "step": 2598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811327904114287e-05,
+      "loss": 0.2724,
+      "step": 2599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111655762916885e-05,
+      "loss": 0.3187,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110031859484554e-05,
+      "loss": 0.309,
+      "step": 2601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810840733097104e-05,
+      "loss": 0.3133,
+      "step": 2602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810678217750155e-05,
+      "loss": 0.3854,
+      "step": 2603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810515639920135e-05,
+      "loss": 0.2874,
+      "step": 2604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810352999619574e-05,
+      "loss": 0.2714,
+      "step": 2605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101902968610082e-05,
+      "loss": 0.3195,
+      "step": 2606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100275316569774e-05,
+      "loss": 0.2803,
+      "step": 2607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809864704020027e-05,
+      "loss": 0.3149,
+      "step": 2608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097018139627068e-05,
+      "loss": 0.2936,
+      "step": 2609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809538861497572e-05,
+      "loss": 0.3329,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093758466371812e-05,
+      "loss": 0.2933,
+      "step": 2611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8092127693940998e-05,
+      "loss": 0.2873,
+      "step": 2612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090496297808962e-05,
+      "loss": 0.2746,
+      "step": 2613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088864278101452e-05,
+      "loss": 0.3024,
+      "step": 2614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808723163494425e-05,
+      "loss": 0.3531,
+      "step": 2615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085598368463194e-05,
+      "loss": 0.2829,
+      "step": 2616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083964478784167e-05,
+      "loss": 0.2899,
+      "step": 2617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8082329966033105e-05,
+      "loss": 0.272,
+      "step": 2618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080694830335985e-05,
+      "loss": 0.2876,
+      "step": 2619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8079059071818828e-05,
+      "loss": 0.3261,
+      "step": 2620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077422690607725e-05,
+      "loss": 0.3115,
+      "step": 2621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807578568682879e-05,
+      "loss": 0.3017,
+      "step": 2622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074148060608197e-05,
+      "loss": 0.2876,
+      "step": 2623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807250981207217e-05,
+      "loss": 0.2832,
+      "step": 2624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807087094134697e-05,
+      "loss": 0.3242,
+      "step": 2625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069231448558923e-05,
+      "loss": 0.3752,
+      "step": 2626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067591333834382e-05,
+      "loss": 0.339,
+      "step": 2627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806595059729977e-05,
+      "loss": 0.2871,
+      "step": 2628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064309239081535e-05,
+      "loss": 0.3015,
+      "step": 2629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062667259306193e-05,
+      "loss": 0.2793,
+      "step": 2630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061024658100298e-05,
+      "loss": 0.291,
+      "step": 2631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805938143559045e-05,
+      "loss": 0.3187,
+      "step": 2632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057737591903306e-05,
+      "loss": 0.2576,
+      "step": 2633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056093127165564e-05,
+      "loss": 0.3286,
+      "step": 2634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054448041503966e-05,
+      "loss": 0.2805,
+      "step": 2635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052802335045315e-05,
+      "loss": 0.2883,
+      "step": 2636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805115600791645e-05,
+      "loss": 0.3074,
+      "step": 2637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804950906024426e-05,
+      "loss": 0.3187,
+      "step": 2638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047861492155687e-05,
+      "loss": 0.3398,
+      "step": 2639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046213303777717e-05,
+      "loss": 0.312,
+      "step": 2640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804456449523738e-05,
+      "loss": 0.3007,
+      "step": 2641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804291506666176e-05,
+      "loss": 0.3564,
+      "step": 2642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804126501817799e-05,
+      "loss": 0.31,
+      "step": 2643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039614349913245e-05,
+      "loss": 0.3165,
+      "step": 2644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037963061994756e-05,
+      "loss": 0.28,
+      "step": 2645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036311154549783e-05,
+      "loss": 0.3203,
+      "step": 2646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803465862770566e-05,
+      "loss": 0.2863,
+      "step": 2647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033005481589746e-05,
+      "loss": 0.2887,
+      "step": 2648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031351716329462e-05,
+      "loss": 0.2966,
+      "step": 2649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029697332052277e-05,
+      "loss": 0.2929,
+      "step": 2650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028042328885694e-05,
+      "loss": 0.29,
+      "step": 2651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8026386706957278e-05,
+      "loss": 0.2834,
+      "step": 2652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024730466394632e-05,
+      "loss": 0.3765,
+      "step": 2653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802307360732541e-05,
+      "loss": 0.2834,
+      "step": 2654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021416129877324e-05,
+      "loss": 0.3094,
+      "step": 2655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019758034178116e-05,
+      "loss": 0.3654,
+      "step": 2656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018099320355586e-05,
+      "loss": 0.3066,
+      "step": 2657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016439988537576e-05,
+      "loss": 0.3427,
+      "step": 2658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014780038851983e-05,
+      "loss": 0.3094,
+      "step": 2659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013119471426748e-05,
+      "loss": 0.2836,
+      "step": 2660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011458286389856e-05,
+      "loss": 0.3178,
+      "step": 2661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009796483869347e-05,
+      "loss": 0.2842,
+      "step": 2662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80081340639933e-05,
+      "loss": 0.2948,
+      "step": 2663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006471026889852e-05,
+      "loss": 0.3229,
+      "step": 2664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004807372687175e-05,
+      "loss": 0.2938,
+      "step": 2665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003143101513502e-05,
+      "loss": 0.3467,
+      "step": 2666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001478213497104e-05,
+      "loss": 0.37,
+      "step": 2667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79998127087663e-05,
+      "loss": 0.2593,
+      "step": 2668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998146587449457e-05,
+      "loss": 0.2997,
+      "step": 2669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996479849675e-05,
+      "loss": 0.3116,
+      "step": 2670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994812495571387e-05,
+      "loss": 0.3419,
+      "step": 2671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799314452526713e-05,
+      "loss": 0.3844,
+      "step": 2672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799147593889079e-05,
+      "loss": 0.3381,
+      "step": 2673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798980673657097e-05,
+      "loss": 0.2622,
+      "step": 2674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988136918436324e-05,
+      "loss": 0.3355,
+      "step": 2675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7986466484615557e-05,
+      "loss": 0.272,
+      "step": 2676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984795435237418e-05,
+      "loss": 0.3593,
+      "step": 2677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7983123770430696e-05,
+      "loss": 0.3203,
+      "step": 2678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798145149032424e-05,
+      "loss": 0.3458,
+      "step": 2679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797977859504694e-05,
+      "loss": 0.3336,
+      "step": 2680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797810508472774e-05,
+      "loss": 0.3352,
+      "step": 2681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976430959495617e-05,
+      "loss": 0.2817,
+      "step": 2682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797475621947961e-05,
+      "loss": 0.2972,
+      "step": 2683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7973080864808795e-05,
+      "loss": 0.2954,
+      "step": 2684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797140489561231e-05,
+      "loss": 0.299,
+      "step": 2685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969728312019316e-05,
+      "loss": 0.3123,
+      "step": 2686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968051114159046e-05,
+      "loss": 0.2874,
+      "step": 2687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796637330216077e-05,
+      "loss": 0.3176,
+      "step": 2688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964694876153802e-05,
+      "loss": 0.3162,
+      "step": 2689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963015836267502e-05,
+      "loss": 0.327,
+      "step": 2690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7961336182631293e-05,
+      "loss": 0.2814,
+      "step": 2691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795965591537463e-05,
+      "loss": 0.2719,
+      "step": 2692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957975034627017e-05,
+      "loss": 0.3038,
+      "step": 2693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795629354051801e-05,
+      "loss": 0.3014,
+      "step": 2694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795461143317721e-05,
+      "loss": 0.3409,
+      "step": 2695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7952928712734266e-05,
+      "loss": 0.3441,
+      "step": 2696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951245379318872e-05,
+      "loss": 0.2802,
+      "step": 2697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949561433060775e-05,
+      "loss": 0.3172,
+      "step": 2698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794787687408976e-05,
+      "loss": 0.3266,
+      "step": 2699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794619170253567e-05,
+      "loss": 0.2952,
+      "step": 2700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944505918528384e-05,
+      "loss": 0.3082,
+      "step": 2701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7942819522197837e-05,
+      "loss": 0.2697,
+      "step": 2702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794113251367401e-05,
+      "loss": 0.2893,
+      "step": 2703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7939444893086925e-05,
+      "loss": 0.2945,
+      "step": 2704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793775666056666e-05,
+      "loss": 0.3228,
+      "step": 2705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793606781624333e-05,
+      "loss": 0.2875,
+      "step": 2706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793437836024711e-05,
+      "loss": 0.2986,
+      "step": 2707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793268829270821e-05,
+      "loss": 0.2694,
+      "step": 2708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930997613756892e-05,
+      "loss": 0.3179,
+      "step": 2709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929306323523463e-05,
+      "loss": 0.3358,
+      "step": 2710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7927614422138286e-05,
+      "loss": 0.3011,
+      "step": 2711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792592190973176e-05,
+      "loss": 0.3267,
+      "step": 2712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924228786434333e-05,
+      "loss": 0.3232,
+      "step": 2713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792253505237651e-05,
+      "loss": 0.307,
+      "step": 2714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920840707688833e-05,
+      "loss": 0.2712,
+      "step": 2715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791914575250189e-05,
+      "loss": 0.2921,
+      "step": 2716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917450186946323e-05,
+      "loss": 0.29,
+      "step": 2717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915754011152815e-05,
+      "loss": 0.2966,
+      "step": 2718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914057225252103e-05,
+      "loss": 0.313,
+      "step": 2719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912359829374963e-05,
+      "loss": 0.2825,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910661823652223e-05,
+      "loss": 0.3318,
+      "step": 2721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790896320821476e-05,
+      "loss": 0.3194,
+      "step": 2722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790726398319349e-05,
+      "loss": 0.329,
+      "step": 2723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905564148719383e-05,
+      "loss": 0.3492,
+      "step": 2724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903863704923453e-05,
+      "loss": 0.2863,
+      "step": 2725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902162651936766e-05,
+      "loss": 0.3461,
+      "step": 2726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900460989890424e-05,
+      "loss": 0.3307,
+      "step": 2727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789875871891559e-05,
+      "loss": 0.3148,
+      "step": 2728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897055839143457e-05,
+      "loss": 0.2863,
+      "step": 2729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895352350705288e-05,
+      "loss": 0.305,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893648253732364e-05,
+      "loss": 0.6641,
+      "step": 2731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7891943548356043e-05,
+      "loss": 0.3036,
+      "step": 2732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890238234707708e-05,
+      "loss": 0.3477,
+      "step": 2733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7888532312918793e-05,
+      "loss": 0.2765,
+      "step": 2734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886825783120786e-05,
+      "loss": 0.3339,
+      "step": 2735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788511864544522e-05,
+      "loss": 0.3472,
+      "step": 2736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7883410900023667e-05,
+      "loss": 0.3389,
+      "step": 2737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788170254698776e-05,
+      "loss": 0.3372,
+      "step": 2738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787999358646916e-05,
+      "loss": 0.2847,
+      "step": 2739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7878284018599594e-05,
+      "loss": 0.2751,
+      "step": 2740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7876573843510822e-05,
+      "loss": 0.2642,
+      "step": 2741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874863061334658e-05,
+      "loss": 0.2982,
+      "step": 2742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787315167220296e-05,
+      "loss": 0.2791,
+      "step": 2743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871439676247632e-05,
+      "loss": 0.2979,
+      "step": 2744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786972707360063e-05,
+      "loss": 0.2662,
+      "step": 2745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786801386439395e-05,
+      "loss": 0.3167,
+      "step": 2746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786630004875964e-05,
+      "loss": 0.614,
+      "step": 2747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7864585626829786e-05,
+      "loss": 0.2914,
+      "step": 2748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862870598736534e-05,
+      "loss": 0.3242,
+      "step": 2749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786115496461207e-05,
+      "loss": 0.3099,
+      "step": 2750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859438724588623e-05,
+      "loss": 0.2908,
+      "step": 2751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857721878798476e-05,
+      "loss": 0.3257,
+      "step": 2752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785600442737395e-05,
+      "loss": 0.2869,
+      "step": 2753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785428637044742e-05,
+      "loss": 0.2927,
+      "step": 2754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852567708151306e-05,
+      "loss": 0.3333,
+      "step": 2755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850848440618075e-05,
+      "loss": 0.3088,
+      "step": 2756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849128567980238e-05,
+      "loss": 0.3142,
+      "step": 2757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847408090370355e-05,
+      "loss": 0.3382,
+      "step": 2758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845687007921034e-05,
+      "loss": 0.3124,
+      "step": 2759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784396532076492e-05,
+      "loss": 0.3123,
+      "step": 2760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784224302903472e-05,
+      "loss": 0.2925,
+      "step": 2761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840520132863173e-05,
+      "loss": 0.2896,
+      "step": 2762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783879663238308e-05,
+      "loss": 0.2937,
+      "step": 2763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837072527727275e-05,
+      "loss": 0.3057,
+      "step": 2764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835347819028642e-05,
+      "loss": 0.3079,
+      "step": 2765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833622506420116e-05,
+      "loss": 0.2964,
+      "step": 2766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783189659003467e-05,
+      "loss": 0.2938,
+      "step": 2767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783017007000534e-05,
+      "loss": 0.3206,
+      "step": 2768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828442946465188e-05,
+      "loss": 0.3374,
+      "step": 2769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7826715219547336e-05,
+      "loss": 0.3456,
+      "step": 2770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824986889384948e-05,
+      "loss": 0.3585,
+      "step": 2771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823257956111233e-05,
+      "loss": 0.2675,
+      "step": 2772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782152841985945e-05,
+      "loss": 0.3141,
+      "step": 2773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819798280762907e-05,
+      "loss": 0.3415,
+      "step": 2774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781806753895495e-05,
+      "loss": 0.3244,
+      "step": 2775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816336194568976e-05,
+      "loss": 0.2383,
+      "step": 2776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781460424773843e-05,
+      "loss": 0.2368,
+      "step": 2777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78128716985968e-05,
+      "loss": 0.3093,
+      "step": 2778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781113854727763e-05,
+      "loss": 0.2943,
+      "step": 2779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780940479391449e-05,
+      "loss": 0.2877,
+      "step": 2780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780767043864102e-05,
+      "loss": 0.308,
+      "step": 2781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780593548159089e-05,
+      "loss": 0.269,
+      "step": 2782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780419992289782e-05,
+      "loss": 0.2917,
+      "step": 2783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802463762695588e-05,
+      "loss": 0.3575,
+      "step": 2784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800727001118e-05,
+      "loss": 0.2968,
+      "step": 2785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779898963829892e-05,
+      "loss": 0.3095,
+      "step": 2786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797251674372253e-05,
+      "loss": 0.3192,
+      "step": 2787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795513109471952e-05,
+      "loss": 0.5713,
+      "step": 2788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793773943732026e-05,
+      "loss": 0.2745,
+      "step": 2789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792034177286508e-05,
+      "loss": 0.368,
+      "step": 2790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77902938102695e-05,
+      "loss": 0.2783,
+      "step": 2791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788552842815136e-05,
+      "loss": 0.3057,
+      "step": 2792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786811275057606e-05,
+      "loss": 0.3386,
+      "step": 2793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778506910713114e-05,
+      "loss": 0.3011,
+      "step": 2794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778332633917001e-05,
+      "loss": 0.3047,
+      "step": 2795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781582971308547e-05,
+      "loss": 0.322,
+      "step": 2796
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777983900368112e-05,
+      "loss": 0.3665,
+      "step": 2797
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777809443642214e-05,
+      "loss": 0.2877,
+      "step": 2798
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7776349269666076e-05,
+      "loss": 0.2598,
+      "step": 2799
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774603503547432e-05,
+      "loss": 0.3298,
+      "step": 2800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772857138200767e-05,
+      "loss": 0.3345,
+      "step": 2801
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777111017376068e-05,
+      "loss": 0.3273,
+      "step": 2802
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769362610361824e-05,
+      "loss": 0.3007,
+      "step": 2803
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767614448138882e-05,
+      "loss": 0.3427,
+      "step": 2804
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77658656872266e-05,
+      "loss": 0.3618,
+      "step": 2805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776411632775976e-05,
+      "loss": 0.3187,
+      "step": 2806
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762366369873204e-05,
+      "loss": 0.2852,
+      "step": 2807
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77606158137018e-05,
+      "loss": 0.3292,
+      "step": 2808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758864659380474e-05,
+      "loss": 0.314,
+      "step": 2809
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77571129070442e-05,
+      "loss": 0.3109,
+      "step": 2810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775536055682799e-05,
+      "loss": 0.3658,
+      "step": 2811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775360760886691e-05,
+      "loss": 0.2958,
+      "step": 2812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751854063296068e-05,
+      "loss": 0.2924,
+      "step": 2813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750099920250616e-05,
+      "loss": 0.2863,
+      "step": 2814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774834517986576e-05,
+      "loss": 0.3532,
+      "step": 2815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774658984227674e-05,
+      "loss": 0.3488,
+      "step": 2816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774483390761885e-05,
+      "loss": 0.308,
+      "step": 2817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743077376027433e-05,
+      "loss": 0.2943,
+      "step": 2818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741320247637875e-05,
+      "loss": 0.2876,
+      "step": 2819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739562522585598e-05,
+      "loss": 0.2805,
+      "step": 2820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737804201006084e-05,
+      "loss": 0.2848,
+      "step": 2821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773604528303486e-05,
+      "loss": 0.285,
+      "step": 2822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773428576880749e-05,
+      "loss": 0.3236,
+      "step": 2823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732525658459586e-05,
+      "loss": 0.3361,
+      "step": 2824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7730764952126813e-05,
+      "loss": 0.2708,
+      "step": 2825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729003649944878e-05,
+      "loss": 0.2791,
+      "step": 2826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772724175204953e-05,
+      "loss": 0.2586,
+      "step": 2827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772547925857657e-05,
+      "loss": 0.3179,
+      "step": 2828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723716169661843e-05,
+      "loss": 0.2417,
+      "step": 2829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7721952485441232e-05,
+      "loss": 0.3043,
+      "step": 2830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720188206050682e-05,
+      "loss": 0.2878,
+      "step": 2831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718423331626175e-05,
+      "loss": 0.2891,
+      "step": 2832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716657862303733e-05,
+      "loss": 0.3344,
+      "step": 2833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7714891798219432e-05,
+      "loss": 0.3486,
+      "step": 2834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771312513950939e-05,
+      "loss": 0.3138,
+      "step": 2835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711357886309777e-05,
+      "loss": 0.294,
+      "step": 2836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77095900387568e-05,
+      "loss": 0.2626,
+      "step": 2837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707821596986715e-05,
+      "loss": 0.3339,
+      "step": 2838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7706052561135826e-05,
+      "loss": 0.3048,
+      "step": 2839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7704282931340488e-05,
+      "loss": 0.6313,
+      "step": 2840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702512707737086e-05,
+      "loss": 0.2995,
+      "step": 2841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770074189046206e-05,
+      "loss": 0.2975,
+      "step": 2842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698970479651904e-05,
+      "loss": 0.2623,
+      "step": 2843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697198475443146e-05,
+      "loss": 0.3027,
+      "step": 2844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769542587797236e-05,
+      "loss": 0.3005,
+      "step": 2845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693652687376173e-05,
+      "loss": 0.2664,
+      "step": 2846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691878903791252e-05,
+      "loss": 0.6333,
+      "step": 2847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690104527354313e-05,
+      "loss": 0.3271,
+      "step": 2848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768832955820211e-05,
+      "loss": 0.283,
+      "step": 2849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768655399647146e-05,
+      "loss": 0.3414,
+      "step": 2850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684777842299206e-05,
+      "loss": 0.3184,
+      "step": 2851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7683001095822245e-05,
+      "loss": 0.3103,
+      "step": 2852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7681223757177526e-05,
+      "loss": 0.3055,
+      "step": 2853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679445826502033e-05,
+      "loss": 0.306,
+      "step": 2854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7677667303932797e-05,
+      "loss": 0.3315,
+      "step": 2855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675888189606907e-05,
+      "loss": 0.3152,
+      "step": 2856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767410848366148e-05,
+      "loss": 0.2598,
+      "step": 2857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7672328186233692e-05,
+      "loss": 0.3183,
+      "step": 2858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7670547297460758e-05,
+      "loss": 0.2806,
+      "step": 2859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668765817479937e-05,
+      "loss": 0.2607,
+      "step": 2860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766698374642854e-05,
+      "loss": 0.3216,
+      "step": 2861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766520108444392e-05,
+      "loss": 0.262,
+      "step": 2862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663417831663474e-05,
+      "loss": 0.3249,
+      "step": 2863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766163398822465e-05,
+      "loss": 0.3141,
+      "step": 2864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765984955426493e-05,
+      "loss": 0.3157,
+      "step": 2865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765806452992186e-05,
+      "loss": 0.3539,
+      "step": 2866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656278915333017e-05,
+      "loss": 0.2562,
+      "step": 2867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765449271063602e-05,
+      "loss": 0.3212,
+      "step": 2868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7652705915968552e-05,
+      "loss": 0.3331,
+      "step": 2869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650918531468326e-05,
+      "loss": 0.321,
+      "step": 2870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76491305572731e-05,
+      "loss": 0.2733,
+      "step": 2871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647341993520687e-05,
+      "loss": 0.3022,
+      "step": 2872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764555284034894e-05,
+      "loss": 0.3089,
+      "step": 2873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764376309789576e-05,
+      "loss": 0.3743,
+      "step": 2874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641972766299088e-05,
+      "loss": 0.2706,
+      "step": 2875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7640181845696914e-05,
+      "loss": 0.2991,
+      "step": 2876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638390336227275e-05,
+      "loss": 0.3155,
+      "step": 2877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636598238028253e-05,
+      "loss": 0.2684,
+      "step": 2878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763480555123797e-05,
+      "loss": 0.2991,
+      "step": 2879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76330122759946e-05,
+      "loss": 0.2953,
+      "step": 2880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7631218412436362e-05,
+      "loss": 0.2729,
+      "step": 2881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629423960701513e-05,
+      "loss": 0.3105,
+      "step": 2882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627628920928366e-05,
+      "loss": 0.3604,
+      "step": 2883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7625833293255268e-05,
+      "loss": 0.311,
+      "step": 2884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762403707782062e-05,
+      "loss": 0.3134,
+      "step": 2885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762224027476287e-05,
+      "loss": 0.3066,
+      "step": 2886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7620442884220495e-05,
+      "loss": 0.2988,
+      "step": 2887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761864490633204e-05,
+      "loss": 0.3286,
+      "step": 2888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616846341236082e-05,
+      "loss": 0.2817,
+      "step": 2889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761504718907124e-05,
+      "loss": 0.2957,
+      "step": 2890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613247449976195e-05,
+      "loss": 0.2975,
+      "step": 2891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761144712408965e-05,
+      "loss": 0.2997,
+      "step": 2892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760964621155037e-05,
+      "loss": 0.3035,
+      "step": 2893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760784471249716e-05,
+      "loss": 0.3041,
+      "step": 2894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760604262706887e-05,
+      "loss": 0.3328,
+      "step": 2895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76042399554044e-05,
+      "loss": 0.3155,
+      "step": 2896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7602436697642686e-05,
+      "loss": 0.3151,
+      "step": 2897
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7600632853922713e-05,
+      "loss": 0.265,
+      "step": 2898
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598828424383522e-05,
+      "loss": 0.3282,
+      "step": 2899
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759702340916418e-05,
+      "loss": 0.288,
+      "step": 2900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595217808403814e-05,
+      "loss": 0.2635,
+      "step": 2901
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593411622241584e-05,
+      "loss": 0.3057,
+      "step": 2902
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591604850816705e-05,
+      "loss": 0.3115,
+      "step": 2903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758979749426844e-05,
+      "loss": 0.2652,
+      "step": 2904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587989552736085e-05,
+      "loss": 0.3064,
+      "step": 2905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586181026358987e-05,
+      "loss": 0.3276,
+      "step": 2906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758437191527654e-05,
+      "loss": 0.2938,
+      "step": 2907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582562219628185e-05,
+      "loss": 0.2686,
+      "step": 2908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580751939553396e-05,
+      "loss": 0.2975,
+      "step": 2909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757894107519171e-05,
+      "loss": 0.3219,
+      "step": 2910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757712962668269e-05,
+      "loss": 0.3459,
+      "step": 2911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575317594165963e-05,
+      "loss": 0.3243,
+      "step": 2912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757350497778118e-05,
+      "loss": 0.3334,
+      "step": 2913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757169177766806e-05,
+      "loss": 0.2744,
+      "step": 2914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756987799396635e-05,
+      "loss": 0.2953,
+      "step": 2915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568063626815844e-05,
+      "loss": 0.3466,
+      "step": 2916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566248676356394e-05,
+      "loss": 0.3323,
+      "step": 2917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564433142727882e-05,
+      "loss": 0.3196,
+      "step": 2918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7562617026070238e-05,
+      "loss": 0.2961,
+      "step": 2919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560800326523442e-05,
+      "loss": 0.3482,
+      "step": 2920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558983044227513e-05,
+      "loss": 0.2996,
+      "step": 2921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7557165179322522e-05,
+      "loss": 0.3297,
+      "step": 2922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555346731948587e-05,
+      "loss": 0.2964,
+      "step": 2923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755352770224585e-05,
+      "loss": 0.2969,
+      "step": 2924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755170809035452e-05,
+      "loss": 0.2765,
+      "step": 2925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7549887896414853e-05,
+      "loss": 0.3168,
+      "step": 2926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548067120567126e-05,
+      "loss": 0.3132,
+      "step": 2927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754624576295168e-05,
+      "loss": 0.2957,
+      "step": 2928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544423823708903e-05,
+      "loss": 0.2717,
+      "step": 2929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542601302979213e-05,
+      "loss": 0.3103,
+      "step": 2930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7540778200903082e-05,
+      "loss": 0.3256,
+      "step": 2931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753895451762103e-05,
+      "loss": 0.2991,
+      "step": 2932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7537130253273613e-05,
+      "loss": 0.2603,
+      "step": 2933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753530540800144e-05,
+      "loss": 0.3221,
+      "step": 2934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533479981945157e-05,
+      "loss": 0.3535,
+      "step": 2935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531653975245463e-05,
+      "loss": 0.3038,
+      "step": 2936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7529827388043093e-05,
+      "loss": 0.2844,
+      "step": 2937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528000220478836e-05,
+      "loss": 0.5901,
+      "step": 2938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526172472693518e-05,
+      "loss": 0.3172,
+      "step": 2939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524344144828015e-05,
+      "loss": 0.327,
+      "step": 2940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522515237023242e-05,
+      "loss": 0.2545,
+      "step": 2941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520685749420164e-05,
+      "loss": 0.274,
+      "step": 2942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518855682159793e-05,
+      "loss": 0.2668,
+      "step": 2943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7517025035383175e-05,
+      "loss": 0.2944,
+      "step": 2944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751519380923141e-05,
+      "loss": 0.2792,
+      "step": 2945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751336200384564e-05,
+      "loss": 0.2963,
+      "step": 2946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7511529619367055e-05,
+      "loss": 0.2906,
+      "step": 2947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509696655936878e-05,
+      "loss": 0.2549,
+      "step": 2948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750786311369639e-05,
+      "loss": 0.3553,
+      "step": 2949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506028992786912e-05,
+      "loss": 0.2766,
+      "step": 2950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504194293349805e-05,
+      "loss": 0.3689,
+      "step": 2951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7502359015526488e-05,
+      "loss": 0.3091,
+      "step": 2952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75005231594584e-05,
+      "loss": 0.2937,
+      "step": 2953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749868672528705e-05,
+      "loss": 0.3443,
+      "step": 2954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749684971315398e-05,
+      "loss": 0.3121,
+      "step": 2955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749501212320078e-05,
+      "loss": 0.2676,
+      "step": 2956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749317395556908e-05,
+      "loss": 0.2698,
+      "step": 2957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491335210400554e-05,
+      "loss": 0.3367,
+      "step": 2958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7489495887836922e-05,
+      "loss": 0.2939,
+      "step": 2959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7487655988019957e-05,
+      "loss": 0.2722,
+      "step": 2960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485815511091466e-05,
+      "loss": 0.3036,
+      "step": 2961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483974457193307e-05,
+      "loss": 0.3098,
+      "step": 2962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748213282646737e-05,
+      "loss": 0.2928,
+      "step": 2963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748029061905561e-05,
+      "loss": 0.295,
+      "step": 2964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747844783510001e-05,
+      "loss": 0.3275,
+      "step": 2965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74766044747426e-05,
+      "loss": 0.2881,
+      "step": 2966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747476053812546e-05,
+      "loss": 0.301,
+      "step": 2967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7472916025390714e-05,
+      "loss": 0.2794,
+      "step": 2968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7471070936680527e-05,
+      "loss": 0.2684,
+      "step": 2969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7469225272137104e-05,
+      "loss": 0.3401,
+      "step": 2970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467379031902707e-05,
+      "loss": 0.3375,
+      "step": 2971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7465532216119628e-05,
+      "loss": 0.3051,
+      "step": 2972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463684824930215e-05,
+      "loss": 0.3447,
+      "step": 2973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7461836858476858e-05,
+      "loss": 0.2483,
+      "step": 2974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459988316901984e-05,
+      "loss": 0.2654,
+      "step": 2975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745813920034807e-05,
+      "loss": 0.6151,
+      "step": 2976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745628950895764e-05,
+      "loss": 0.3342,
+      "step": 2977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454439242873257e-05,
+      "loss": 0.2723,
+      "step": 2978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452588402237525e-05,
+      "loss": 0.613,
+      "step": 2979
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450736987193113e-05,
+      "loss": 0.3793,
+      "step": 2980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448884997882706e-05,
+      "loss": 0.3612,
+      "step": 2981
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7447032434449045e-05,
+      "loss": 0.2988,
+      "step": 2982
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445179297034925e-05,
+      "loss": 0.3465,
+      "step": 2983
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744332558578317e-05,
+      "loss": 0.3279,
+      "step": 2984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744147130083666e-05,
+      "loss": 0.3226,
+      "step": 2985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743961644233831e-05,
+      "loss": 0.3051,
+      "step": 2986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437761010431083e-05,
+      "loss": 0.2554,
+      "step": 2987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743590500525799e-05,
+      "loss": 0.2751,
+      "step": 2988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434048426962086e-05,
+      "loss": 0.2797,
+      "step": 2989
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432191275686454e-05,
+      "loss": 0.3446,
+      "step": 2990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430333551574247e-05,
+      "loss": 0.6001,
+      "step": 2991
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742847525476864e-05,
+      "loss": 0.2774,
+      "step": 2992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742661638541287e-05,
+      "loss": 0.3793,
+      "step": 2993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7424756943650203e-05,
+      "loss": 0.3013,
+      "step": 2994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7422896929623957e-05,
+      "loss": 0.2703,
+      "step": 2995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421036343477498e-05,
+      "loss": 0.3277,
+      "step": 2996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741917518535422e-05,
+      "loss": 0.4084,
+      "step": 2997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741731345539758e-05,
+      "loss": 0.3613,
+      "step": 2998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415451153751068e-05,
+      "loss": 0.3193,
+      "step": 2999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7413588280558223e-05,
+      "loss": 0.3335,
+      "step": 3000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411724835962623e-05,
+      "loss": 0.2513,
+      "step": 3001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74098608201079e-05,
+      "loss": 0.292,
+      "step": 3002
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407996233137713e-05,
+      "loss": 0.2734,
+      "step": 3003
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406131075195784e-05,
+      "loss": 0.2944,
+      "step": 3004
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404265346425867e-05,
+      "loss": 0.3594,
+      "step": 3005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740239904697176e-05,
+      "loss": 0.269,
+      "step": 3006
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740053217697731e-05,
+      "loss": 0.2882,
+      "step": 3007
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739866473658641e-05,
+      "loss": 0.2902,
+      "step": 3008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396796725942986e-05,
+      "loss": 0.3079,
+      "step": 3009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739492814519102e-05,
+      "loss": 0.2763,
+      "step": 3010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7393058994474535e-05,
+      "loss": 0.2678,
+      "step": 3011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739118927393759e-05,
+      "loss": 0.3147,
+      "step": 3012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73893189837243e-05,
+      "loss": 0.3236,
+      "step": 3013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7387448123978813e-05,
+      "loss": 0.2949,
+      "step": 3014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385576694845324e-05,
+      "loss": 0.3339,
+      "step": 3015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738370469646808e-05,
+      "loss": 0.307,
+      "step": 3016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738183212899136e-05,
+      "loss": 0.2856,
+      "step": 3017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7379958992559494e-05,
+      "loss": 0.2893,
+      "step": 3018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7378085287316853e-05,
+      "loss": 0.3495,
+      "step": 3019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737621101340786e-05,
+      "loss": 0.3138,
+      "step": 3020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7374336170976964e-05,
+      "loss": 0.2986,
+      "step": 3021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7372460760168676e-05,
+      "loss": 0.3237,
+      "step": 3022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737058478112754e-05,
+      "loss": 0.2622,
+      "step": 3023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7368708233998148e-05,
+      "loss": 0.2722,
+      "step": 3024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366831118925133e-05,
+      "loss": 0.3397,
+      "step": 3025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736495343605318e-05,
+      "loss": 0.3392,
+      "step": 3026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7363075185527007e-05,
+      "loss": 0.2745,
+      "step": 3027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361196367491378e-05,
+      "loss": 0.2632,
+      "step": 3028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735931698209111e-05,
+      "loss": 0.302,
+      "step": 3029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735743702947105e-05,
+      "loss": 0.3304,
+      "step": 3030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355556509776093e-05,
+      "loss": 0.3356,
+      "step": 3031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7353675423151194e-05,
+      "loss": 0.2728,
+      "step": 3032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351793769741326e-05,
+      "loss": 0.2748,
+      "step": 3033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734991154969152e-05,
+      "loss": 0.33,
+      "step": 3034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348028763146843e-05,
+      "loss": 0.3058,
+      "step": 3035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346145410252422e-05,
+      "loss": 0.3177,
+      "step": 3036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344261491153412e-05,
+      "loss": 0.2975,
+      "step": 3037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7342377005995014e-05,
+      "loss": 0.3472,
+      "step": 3038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340491954922474e-05,
+      "loss": 0.2886,
+      "step": 3039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733860633808109e-05,
+      "loss": 0.2708,
+      "step": 3040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336720155616186e-05,
+      "loss": 0.2965,
+      "step": 3041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7334833407673145e-05,
+      "loss": 0.6443,
+      "step": 3042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733294609439739e-05,
+      "loss": 0.2844,
+      "step": 3043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733105821593438e-05,
+      "loss": 0.283,
+      "step": 3044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329169772429628e-05,
+      "loss": 0.3101,
+      "step": 3045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7327280764028683e-05,
+      "loss": 0.3412,
+      "step": 3046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325391190877144e-05,
+      "loss": 0.6011,
+      "step": 3047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732350105312065e-05,
+      "loss": 0.3168,
+      "step": 3048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7321610350904877e-05,
+      "loss": 0.2737,
+      "step": 3049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319719084375556e-05,
+      "loss": 0.3342,
+      "step": 3050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317827253678456e-05,
+      "loss": 0.2621,
+      "step": 3051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731593485895939e-05,
+      "loss": 0.2642,
+      "step": 3052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314041900364215e-05,
+      "loss": 0.3472,
+      "step": 3053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731214837803883e-05,
+      "loss": 0.2865,
+      "step": 3054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310254292129175e-05,
+      "loss": 0.5972,
+      "step": 3055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730835964278124e-05,
+      "loss": 0.3008,
+      "step": 3056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730646443014106e-05,
+      "loss": 0.2531,
+      "step": 3057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7304568654354703e-05,
+      "loss": 0.3138,
+      "step": 3058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7302672315568284e-05,
+      "loss": 0.2844,
+      "step": 3059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730077541392797e-05,
+      "loss": 0.3058,
+      "step": 3060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298877949579962e-05,
+      "loss": 0.2876,
+      "step": 3061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7296979922670502e-05,
+      "loss": 0.324,
+      "step": 3062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7295081333345887e-05,
+      "loss": 0.3231,
+      "step": 3063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729318218175245e-05,
+      "loss": 0.2648,
+      "step": 3064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729128246803657e-05,
+      "loss": 0.317,
+      "step": 3065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728938219234466e-05,
+      "loss": 0.2813,
+      "step": 3066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287481354823187e-05,
+      "loss": 0.276,
+      "step": 3067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7285579955618663e-05,
+      "loss": 0.3404,
+      "step": 3068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283677994877634e-05,
+      "loss": 0.3178,
+      "step": 3069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281775472746695e-05,
+      "loss": 0.3302,
+      "step": 3070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279872389372484e-05,
+      "loss": 0.2679,
+      "step": 3071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727796874490168e-05,
+      "loss": 0.3051,
+      "step": 3072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7276064539481007e-05,
+      "loss": 0.3224,
+      "step": 3073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7274159773257227e-05,
+      "loss": 0.329,
+      "step": 3074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727225444637716e-05,
+      "loss": 0.3153,
+      "step": 3075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727034855898765e-05,
+      "loss": 0.335,
+      "step": 3076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72684421112356e-05,
+      "loss": 0.3243,
+      "step": 3077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266535103267943e-05,
+      "loss": 0.295,
+      "step": 3078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264627535231667e-05,
+      "loss": 0.3094,
+      "step": 3079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7262719407273795e-05,
+      "loss": 0.2799,
+      "step": 3080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72608107195414e-05,
+      "loss": 0.3039,
+      "step": 3081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7258901472181587e-05,
+      "loss": 0.2769,
+      "step": 3082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725699166534152e-05,
+      "loss": 0.3543,
+      "step": 3083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255081299168393e-05,
+      "loss": 0.6128,
+      "step": 3084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7253170373809447e-05,
+      "loss": 0.3118,
+      "step": 3085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7251258889411964e-05,
+      "loss": 0.3437,
+      "step": 3086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724934684612328e-05,
+      "loss": 0.2847,
+      "step": 3087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724743424409076e-05,
+      "loss": 0.2795,
+      "step": 3088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724552108346182e-05,
+      "loss": 0.297,
+      "step": 3089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243607364383916e-05,
+      "loss": 0.3001,
+      "step": 3090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7241693087004546e-05,
+      "loss": 0.303,
+      "step": 3091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239778251471255e-05,
+      "loss": 0.2795,
+      "step": 3092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723786285793163e-05,
+      "loss": 0.3063,
+      "step": 3093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72359469065333e-05,
+      "loss": 0.3005,
+      "step": 3094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234030397423935e-05,
+      "loss": 0.3278,
+      "step": 3095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723211333075125e-05,
+      "loss": 0.2925,
+      "step": 3096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723019570666301e-05,
+      "loss": 0.3046,
+      "step": 3097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228277525307007e-05,
+      "loss": 0.2824,
+      "step": 3098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226358786831087e-05,
+      "loss": 0.2944,
+      "step": 3099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722443949138314e-05,
+      "loss": 0.3724,
+      "step": 3100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222519639111094e-05,
+      "loss": 0.3059,
+      "step": 3101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7220599230162917e-05,
+      "loss": 0.29,
+      "step": 3102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218678264686634e-05,
+      "loss": 0.3082,
+      "step": 3103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72167567428303e-05,
+      "loss": 0.2892,
+      "step": 3104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7214834664742014e-05,
+      "loss": 0.2596,
+      "step": 3105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212912030569923e-05,
+      "loss": 0.2877,
+      "step": 3106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210988840462207e-05,
+      "loss": 0.2654,
+      "step": 3107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7209065094567107e-05,
+      "loss": 0.2755,
+      "step": 3108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207140793032892e-05,
+      "loss": 0.2919,
+      "step": 3109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720521593600787e-05,
+      "loss": 0.316,
+      "step": 3110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720329052364041e-05,
+      "loss": 0.308,
+      "step": 3111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201364556078908e-05,
+      "loss": 0.2589,
+      "step": 3112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199438033471812e-05,
+      "loss": 0.3171,
+      "step": 3113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71975109559676e-05,
+      "loss": 0.2607,
+      "step": 3114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195583323714812e-05,
+      "loss": 0.342,
+      "step": 3115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193655136862016e-05,
+      "loss": 0.3469,
+      "step": 3116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719172639555782e-05,
+      "loss": 0.2828,
+      "step": 3117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7189797099950895e-05,
+      "loss": 0.3389,
+      "step": 3118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187867250189936e-05,
+      "loss": 0.322,
+      "step": 3119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7185936846423686e-05,
+      "loss": 0.6334,
+      "step": 3120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718400588880093e-05,
+      "loss": 0.2861,
+      "step": 3121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182074377470494e-05,
+      "loss": 0.2823,
+      "step": 3122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180142312581253e-05,
+      "loss": 0.2714,
+      "step": 3123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717820969428212e-05,
+      "loss": 0.3376,
+      "step": 3124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7176276522722054e-05,
+      "loss": 0.2822,
+      "step": 3125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174342798050056e-05,
+      "loss": 0.319,
+      "step": 3126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717240852041516e-05,
+      "loss": 0.3087,
+      "step": 3127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717047368996646e-05,
+      "loss": 0.298,
+      "step": 3128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168538306853075e-05,
+      "loss": 0.2981,
+      "step": 3129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166602371224178e-05,
+      "loss": 0.2862,
+      "step": 3130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7164665883228982e-05,
+      "loss": 0.2809,
+      "step": 3131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716272884301674e-05,
+      "loss": 0.3267,
+      "step": 3132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160791250736754e-05,
+      "loss": 0.3401,
+      "step": 3133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7158853106538358e-05,
+      "loss": 0.27,
+      "step": 3134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156914410570937e-05,
+      "loss": 0.29,
+      "step": 3135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154975162983917e-05,
+      "loss": 0.2925,
+      "step": 3136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153035363926766e-05,
+      "loss": 0.2865,
+      "step": 3137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151095013548996e-05,
+      "loss": 0.3004,
+      "step": 3138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7149154112000154e-05,
+      "loss": 0.3009,
+      "step": 3139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7147212659429837e-05,
+      "loss": 0.2823,
+      "step": 3140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145270655987686e-05,
+      "loss": 0.2711,
+      "step": 3141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714332810182338e-05,
+      "loss": 0.3141,
+      "step": 3142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141384997086638e-05,
+      "loss": 0.299,
+      "step": 3143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7139441341927224e-05,
+      "loss": 0.2684,
+      "step": 3144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7137497136494953e-05,
+      "loss": 0.2828,
+      "step": 3145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713555238093967e-05,
+      "loss": 0.3049,
+      "step": 3146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133607075411266e-05,
+      "loss": 0.3228,
+      "step": 3147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131661220059675e-05,
+      "loss": 0.3062,
+      "step": 3148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129714815034876e-05,
+      "loss": 0.2823,
+      "step": 3149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7127767860486892e-05,
+      "loss": 0.3042,
+      "step": 3150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125820356565776e-05,
+      "loss": 0.2822,
+      "step": 3151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712387230342164e-05,
+      "loss": 0.308,
+      "step": 3152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121923701204623e-05,
+      "loss": 0.3065,
+      "step": 3153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711997455006492e-05,
+      "loss": 0.3438,
+      "step": 3154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118024850152763e-05,
+      "loss": 0.3024,
+      "step": 3155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116074601618418e-05,
+      "loss": 0.2725,
+      "step": 3156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7114123804612205e-05,
+      "loss": 0.3035,
+      "step": 3157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112172459284478e-05,
+      "loss": 0.2926,
+      "step": 3158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110220565785644e-05,
+      "loss": 0.2766,
+      "step": 3159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710826812426614e-05,
+      "loss": 0.2586,
+      "step": 3160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710631513487645e-05,
+      "loss": 0.2908,
+      "step": 3161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7104361597767107e-05,
+      "loss": 0.3331,
+      "step": 3162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102407513088676e-05,
+      "loss": 0.2949,
+      "step": 3163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100452880991764e-05,
+      "loss": 0.3762,
+      "step": 3164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7098497701627027e-05,
+      "loss": 0.3291,
+      "step": 3165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709654197514517e-05,
+      "loss": 0.288,
+      "step": 3166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094585701696916e-05,
+      "loss": 0.2801,
+      "step": 3167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7092628881433052e-05,
+      "loss": 0.2711,
+      "step": 3168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70906715145044e-05,
+      "loss": 0.2989,
+      "step": 3169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088713601061823e-05,
+      "loss": 0.264,
+      "step": 3170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708675514125623e-05,
+      "loss": 0.2844,
+      "step": 3171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084796135238566e-05,
+      "loss": 0.3286,
+      "step": 3172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082836583159826e-05,
+      "loss": 0.304,
+      "step": 3173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7080876485171035e-05,
+      "loss": 0.285,
+      "step": 3174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7078915841423273e-05,
+      "loss": 0.2709,
+      "step": 3175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7076954652067657e-05,
+      "loss": 0.2842,
+      "step": 3176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074992917255343e-05,
+      "loss": 0.3191,
+      "step": 3177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073030637137535e-05,
+      "loss": 0.3039,
+      "step": 3178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.2858,
+      "step": 3179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706910444159045e-05,
+      "loss": 0.289,
+      "step": 3180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7067140526463778e-05,
+      "loss": 0.3314,
+      "step": 3181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065176066636836e-05,
+      "loss": 0.2966,
+      "step": 3182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063211062261034e-05,
+      "loss": 0.3057,
+      "step": 3183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061245513487824e-05,
+      "loss": 0.28,
+      "step": 3184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70592794204687e-05,
+      "loss": 0.3273,
+      "step": 3185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70573127833552e-05,
+      "loss": 0.3187,
+      "step": 3186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70553456022989e-05,
+      "loss": 0.3081,
+      "step": 3187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053377877451424e-05,
+      "loss": 0.365,
+      "step": 3188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051409608964433e-05,
+      "loss": 0.2942,
+      "step": 3189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704944079698963e-05,
+      "loss": 0.299,
+      "step": 3190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7047471441678764e-05,
+      "loss": 0.2977,
+      "step": 3191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704550154318362e-05,
+      "loss": 0.2982,
+      "step": 3192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704353110165603e-05,
+      "loss": 0.3495,
+      "step": 3193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704156011724787e-05,
+      "loss": 0.2707,
+      "step": 3194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039588590111045e-05,
+      "loss": 0.2767,
+      "step": 3195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037616520397515e-05,
+      "loss": 0.3471,
+      "step": 3196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7035643908259278e-05,
+      "loss": 0.2956,
+      "step": 3197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033670753848373e-05,
+      "loss": 0.3131,
+      "step": 3198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031697057316883e-05,
+      "loss": 0.3559,
+      "step": 3199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702972281881693e-05,
+      "loss": 0.2711,
+      "step": 3200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702774803850067e-05,
+      "loss": 0.2862,
+      "step": 3201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025772716520324e-05,
+      "loss": 0.2958,
+      "step": 3202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7023796853028125e-05,
+      "loss": 0.3347,
+      "step": 3203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7021820448176372e-05,
+      "loss": 0.2975,
+      "step": 3204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019843502117398e-05,
+      "loss": 0.3403,
+      "step": 3205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701786601500357e-05,
+      "loss": 0.2828,
+      "step": 3206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701588798698731e-05,
+      "loss": 0.2838,
+      "step": 3207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7013909418221065e-05,
+      "loss": 0.3041,
+      "step": 3208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701193030885734e-05,
+      "loss": 0.2836,
+      "step": 3209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009950659048677e-05,
+      "loss": 0.3325,
+      "step": 3210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007970468947653e-05,
+      "loss": 0.325,
+      "step": 3211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7005989738706892e-05,
+      "loss": 0.3261,
+      "step": 3212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700400846847906e-05,
+      "loss": 0.2954,
+      "step": 3213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002026658416862e-05,
+      "loss": 0.3056,
+      "step": 3214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700004430867305e-05,
+      "loss": 0.3188,
+      "step": 3215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6998061419400408e-05,
+      "loss": 0.2596,
+      "step": 3216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699607799075177e-05,
+      "loss": 0.3025,
+      "step": 3217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699409402288001e-05,
+      "loss": 0.3573,
+      "step": 3218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992109515938042e-05,
+      "loss": 0.2955,
+      "step": 3219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699012447007882e-05,
+      "loss": 0.3124,
+      "step": 3220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698813888545535e-05,
+      "loss": 0.2769,
+      "step": 3221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6986152762220655e-05,
+      "loss": 0.3116,
+      "step": 3222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698416610052783e-05,
+      "loss": 0.3093,
+      "step": 3223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6982178900529988e-05,
+      "loss": 0.2751,
+      "step": 3224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6980191162380298e-05,
+      "loss": 0.3327,
+      "step": 3225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978202886231963e-05,
+      "loss": 0.287,
+      "step": 3226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697621407223823e-05,
+      "loss": 0.3455,
+      "step": 3227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697422472055239e-05,
+      "loss": 0.2659,
+      "step": 3228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6972234831327767e-05,
+      "loss": 0.2847,
+      "step": 3229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6970244404717732e-05,
+      "loss": 0.3066,
+      "step": 3230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6968253440875702e-05,
+      "loss": 0.3514,
+      "step": 3231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6966261939955125e-05,
+      "loss": 0.2875,
+      "step": 3232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69642699021095e-05,
+      "loss": 0.2546,
+      "step": 3233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962277327492366e-05,
+      "loss": 0.3536,
+      "step": 3234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960284216257293e-05,
+      "loss": 0.3553,
+      "step": 3235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958290568557905e-05,
+      "loss": 0.2839,
+      "step": 3236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695629638454786e-05,
+      "loss": 0.3006,
+      "step": 3237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954301664380867e-05,
+      "loss": 0.2579,
+      "step": 3238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952306408210663e-05,
+      "loss": 0.284,
+      "step": 3239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695031061619103e-05,
+      "loss": 0.2819,
+      "step": 3240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948314288475796e-05,
+      "loss": 0.3775,
+      "step": 3241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946317425218834e-05,
+      "loss": 0.29,
+      "step": 3242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944320026574047e-05,
+      "loss": 0.3154,
+      "step": 3243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694232209269538e-05,
+      "loss": 0.3047,
+      "step": 3244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940323623736835e-05,
+      "loss": 0.2916,
+      "step": 3245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6938324619852435e-05,
+      "loss": 0.3168,
+      "step": 3246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693632508119626e-05,
+      "loss": 0.2517,
+      "step": 3247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934325007922418e-05,
+      "loss": 0.2792,
+      "step": 3248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932324400185073e-05,
+      "loss": 0.3254,
+      "step": 3249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693032325813841e-05,
+      "loss": 0.3464,
+      "step": 3250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928321581936676e-05,
+      "loss": 0.2496,
+      "step": 3251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692631937173415e-05,
+      "loss": 0.2857,
+      "step": 3252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692431662768515e-05,
+      "loss": 0.3168,
+      "step": 3253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922313349944037e-05,
+      "loss": 0.2949,
+      "step": 3254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920309538665215e-05,
+      "loss": 0.3461,
+      "step": 3255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691830519400313e-05,
+      "loss": 0.2751,
+      "step": 3256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916300316112265e-05,
+      "loss": 0.3233,
+      "step": 3257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6914294905147144e-05,
+      "loss": 0.2827,
+      "step": 3258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691228896126234e-05,
+      "loss": 0.2757,
+      "step": 3259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6910282484612452e-05,
+      "loss": 0.266,
+      "step": 3260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690827547535214e-05,
+      "loss": 0.2808,
+      "step": 3261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6906267933636087e-05,
+      "loss": 0.3284,
+      "step": 3262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6904259859619028e-05,
+      "loss": 0.2706,
+      "step": 3263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690225125345573e-05,
+      "loss": 0.2667,
+      "step": 3264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900242115301014e-05,
+      "loss": 0.2571,
+      "step": 3265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689823244530973e-05,
+      "loss": 0.2697,
+      "step": 3266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6896222243636775e-05,
+      "loss": 0.3256,
+      "step": 3267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894211510437086e-05,
+      "loss": 0.3239,
+      "step": 3268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892200245865635e-05,
+      "loss": 0.2856,
+      "step": 3269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890188450077445e-05,
+      "loss": 0.2835,
+      "step": 3270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888176123227576e-05,
+      "loss": 0.3461,
+      "step": 3271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886163265471127e-05,
+      "loss": 0.3284,
+      "step": 3272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688414987696324e-05,
+      "loss": 0.2874,
+      "step": 3273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882135957859095e-05,
+      "loss": 0.2784,
+      "step": 3274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880121508313916e-05,
+      "loss": 0.3162,
+      "step": 3275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6878106528482968e-05,
+      "loss": 0.2836,
+      "step": 3276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687609101852155e-05,
+      "loss": 0.2795,
+      "step": 3277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874074978585018e-05,
+      "loss": 0.3029,
+      "step": 3278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687205840882875e-05,
+      "loss": 0.265,
+      "step": 3279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870041309408174e-05,
+      "loss": 0.3556,
+      "step": 3280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868023680478763e-05,
+      "loss": 0.2484,
+      "step": 3281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686600552219602e-05,
+      "loss": 0.3536,
+      "step": 3282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6863986834715497e-05,
+      "loss": 0.2745,
+      "step": 3283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686196761819279e-05,
+      "loss": 0.2825,
+      "step": 3284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859947872783517e-05,
+      "loss": 0.2963,
+      "step": 3285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6857927598643362e-05,
+      "loss": 0.2841,
+      "step": 3286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6855906795928033e-05,
+      "loss": 0.3093,
+      "step": 3287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853885464793287e-05,
+      "loss": 0.2869,
+      "step": 3288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685186360539491e-05,
+      "loss": 0.2253,
+      "step": 3289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849841217888748e-05,
+      "loss": 0.2869,
+      "step": 3290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684781830243067e-05,
+      "loss": 0.6086,
+      "step": 3291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684579485917659e-05,
+      "loss": 0.2919,
+      "step": 3292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684377088828247e-05,
+      "loss": 0.3093,
+      "step": 3293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6841746389904306e-05,
+      "loss": 0.2957,
+      "step": 3294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6839721364198134e-05,
+      "loss": 0.3026,
+      "step": 3295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683769581132004e-05,
+      "loss": 0.2643,
+      "step": 3296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835669731426137e-05,
+      "loss": 0.2872,
+      "step": 3297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833643124672586e-05,
+      "loss": 0.3674,
+      "step": 3298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683161599121559e-05,
+      "loss": 0.3269,
+      "step": 3299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682958833121139e-05,
+      "loss": 0.2796,
+      "step": 3300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827560144816266e-05,
+      "loss": 0.285,
+      "step": 3301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6825531432186545e-05,
+      "loss": 0.2755,
+      "step": 3302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823502193478583e-05,
+      "loss": 0.3187,
+      "step": 3303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6821472428848788e-05,
+      "loss": 0.283,
+      "step": 3304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6819442138453605e-05,
+      "loss": 0.3214,
+      "step": 3305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681741132244952e-05,
+      "loss": 0.3398,
+      "step": 3306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6815379980993055e-05,
+      "loss": 0.3513,
+      "step": 3307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813348114240775e-05,
+      "loss": 0.2922,
+      "step": 3308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681131572234929e-05,
+      "loss": 0.379,
+      "step": 3309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809282805475243e-05,
+      "loss": 0.2955,
+      "step": 3310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807249363775326e-05,
+      "loss": 0.2535,
+      "step": 3311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805215397406264e-05,
+      "loss": 0.3207,
+      "step": 3312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803180906524822e-05,
+      "loss": 0.34,
+      "step": 3313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680114589128781e-05,
+      "loss": 0.3007,
+      "step": 3314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679911035185208e-05,
+      "loss": 0.3101,
+      "step": 3315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679707428837452e-05,
+      "loss": 0.5824,
+      "step": 3316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679503770101206e-05,
+      "loss": 0.3018,
+      "step": 3317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793000589921666e-05,
+      "loss": 0.2902,
+      "step": 3318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679096295526035e-05,
+      "loss": 0.2965,
+      "step": 3319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6788924797185174e-05,
+      "loss": 0.2875,
+      "step": 3320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786886115853214e-05,
+      "loss": 0.3323,
+      "step": 3321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784846911421605e-05,
+      "loss": 0.2855,
+      "step": 3322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782807184047524e-05,
+      "loss": 0.3079,
+      "step": 3323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678076693388818e-05,
+      "loss": 0.5862,
+      "step": 3324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6778726161100825e-05,
+      "loss": 0.2911,
+      "step": 3325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776684865842748e-05,
+      "loss": 0.2958,
+      "step": 3326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677464304827129e-05,
+      "loss": 0.2935,
+      "step": 3327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6772600708543822e-05,
+      "loss": 0.3343,
+      "step": 3328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770557846817754e-05,
+      "loss": 0.2747,
+      "step": 3329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6768514463250544e-05,
+      "loss": 0.3183,
+      "step": 3330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676647055799968e-05,
+      "loss": 0.2773,
+      "step": 3331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67644261312227e-05,
+      "loss": 0.3177,
+      "step": 3332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6762381183077178e-05,
+      "loss": 0.3218,
+      "step": 3333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6760335713720727e-05,
+      "loss": 0.3254,
+      "step": 3334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758289723311007e-05,
+      "loss": 0.278,
+      "step": 3335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756243212005704e-05,
+      "loss": 0.2692,
+      "step": 3336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6754196179962563e-05,
+      "loss": 0.3147,
+      "step": 3337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675214862733935e-05,
+      "loss": 0.324,
+      "step": 3338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750100554293886e-05,
+      "loss": 0.2687,
+      "step": 3339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674805196098402e-05,
+      "loss": 0.2871,
+      "step": 3340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746002847567656e-05,
+      "loss": 0.3069,
+      "step": 3341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674395321420273e-05,
+      "loss": 0.2973,
+      "step": 3342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6741903061047204e-05,
+      "loss": 0.2627,
+      "step": 3343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6739852388259107e-05,
+      "loss": 0.2808,
+      "step": 3344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737801195996492e-05,
+      "loss": 0.3117,
+      "step": 3345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6735749484417452e-05,
+      "loss": 0.2991,
+      "step": 3346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6733697253680124e-05,
+      "loss": 0.3022,
+      "step": 3347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731644503942684e-05,
+      "loss": 0.2876,
+      "step": 3348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6729591235363346e-05,
+      "loss": 0.2958,
+      "step": 3349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672753744810037e-05,
+      "loss": 0.2692,
+      "step": 3350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725483142312046e-05,
+      "loss": 0.2693,
+      "step": 3351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723428318156715e-05,
+      "loss": 0.3182,
+      "step": 3352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6721372975792752e-05,
+      "loss": 0.3208,
+      "step": 3353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671931711537857e-05,
+      "loss": 0.2614,
+      "step": 3354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6717260737072628e-05,
+      "loss": 0.3231,
+      "step": 3355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715203841033417e-05,
+      "loss": 0.3244,
+      "step": 3356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713146427419473e-05,
+      "loss": 0.2794,
+      "step": 3357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711088496389375e-05,
+      "loss": 0.3044,
+      "step": 3358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6709030048101738e-05,
+      "loss": 0.2933,
+      "step": 3359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706971082715212e-05,
+      "loss": 0.2614,
+      "step": 3360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704911600388496e-05,
+      "loss": 0.3378,
+      "step": 3361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702851601280322e-05,
+      "loss": 0.3207,
+      "step": 3362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6700791085549474e-05,
+      "loss": 0.3201,
+      "step": 3363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669873005335475e-05,
+      "loss": 0.2818,
+      "step": 3364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6696668504855016e-05,
+      "loss": 0.3013,
+      "step": 3365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6694606440209163e-05,
+      "loss": 0.2685,
+      "step": 3366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6692543859576124e-05,
+      "loss": 0.292,
+      "step": 3367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6690480763114877e-05,
+      "loss": 0.2921,
+      "step": 3368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688417150984423e-05,
+      "loss": 0.2457,
+      "step": 3369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668635302334383e-05,
+      "loss": 0.326,
+      "step": 3370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6684288380352182e-05,
+      "loss": 0.299,
+      "step": 3371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6682223222168614e-05,
+      "loss": 0.3047,
+      "step": 3372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66801575489523e-05,
+      "loss": 0.3005,
+      "step": 3373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6678091360862447e-05,
+      "loss": 0.278,
+      "step": 3374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667602465805831e-05,
+      "loss": 0.3246,
+      "step": 3375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6673957440699183e-05,
+      "loss": 0.2977,
+      "step": 3376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667188970894439e-05,
+      "loss": 0.3099,
+      "step": 3377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669821462953303e-05,
+      "loss": 0.238,
+      "step": 3378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666775270288534e-05,
+      "loss": 0.2957,
+      "step": 3379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666568342889994e-05,
+      "loss": 0.2776,
+      "step": 3380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66636136411566e-05,
+      "loss": 0.3134,
+      "step": 3381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661543339814847e-05,
+      "loss": 0.3475,
+      "step": 3382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659472525034245e-05,
+      "loss": 0.2585,
+      "step": 3383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657401196974405e-05,
+      "loss": 0.3091,
+      "step": 3384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6655329355794982e-05,
+      "loss": 0.2728,
+      "step": 3385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653257001655652e-05,
+      "loss": 0.295,
+      "step": 3386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651184134716156e-05,
+      "loss": 0.2575,
+      "step": 3387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649110755136242e-05,
+      "loss": 0.2902,
+      "step": 3388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664703686307573e-05,
+      "loss": 0.2437,
+      "step": 3389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644962458694457e-05,
+      "loss": 0.3054,
+      "step": 3390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6642887542152312e-05,
+      "loss": 0.3232,
+      "step": 3391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664081211360922e-05,
+      "loss": 0.3205,
+      "step": 3392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663873617322514e-05,
+      "loss": 0.2703,
+      "step": 3393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6636659721160088e-05,
+      "loss": 0.3407,
+      "step": 3394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663458275757409e-05,
+      "loss": 0.3225,
+      "step": 3395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6632505282627238e-05,
+      "loss": 0.3024,
+      "step": 3396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663042729647965e-05,
+      "loss": 0.2923,
+      "step": 3397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662834879929149e-05,
+      "loss": 0.2529,
+      "step": 3398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662626979122295e-05,
+      "loss": 0.3021,
+      "step": 3399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6624190272434282e-05,
+      "loss": 0.321,
+      "step": 3400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662211024308576e-05,
+      "loss": 0.3032,
+      "step": 3401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620029703337697e-05,
+      "loss": 0.3085,
+      "step": 3402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6617948653350455e-05,
+      "loss": 0.3025,
+      "step": 3403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6615867093284434e-05,
+      "loss": 0.308,
+      "step": 3404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6613785023300063e-05,
+      "loss": 0.2805,
+      "step": 3405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6611702443557826e-05,
+      "loss": 0.259,
+      "step": 3406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660961935421823e-05,
+      "loss": 0.3278,
+      "step": 3407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607535755441837e-05,
+      "loss": 0.3074,
+      "step": 3408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6605451647389238e-05,
+      "loss": 0.28,
+      "step": 3409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660336703022106e-05,
+      "loss": 0.2719,
+      "step": 3410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6601281904097985e-05,
+      "loss": 0.6416,
+      "step": 3411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6599196269180712e-05,
+      "loss": 0.3318,
+      "step": 3412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597110125630004e-05,
+      "loss": 0.3199,
+      "step": 3413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659502347360664e-05,
+      "loss": 0.5948,
+      "step": 3414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6592936313271456e-05,
+      "loss": 0.3248,
+      "step": 3415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659084864478532e-05,
+      "loss": 0.2456,
+      "step": 3416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588760468309134e-05,
+      "loss": 0.2679,
+      "step": 3417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6586671784003846e-05,
+      "loss": 0.2751,
+      "step": 3418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6584582592030444e-05,
+      "loss": 0.3278,
+      "step": 3419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582492892549954e-05,
+      "loss": 0.3103,
+      "step": 3420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580402685723434e-05,
+      "loss": 0.3233,
+      "step": 3421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657831197171199e-05,
+      "loss": 0.329,
+      "step": 3422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576220750676763e-05,
+      "loss": 0.2979,
+      "step": 3423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574129022778936e-05,
+      "loss": 0.3029,
+      "step": 3424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6572036788179728e-05,
+      "loss": 0.259,
+      "step": 3425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569944047040394e-05,
+      "loss": 0.2759,
+      "step": 3426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567850799522237e-05,
+      "loss": 0.3139,
+      "step": 3427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565757045786595e-05,
+      "loss": 0.2843,
+      "step": 3428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6563662785994843e-05,
+      "loss": 0.319,
+      "step": 3429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561568020308397e-05,
+      "loss": 0.2744,
+      "step": 3430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559472748888703e-05,
+      "loss": 0.2747,
+      "step": 3431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655737697189727e-05,
+      "loss": 0.6115,
+      "step": 3432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555280689495613e-05,
+      "loss": 0.3093,
+      "step": 3433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6553183901845313e-05,
+      "loss": 0.3391,
+      "step": 3434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655108660910798e-05,
+      "loss": 0.2307,
+      "step": 3435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654898881144526e-05,
+      "loss": 0.2841,
+      "step": 3436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654689050901884e-05,
+      "loss": 0.292,
+      "step": 3437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654479170199045e-05,
+      "loss": 0.3267,
+      "step": 3438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654269239052186e-05,
+      "loss": 0.2788,
+      "step": 3439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654059257477486e-05,
+      "loss": 0.3215,
+      "step": 3440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6538492254911305e-05,
+      "loss": 0.2996,
+      "step": 3441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6536391431093077e-05,
+      "loss": 0.3179,
+      "step": 3442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534290103482094e-05,
+      "loss": 0.2978,
+      "step": 3443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532188272240314e-05,
+      "loss": 0.2497,
+      "step": 3444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6530085937529743e-05,
+      "loss": 0.3055,
+      "step": 3445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527983099512414e-05,
+      "loss": 0.2488,
+      "step": 3446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65258797583504e-05,
+      "loss": 0.2935,
+      "step": 3447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652377591420582e-05,
+      "loss": 0.3154,
+      "step": 3448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6521671567240826e-05,
+      "loss": 0.2964,
+      "step": 3449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6519566717617616e-05,
+      "loss": 0.2855,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517461365498416e-05,
+      "loss": 0.2857,
+      "step": 3451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515355511045495e-05,
+      "loss": 0.3333,
+      "step": 3452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651324915442117e-05,
+      "loss": 0.3373,
+      "step": 3453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651114229578778e-05,
+      "loss": 0.3002,
+      "step": 3454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509034935307716e-05,
+      "loss": 0.2891,
+      "step": 3455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506927073143396e-05,
+      "loss": 0.2403,
+      "step": 3456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6504818709457294e-05,
+      "loss": 0.3245,
+      "step": 3457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502709844411907e-05,
+      "loss": 0.2944,
+      "step": 3458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500600478169776e-05,
+      "loss": 0.2986,
+      "step": 3459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498490610893478e-05,
+      "loss": 0.2884,
+      "step": 3460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6496380242745635e-05,
+      "loss": 0.3492,
+      "step": 3461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6494269373888902e-05,
+      "loss": 0.3046,
+      "step": 3462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6492158004485977e-05,
+      "loss": 0.269,
+      "step": 3463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6490046134699586e-05,
+      "loss": 0.2934,
+      "step": 3464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487933764692514e-05,
+      "loss": 0.27,
+      "step": 3465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648582089462756e-05,
+      "loss": 0.2817,
+      "step": 3466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648370752466758e-05,
+      "loss": 0.2922,
+      "step": 3467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481593654975466e-05,
+      "loss": 0.3057,
+      "step": 3468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6479479285714133e-05,
+      "loss": 0.2835,
+      "step": 3469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647736441704656e-05,
+      "loss": 0.3217,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475249049135736e-05,
+      "loss": 0.3239,
+      "step": 3471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6473133182144715e-05,
+      "loss": 0.2886,
+      "step": 3472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647101681623657e-05,
+      "loss": 0.312,
+      "step": 3473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6468899951574423e-05,
+      "loss": 0.2891,
+      "step": 3474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646678258832143e-05,
+      "loss": 0.2823,
+      "step": 3475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6464664726640793e-05,
+      "loss": 0.3081,
+      "step": 3476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646254636669574e-05,
+      "loss": 0.2733,
+      "step": 3477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460427508649546e-05,
+      "loss": 0.3147,
+      "step": 3478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6458308152665522e-05,
+      "loss": 0.3237,
+      "step": 3479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456188298907015e-05,
+      "loss": 0.3083,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6454067947537417e-05,
+      "loss": 0.2651,
+      "step": 3481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6451947098720148e-05,
+      "loss": 0.3128,
+      "step": 3482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644982575261868e-05,
+      "loss": 0.3183,
+      "step": 3483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447703909396514e-05,
+      "loss": 0.2783,
+      "step": 3484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6445581569217184e-05,
+      "loss": 0.3288,
+      "step": 3485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644345873224428e-05,
+      "loss": 0.2936,
+      "step": 3486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6441335398641417e-05,
+      "loss": 0.3128,
+      "step": 3487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643921156857224e-05,
+      "loss": 0.2819,
+      "step": 3488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6437087242200462e-05,
+      "loss": 0.3202,
+      "step": 3489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6434962419689803e-05,
+      "loss": 0.2733,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432837101204036e-05,
+      "loss": 0.2684,
+      "step": 3491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6430711286906968e-05,
+      "loss": 0.2916,
+      "step": 3492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642858497696245e-05,
+      "loss": 0.6289,
+      "step": 3493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642645817153437e-05,
+      "loss": 0.3239,
+      "step": 3494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6424330870786648e-05,
+      "loss": 0.2929,
+      "step": 3495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422203074883247e-05,
+      "loss": 0.2653,
+      "step": 3496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6420074783988164e-05,
+      "loss": 0.2754,
+      "step": 3497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6417945998265436e-05,
+      "loss": 0.3062,
+      "step": 3498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415816717879146e-05,
+      "loss": 0.3181,
+      "step": 3499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6413686942993405e-05,
+      "loss": 0.2699,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411556673772365e-05,
+      "loss": 0.3083,
+      "step": 3501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409425910380215e-05,
+      "loss": 0.2824,
+      "step": 3502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407294652981187e-05,
+      "loss": 0.3623,
+      "step": 3503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6405162901739545e-05,
+      "loss": 0.6411,
+      "step": 3504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6403030656819597e-05,
+      "loss": 0.3047,
+      "step": 3505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6400897918385687e-05,
+      "loss": 0.337,
+      "step": 3506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398764686602188e-05,
+      "loss": 0.5853,
+      "step": 3507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639663096163353e-05,
+      "loss": 0.2921,
+      "step": 3508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639449674364416e-05,
+      "loss": 0.2695,
+      "step": 3509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392362032798578e-05,
+      "loss": 0.2542,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390226829261317e-05,
+      "loss": 0.2558,
+      "step": 3511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388091133196946e-05,
+      "loss": 0.2843,
+      "step": 3512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385954944770076e-05,
+      "loss": 0.2916,
+      "step": 3513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638381826414535e-05,
+      "loss": 0.2812,
+      "step": 3514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638168109148746e-05,
+      "loss": 0.2671,
+      "step": 3515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6379543426961122e-05,
+      "loss": 0.338,
+      "step": 3516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6377405270731097e-05,
+      "loss": 0.2952,
+      "step": 3517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375266622962188e-05,
+      "loss": 0.3425,
+      "step": 3518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6373127483819227e-05,
+      "loss": 0.3367,
+      "step": 3519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637098785346709e-05,
+      "loss": 0.2607,
+      "step": 3520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636884773207069e-05,
+      "loss": 0.308,
+      "step": 3521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366707119794978e-05,
+      "loss": 0.2767,
+      "step": 3522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636456601680494e-05,
+      "loss": 0.2756,
+      "step": 3523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63624244232656e-05,
+      "loss": 0.2422,
+      "step": 3524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360282339342023e-05,
+      "loss": 0.3226,
+      "step": 3525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635813976519931e-05,
+      "loss": 0.2922,
+      "step": 3526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355996701002603e-05,
+      "loss": 0.3536,
+      "step": 3527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353853146917073e-05,
+      "loss": 0.2979,
+      "step": 3528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635170910310794e-05,
+      "loss": 0.2936,
+      "step": 3529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349564569740454e-05,
+      "loss": 0.2738,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6347419546979902e-05,
+      "loss": 0.3466,
+      "step": 3531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345274034991615e-05,
+      "loss": 0.2838,
+      "step": 3532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343128033940962e-05,
+      "loss": 0.277,
+      "step": 3533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634098154399334e-05,
+      "loss": 0.2803,
+      "step": 3534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6338834565314193e-05,
+      "loss": 0.2666,
+      "step": 3535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6336687098069e-05,
+      "loss": 0.3102,
+      "step": 3536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6334539142423275e-05,
+      "loss": 0.3169,
+      "step": 3537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633239069854257e-05,
+      "loss": 0.2339,
+      "step": 3538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6330241766592485e-05,
+      "loss": 0.3302,
+      "step": 3539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632809234673864e-05,
+      "loss": 0.3147,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6325942439146705e-05,
+      "loss": 0.2712,
+      "step": 3541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632379204398238e-05,
+      "loss": 0.2921,
+      "step": 3542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632164116141142e-05,
+      "loss": 0.2944,
+      "step": 3543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631948979159959e-05,
+      "loss": 0.2791,
+      "step": 3544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631733793471271e-05,
+      "loss": 0.2876,
+      "step": 3545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6315185590916644e-05,
+      "loss": 0.2985,
+      "step": 3546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631303276037727e-05,
+      "loss": 0.2849,
+      "step": 3547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631087944326053e-05,
+      "loss": 0.2609,
+      "step": 3548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630872563973238e-05,
+      "loss": 0.269,
+      "step": 3549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6306571349958833e-05,
+      "loss": 0.3069,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630441657410593e-05,
+      "loss": 0.3502,
+      "step": 3551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6302261312339745e-05,
+      "loss": 0.2947,
+      "step": 3552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63001055648264e-05,
+      "loss": 0.2584,
+      "step": 3553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297949331732047e-05,
+      "loss": 0.2751,
+      "step": 3554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629579261322288e-05,
+      "loss": 0.278,
+      "step": 3555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6293635409465127e-05,
+      "loss": 0.2886,
+      "step": 3556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6291477720625052e-05,
+      "loss": 0.2904,
+      "step": 3557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6289319546868966e-05,
+      "loss": 0.2527,
+      "step": 3558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62871608883632e-05,
+      "loss": 0.3637,
+      "step": 3559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285001745274143e-05,
+      "loss": 0.2681,
+      "step": 3560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62828421177682e-05,
+      "loss": 0.3171,
+      "step": 3561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628068200601184e-05,
+      "loss": 0.256,
+      "step": 3562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6278521410171538e-05,
+      "loss": 0.2552,
+      "step": 3563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627636033041383e-05,
+      "loss": 0.3043,
+      "step": 3564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274198766905286e-05,
+      "loss": 0.267,
+      "step": 3565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6272036719812496e-05,
+      "loss": 0.2993,
+      "step": 3566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626987418930211e-05,
+      "loss": 0.3393,
+      "step": 3567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6267711175540795e-05,
+      "loss": 0.3575,
+      "step": 3568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626554767869528e-05,
+      "loss": 0.3052,
+      "step": 3569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6263383698932307e-05,
+      "loss": 0.2604,
+      "step": 3570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261219236418667e-05,
+      "loss": 0.25,
+      "step": 3571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259054291321186e-05,
+      "loss": 0.2901,
+      "step": 3572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256888863806724e-05,
+      "loss": 0.3086,
+      "step": 3573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625472295404219e-05,
+      "loss": 0.3322,
+      "step": 3574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6252556562194514e-05,
+      "loss": 0.2962,
+      "step": 3575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625038968843067e-05,
+      "loss": 0.2575,
+      "step": 3576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6248222332917672e-05,
+      "loss": 0.2944,
+      "step": 3577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6246054495822575e-05,
+      "loss": 0.2662,
+      "step": 3578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624388617731246e-05,
+      "loss": 0.3342,
+      "step": 3579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624171737755445e-05,
+      "loss": 0.3099,
+      "step": 3580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6239548096715703e-05,
+      "loss": 0.3141,
+      "step": 3581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237378334963422e-05,
+      "loss": 0.2843,
+      "step": 3582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6235208092464832e-05,
+      "loss": 0.2454,
+      "step": 3583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6233037369387223e-05,
+      "loss": 0.3518,
+      "step": 3584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6230866165897882e-05,
+      "loss": 0.2885,
+      "step": 3585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228694482164167e-05,
+      "loss": 0.3163,
+      "step": 3586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6226522318353462e-05,
+      "loss": 0.3243,
+      "step": 3587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224349674633178e-05,
+      "loss": 0.2996,
+      "step": 3588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622217655117078e-05,
+      "loss": 0.6091,
+      "step": 3589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6220002948133756e-05,
+      "loss": 0.3337,
+      "step": 3590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621782886568964e-05,
+      "loss": 0.3636,
+      "step": 3591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215654304005995e-05,
+      "loss": 0.2984,
+      "step": 3592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213479263250433e-05,
+      "loss": 0.2817,
+      "step": 3593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621130374359059e-05,
+      "loss": 0.2816,
+      "step": 3594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6209127745194143e-05,
+      "loss": 0.2599,
+      "step": 3595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620695126822881e-05,
+      "loss": 0.3114,
+      "step": 3596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6204774312862346e-05,
+      "loss": 0.2878,
+      "step": 3597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202596879262536e-05,
+      "loss": 0.2787,
+      "step": 3598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200418967597205e-05,
+      "loss": 0.3201,
+      "step": 3599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6198240578034216e-05,
+      "loss": 0.2548,
+      "step": 3600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6196061710741472e-05,
+      "loss": 0.2433,
+      "step": 3601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193882365886905e-05,
+      "loss": 0.3009,
+      "step": 3602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6191702543638493e-05,
+      "loss": 0.3177,
+      "step": 3603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618952224416424e-05,
+      "loss": 0.2981,
+      "step": 3604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6187341467632198e-05,
+      "loss": 0.621,
+      "step": 3605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185160214210447e-05,
+      "loss": 0.337,
+      "step": 3606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182978484067106e-05,
+      "loss": 0.2773,
+      "step": 3607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618079627737034e-05,
+      "loss": 0.649,
+      "step": 3608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617861359428833e-05,
+      "loss": 0.282,
+      "step": 3609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617643043498932e-05,
+      "loss": 0.3079,
+      "step": 3610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617424679964157e-05,
+      "loss": 0.2839,
+      "step": 3611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617206268841338e-05,
+      "loss": 0.3082,
+      "step": 3612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61698781014731e-05,
+      "loss": 0.3101,
+      "step": 3613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167693038989098e-05,
+      "loss": 0.2913,
+      "step": 3614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6165507501129796e-05,
+      "loss": 0.2648,
+      "step": 3615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163321488063636e-05,
+      "loss": 0.2505,
+      "step": 3616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161134999959115e-05,
+      "loss": 0.3237,
+      "step": 3617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615894803698475e-05,
+      "loss": 0.2932,
+      "step": 3618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61567605993091e-05,
+      "loss": 0.2676,
+      "step": 3619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154572687100766e-05,
+      "loss": 0.3368,
+      "step": 3620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152384300528375e-05,
+      "loss": 0.29,
+      "step": 3621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615019543976061e-05,
+      "loss": 0.2947,
+      "step": 3622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148006104966164e-05,
+      "loss": 0.3134,
+      "step": 3623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614581629631379e-05,
+      "loss": 0.259,
+      "step": 3624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614362601397226e-05,
+      "loss": 0.3729,
+      "step": 3625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141435258110397e-05,
+      "loss": 0.2959,
+      "step": 3626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139244028897044e-05,
+      "loss": 0.2861,
+      "step": 3627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137052326501098e-05,
+      "loss": 0.2738,
+      "step": 3628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613486015109149e-05,
+      "loss": 0.246,
+      "step": 3629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6132667502837164e-05,
+      "loss": 0.3085,
+      "step": 3630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6130474381907135e-05,
+      "loss": 0.3641,
+      "step": 3631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128280788470432e-05,
+      "loss": 0.3244,
+      "step": 3632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126086722696123e-05,
+      "loss": 0.2672,
+      "step": 3633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6123892184753324e-05,
+      "loss": 0.3281,
+      "step": 3634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612169717481117e-05,
+      "loss": 0.3254,
+      "step": 3635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611950169303885e-05,
+      "loss": 0.2674,
+      "step": 3636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6117305739605574e-05,
+      "loss": 0.3234,
+      "step": 3637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115109314680603e-05,
+      "loss": 0.268,
+      "step": 3638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6112912418433218e-05,
+      "loss": 0.2967,
+      "step": 3639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6110715051032748e-05,
+      "loss": 0.2777,
+      "step": 3640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6108517212648556e-05,
+      "loss": 0.2784,
+      "step": 3641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106318903450042e-05,
+      "loss": 0.2413,
+      "step": 3642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610412012360664e-05,
+      "loss": 0.2925,
+      "step": 3643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101920873287815e-05,
+      "loss": 0.2543,
+      "step": 3644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6099721152663084e-05,
+      "loss": 0.2842,
+      "step": 3645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6097520961901983e-05,
+      "loss": 0.2676,
+      "step": 3646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6095320301174097e-05,
+      "loss": 0.2871,
+      "step": 3647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609311917064904e-05,
+      "loss": 0.2686,
+      "step": 3648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090917570496465e-05,
+      "loss": 0.3093,
+      "step": 3649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608871550088606e-05,
+      "loss": 0.3452,
+      "step": 3650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086512961987548e-05,
+      "loss": 0.3013,
+      "step": 3651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608430995397069e-05,
+      "loss": 0.3232,
+      "step": 3652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608210647700529e-05,
+      "loss": 0.2588,
+      "step": 3653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607990253126117e-05,
+      "loss": 0.2703,
+      "step": 3654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077698116908204e-05,
+      "loss": 0.2668,
+      "step": 3655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60754932341163e-05,
+      "loss": 0.32,
+      "step": 3656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60732878830554e-05,
+      "loss": 0.3237,
+      "step": 3657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071082063895476e-05,
+      "loss": 0.3441,
+      "step": 3658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606887577680654e-05,
+      "loss": 0.3376,
+      "step": 3659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6066669021958653e-05,
+      "loss": 0.3409,
+      "step": 3660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064461799521892e-05,
+      "loss": 0.2643,
+      "step": 3661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6062254109666383e-05,
+      "loss": 0.2762,
+      "step": 3662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606004595256228e-05,
+      "loss": 0.2946,
+      "step": 3663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057837328379778e-05,
+      "loss": 0.2802,
+      "step": 3664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6055628237289103e-05,
+      "loss": 0.3182,
+      "step": 3665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053418679460534e-05,
+      "loss": 0.2902,
+      "step": 3666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605120865506436e-05,
+      "loss": 0.2957,
+      "step": 3667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604899816427092e-05,
+      "loss": 0.2753,
+      "step": 3668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6046787207250597e-05,
+      "loss": 0.3199,
+      "step": 3669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604457578417379e-05,
+      "loss": 0.3224,
+      "step": 3670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6042363895210948e-05,
+      "loss": 0.3215,
+      "step": 3671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6040151540532553e-05,
+      "loss": 0.2979,
+      "step": 3672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037938720309122e-05,
+      "loss": 0.3066,
+      "step": 3673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603572543471121e-05,
+      "loss": 0.3035,
+      "step": 3674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6033511683909406e-05,
+      "loss": 0.2681,
+      "step": 3675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031297468074335e-05,
+      "loss": 0.2849,
+      "step": 3676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6029082787376653e-05,
+      "loss": 0.2579,
+      "step": 3677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602686764198706e-05,
+      "loss": 0.582,
+      "step": 3678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024652032076295e-05,
+      "loss": 0.2867,
+      "step": 3679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022435957815116e-05,
+      "loss": 0.283,
+      "step": 3680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602021941937433e-05,
+      "loss": 0.2457,
+      "step": 3681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601800241692478e-05,
+      "loss": 0.3286,
+      "step": 3682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015784950637338e-05,
+      "loss": 0.3573,
+      "step": 3683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013567020682917e-05,
+      "loss": 0.3053,
+      "step": 3684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6011348627232463e-05,
+      "loss": 0.2863,
+      "step": 3685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009129770456962e-05,
+      "loss": 0.3561,
+      "step": 3686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6006910450527428e-05,
+      "loss": 0.2638,
+      "step": 3687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600469066761492e-05,
+      "loss": 0.34,
+      "step": 3688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002470421890522e-05,
+      "loss": 0.2606,
+      "step": 3689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000249713525366e-05,
+      "loss": 0.2829,
+      "step": 3690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599802854269061e-05,
+      "loss": 0.3141,
+      "step": 3691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599580690955745e-05,
+      "loss": 0.6311,
+      "step": 3692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599358481429712e-05,
+      "loss": 0.2985,
+      "step": 3693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599136225708089e-05,
+      "loss": 0.3333,
+      "step": 3694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598913923808006e-05,
+      "loss": 0.2717,
+      "step": 3695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986915757465968e-05,
+      "loss": 0.3118,
+      "step": 3696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598469181540999e-05,
+      "loss": 0.257,
+      "step": 3697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982467412083543e-05,
+      "loss": 0.2535,
+      "step": 3698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5980242547658068e-05,
+      "loss": 0.2959,
+      "step": 3699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978017222305046e-05,
+      "loss": 0.3043,
+      "step": 3700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5975791436195994e-05,
+      "loss": 0.2791,
+      "step": 3701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973565189502463e-05,
+      "loss": 0.3318,
+      "step": 3702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597133848239605e-05,
+      "loss": 0.2953,
+      "step": 3703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969111315048365e-05,
+      "loss": 0.2986,
+      "step": 3704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5966883687631075e-05,
+      "loss": 0.2959,
+      "step": 3705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596465560031588e-05,
+      "loss": 0.2951,
+      "step": 3706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5962427053274495e-05,
+      "loss": 0.2969,
+      "step": 3707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59601980466787e-05,
+      "loss": 0.2968,
+      "step": 3708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595796858070029e-05,
+      "loss": 0.293,
+      "step": 3709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5955738655511094e-05,
+      "loss": 0.3122,
+      "step": 3710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953508271282997e-05,
+      "loss": 0.2566,
+      "step": 3711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59512774281879e-05,
+      "loss": 0.2891,
+      "step": 3712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594904612639774e-05,
+      "loss": 0.2946,
+      "step": 3713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5946814366084505e-05,
+      "loss": 0.317,
+      "step": 3714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59445821474202e-05,
+      "loss": 0.2724,
+      "step": 3715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5942349470576878e-05,
+      "loss": 0.3065,
+      "step": 3716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940116335726615e-05,
+      "loss": 0.2784,
+      "step": 3717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5937882743041543e-05,
+      "loss": 0.3358,
+      "step": 3718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935648692693805e-05,
+      "loss": 0.3549,
+      "step": 3719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5933414184855597e-05,
+      "loss": 0.3517,
+      "step": 3720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5931179219699144e-05,
+      "loss": 0.2758,
+      "step": 3721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928943797396695e-05,
+      "loss": 0.3008,
+      "step": 3722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592670791812056e-05,
+      "loss": 0.3405,
+      "step": 3723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592447158204306e-05,
+      "loss": 0.2762,
+      "step": 3724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5922234789336567e-05,
+      "loss": 0.3245,
+      "step": 3725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591999754017348e-05,
+      "loss": 0.3244,
+      "step": 3726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5917759834726233e-05,
+      "loss": 0.2568,
+      "step": 3727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915521673167296e-05,
+      "loss": 0.3459,
+      "step": 3728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591328305566918e-05,
+      "loss": 0.3571,
+      "step": 3729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5911043982404426e-05,
+      "loss": 0.2926,
+      "step": 3730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5908804453545608e-05,
+      "loss": 0.3193,
+      "step": 3731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590656446926534e-05,
+      "loss": 0.3146,
+      "step": 3732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5904324029736266e-05,
+      "loss": 0.298,
+      "step": 3733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5902083135131067e-05,
+      "loss": 0.2803,
+      "step": 3734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899841785622468e-05,
+      "loss": 0.2667,
+      "step": 3735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5897599981383214e-05,
+      "loss": 0.2776,
+      "step": 3736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895357722586093e-05,
+      "loss": 0.2954,
+      "step": 3737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5893115009403932e-05,
+      "loss": 0.3359,
+      "step": 3738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5890871842009582e-05,
+      "loss": 0.3189,
+      "step": 3739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588862822057594e-05,
+      "loss": 0.615,
+      "step": 3740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588638414527593e-05,
+      "loss": 0.3118,
+      "step": 3741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884139616282517e-05,
+      "loss": 0.3116,
+      "step": 3742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881894633768697e-05,
+      "loss": 0.324,
+      "step": 3743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58796491979075e-05,
+      "loss": 0.3676,
+      "step": 3744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877403308871997e-05,
+      "loss": 0.3316,
+      "step": 3745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5875156966835285e-05,
+      "loss": 0.2591,
+      "step": 3746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5872910171970506e-05,
+      "loss": 0.3391,
+      "step": 3747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587066292445083e-05,
+      "loss": 0.5974,
+      "step": 3748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868415224449463e-05,
+      "loss": 0.3163,
+      "step": 3749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866167072139645e-05,
+      "loss": 0.31,
+      "step": 3750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586391846769466e-05,
+      "loss": 0.3144,
+      "step": 3751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586166941128781e-05,
+      "loss": 0.3107,
+      "step": 3752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585941990309245e-05,
+      "loss": 0.2744,
+      "step": 3753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5857169943281948e-05,
+      "loss": 0.2595,
+      "step": 3754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854919532029734e-05,
+      "loss": 0.2889,
+      "step": 3755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5852668669509252e-05,
+      "loss": 0.3166,
+      "step": 3756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850417355893984e-05,
+      "loss": 0.2836,
+      "step": 3757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848165591357458e-05,
+      "loss": 0.3284,
+      "step": 3758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845913376073222e-05,
+      "loss": 0.2581,
+      "step": 3759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843660710214872e-05,
+      "loss": 0.4052,
+      "step": 3760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5841407593956026e-05,
+      "loss": 0.2761,
+      "step": 3761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839154027470346e-05,
+      "loss": 0.3372,
+      "step": 3762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836900010931527e-05,
+      "loss": 0.3278,
+      "step": 3763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834645544513296e-05,
+      "loss": 0.296,
+      "step": 3764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5832390628389417e-05,
+      "loss": 0.3231,
+      "step": 3765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830135262733684e-05,
+      "loss": 0.3066,
+      "step": 3766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5827879447719932e-05,
+      "loss": 0.2812,
+      "step": 3767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582562318352203e-05,
+      "loss": 0.2911,
+      "step": 3768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582336647031388e-05,
+      "loss": 0.2672,
+      "step": 3769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5821109308269416e-05,
+      "loss": 0.3187,
+      "step": 3770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581885169756261e-05,
+      "loss": 0.2862,
+      "step": 3771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581659363836747e-05,
+      "loss": 0.2764,
+      "step": 3772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814335130858026e-05,
+      "loss": 0.257,
+      "step": 3773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581207617520836e-05,
+      "loss": 0.261,
+      "step": 3774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5809816771592584e-05,
+      "loss": 0.2747,
+      "step": 3775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807556920184837e-05,
+      "loss": 0.3214,
+      "step": 3776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58052966211593e-05,
+      "loss": 0.2962,
+      "step": 3777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5803035874690186e-05,
+      "loss": 0.2794,
+      "step": 3778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800774680951736e-05,
+      "loss": 0.3008,
+      "step": 3779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579851304011824e-05,
+      "loss": 0.2864,
+      "step": 3780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796250952364008e-05,
+      "loss": 0.2719,
+      "step": 3781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579398841786339e-05,
+      "loss": 0.3212,
+      "step": 3782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579172543679078e-05,
+      "loss": 0.2466,
+      "step": 3783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789462009320586e-05,
+      "loss": 0.27,
+      "step": 3784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578719813562727e-05,
+      "loss": 0.2386,
+      "step": 3785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5784933815885315e-05,
+      "loss": 0.2562,
+      "step": 3786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5782669050269243e-05,
+      "loss": 0.3107,
+      "step": 3787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578040383895362e-05,
+      "loss": 0.2988,
+      "step": 3788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5778138182113027e-05,
+      "loss": 0.3156,
+      "step": 3789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5775872079922098e-05,
+      "loss": 0.2906,
+      "step": 3790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5773605532555484e-05,
+      "loss": 0.287,
+      "step": 3791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5771338540187883e-05,
+      "loss": 0.2913,
+      "step": 3792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769071102994024e-05,
+      "loss": 0.5977,
+      "step": 3793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766803221148676e-05,
+      "loss": 0.3246,
+      "step": 3794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764534894826623e-05,
+      "loss": 0.2846,
+      "step": 3795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762266124202708e-05,
+      "loss": 0.28,
+      "step": 3796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759996909451795e-05,
+      "loss": 0.2986,
+      "step": 3797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757727250748773e-05,
+      "loss": 0.302,
+      "step": 3798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575545714826859e-05,
+      "loss": 0.356,
+      "step": 3799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753186602186207e-05,
+      "loss": 0.3202,
+      "step": 3800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575091561267663e-05,
+      "loss": 0.307,
+      "step": 3801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574864417991489e-05,
+      "loss": 0.2645,
+      "step": 3802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746372304076065e-05,
+      "loss": 0.2705,
+      "step": 3803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5744099985335255e-05,
+      "loss": 0.3011,
+      "step": 3804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5741827223867602e-05,
+      "loss": 0.6638,
+      "step": 3805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5739554019848274e-05,
+      "loss": 0.3149,
+      "step": 3806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737280373452487e-05,
+      "loss": 0.2958,
+      "step": 3807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735006284855473e-05,
+      "loss": 0.3245,
+      "step": 3808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5732731754232516e-05,
+      "loss": 0.2841,
+      "step": 3809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573045678175892e-05,
+      "loss": 0.3588,
+      "step": 3810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572818136761003e-05,
+      "loss": 0.2849,
+      "step": 3811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5725905511961226e-05,
+      "loss": 0.3043,
+      "step": 3812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5723629214987915e-05,
+      "loss": 0.3508,
+      "step": 3813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5721352476865546e-05,
+      "loss": 0.2888,
+      "step": 3814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5719075297769596e-05,
+      "loss": 0.2879,
+      "step": 3815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5716797677875586e-05,
+      "loss": 0.3232,
+      "step": 3816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5714519617359054e-05,
+      "loss": 0.2796,
+      "step": 3817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571224111639559e-05,
+      "loss": 0.2831,
+      "step": 3818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5709962175160806e-05,
+      "loss": 0.2746,
+      "step": 3819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707682793830347e-05,
+      "loss": 0.2793,
+      "step": 3820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5705402972579902e-05,
+      "loss": 0.3566,
+      "step": 3821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570312271158519e-05,
+      "loss": 0.2615,
+      "step": 3822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5700842011021954e-05,
+      "loss": 0.2676,
+      "step": 3823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5698560871065986e-05,
+      "loss": 0.3256,
+      "step": 3824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5696279291893107e-05,
+      "loss": 0.2972,
+      "step": 3825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693997273679165e-05,
+      "loss": 0.2914,
+      "step": 3826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691714816600045e-05,
+      "loss": 0.303,
+      "step": 3827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689431920831676e-05,
+      "loss": 0.3151,
+      "step": 3828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5687148586550003e-05,
+      "loss": 0.2988,
+      "step": 3829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568486481393102e-05,
+      "loss": 0.2723,
+      "step": 3830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682580603150742e-05,
+      "loss": 0.2534,
+      "step": 3831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5680295954385235e-05,
+      "loss": 0.2994,
+      "step": 3832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5678010867810583e-05,
+      "loss": 0.3505,
+      "step": 3833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675725343602904e-05,
+      "loss": 0.2736,
+      "step": 3834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5673439381938365e-05,
+      "loss": 0.2769,
+      "step": 3835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567115298299315e-05,
+      "loss": 0.2642,
+      "step": 3836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668866146943484e-05,
+      "loss": 0.2465,
+      "step": 3837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5666578873965627e-05,
+      "loss": 0.2972,
+      "step": 3838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566429116423587e-05,
+      "loss": 0.293,
+      "step": 3839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566200301793054e-05,
+      "loss": 0.2595,
+      "step": 3840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659714435225993e-05,
+      "loss": 0.3223,
+      "step": 3841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5657425416298623e-05,
+      "loss": 0.2939,
+      "step": 3842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5655135961324856e-05,
+      "loss": 0.2733,
+      "step": 3843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565284607048115e-05,
+      "loss": 0.2607,
+      "step": 3844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5650555743944002e-05,
+      "loss": 0.3189,
+      "step": 3845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5648264981889936e-05,
+      "loss": 0.2677,
+      "step": 3846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5645973784495517e-05,
+      "loss": 0.2839,
+      "step": 3847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5643682151937333e-05,
+      "loss": 0.296,
+      "step": 3848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564139008439202e-05,
+      "loss": 0.2631,
+      "step": 3849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5639097582036226e-05,
+      "loss": 0.3555,
+      "step": 3850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563680464504666e-05,
+      "loss": 0.2966,
+      "step": 3851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5634511273600042e-05,
+      "loss": 0.2507,
+      "step": 3852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563221746787314e-05,
+      "loss": 0.2749,
+      "step": 3853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562992322804274e-05,
+      "loss": 0.2935,
+      "step": 3854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627628554285678e-05,
+      "loss": 0.3058,
+      "step": 3855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5625333446778812e-05,
+      "loss": 0.2646,
+      "step": 3856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623037905699043e-05,
+      "loss": 0.2944,
+      "step": 3857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620741931223292e-05,
+      "loss": 0.2922,
+      "step": 3858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5618445523528533e-05,
+      "loss": 0.3265,
+      "step": 3859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561614868279175e-05,
+      "loss": 0.287,
+      "step": 3860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5613851409189974e-05,
+      "loss": 0.2684,
+      "step": 3861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5611553702900275e-05,
+      "loss": 0.301,
+      "step": 3862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560925556409974e-05,
+      "loss": 0.2579,
+      "step": 3863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5606956992965504e-05,
+      "loss": 0.2759,
+      "step": 3864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560465798967473e-05,
+      "loss": 0.2783,
+      "step": 3865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602358554404613e-05,
+      "loss": 0.2875,
+      "step": 3866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600058687332375e-05,
+      "loss": 0.2493,
+      "step": 3867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597758388635288e-05,
+      "loss": 0.2921,
+      "step": 3868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5595457658490643e-05,
+      "loss": 0.2942,
+      "step": 3869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593156497075767e-05,
+      "loss": 0.3015,
+      "step": 3870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559085490456803e-05,
+      "loss": 0.3109,
+      "step": 3871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5588552881144814e-05,
+      "loss": 0.6384,
+      "step": 3872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586250426983566e-05,
+      "loss": 0.333,
+      "step": 3873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558394754226173e-05,
+      "loss": 0.3302,
+      "step": 3874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5581644227156815e-05,
+      "loss": 0.3281,
+      "step": 3875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5579340481846338e-05,
+      "loss": 0.2907,
+      "step": 3876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5577036306507863e-05,
+      "loss": 0.3022,
+      "step": 3877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5574731701318987e-05,
+      "loss": 0.2982,
+      "step": 3878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5572426666457342e-05,
+      "loss": 0.3142,
+      "step": 3879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557012120210058e-05,
+      "loss": 0.3098,
+      "step": 3880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55678153084264e-05,
+      "loss": 0.2721,
+      "step": 3881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5565508985612525e-05,
+      "loss": 0.2855,
+      "step": 3882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556320223383672e-05,
+      "loss": 0.3185,
+      "step": 3883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556089505327677e-05,
+      "loss": 0.265,
+      "step": 3884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555858744411052e-05,
+      "loss": 0.3051,
+      "step": 3885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556279406515802e-05,
+      "loss": 0.2719,
+      "step": 3886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553970940670527e-05,
+      "loss": 0.304,
+      "step": 3887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551662046752612e-05,
+      "loss": 0.2966,
+      "step": 3888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554935272494002e-05,
+      "loss": 0.3229,
+      "step": 3889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554704297541074e-05,
+      "loss": 0.239,
+      "step": 3890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5544732798342798e-05,
+      "loss": 0.2798,
+      "step": 3891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554242219391425e-05,
+      "loss": 0.2979,
+      "step": 3892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554011116230318e-05,
+      "loss": 0.3093,
+      "step": 3893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553779970368772e-05,
+      "loss": 0.332,
+      "step": 3894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5535487818246023e-05,
+      "loss": 0.3015,
+      "step": 3895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553317550615627e-05,
+      "loss": 0.3169,
+      "step": 3896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5530862767596697e-05,
+      "loss": 0.3001,
+      "step": 3897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5528549602745545e-05,
+      "loss": 0.3043,
+      "step": 3898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552623601178111e-05,
+      "loss": 0.3199,
+      "step": 3899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552392199488171e-05,
+      "loss": 0.2902,
+      "step": 3900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5521607552225698e-05,
+      "loss": 0.2822,
+      "step": 3901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5519292683991455e-05,
+      "loss": 0.2686,
+      "step": 3902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5516977390357405e-05,
+      "loss": 0.3297,
+      "step": 3903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514661671502e-05,
+      "loss": 0.2429,
+      "step": 3904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5512345527603718e-05,
+      "loss": 0.3002,
+      "step": 3905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5510028958841085e-05,
+      "loss": 0.2615,
+      "step": 3906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507711965392643e-05,
+      "loss": 0.2908,
+      "step": 3907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5505394547436976e-05,
+      "loss": 0.2653,
+      "step": 3908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.55030767051527e-05,
+      "loss": 0.2932,
+      "step": 3909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5500758438718463e-05,
+      "loss": 0.3299,
+      "step": 3910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549843974831295e-05,
+      "loss": 0.3065,
+      "step": 3911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5496120634114865e-05,
+      "loss": 0.5555,
+      "step": 3912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5493801096302964e-05,
+      "loss": 0.2878,
+      "step": 3913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491481135056012e-05,
+      "loss": 0.2618,
+      "step": 3914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5489160750552833e-05,
+      "loss": 0.3099,
+      "step": 3915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548683994297227e-05,
+      "loss": 0.3718,
+      "step": 3916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5484518712493188e-05,
+      "loss": 0.2573,
+      "step": 3917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548219705929451e-05,
+      "loss": 0.3022,
+      "step": 3918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5479874983555166e-05,
+      "loss": 0.2458,
+      "step": 3919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477552485454136e-05,
+      "loss": 0.2794,
+      "step": 3920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475229565170428e-05,
+      "loss": 0.3146,
+      "step": 3921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472906222883075e-05,
+      "loss": 0.3178,
+      "step": 3922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547058245877116e-05,
+      "loss": 0.2927,
+      "step": 3923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5468258273013773e-05,
+      "loss": 0.3034,
+      "step": 3924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546593366579006e-05,
+      "loss": 0.3246,
+      "step": 3925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546360863727919e-05,
+      "loss": 0.3193,
+      "step": 3926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546128318766036e-05,
+      "loss": 0.2811,
+      "step": 3927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545895731711281e-05,
+      "loss": 0.2975,
+      "step": 3928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54566310258158e-05,
+      "loss": 0.3044,
+      "step": 3929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5454304313948635e-05,
+      "loss": 0.2925,
+      "step": 3930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545197718169064e-05,
+      "loss": 0.2997,
+      "step": 3931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544964962922119e-05,
+      "loss": 0.2923,
+      "step": 3932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5447321656719668e-05,
+      "loss": 0.3475,
+      "step": 3933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544499326436551e-05,
+      "loss": 0.2952,
+      "step": 3934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5442664452338178e-05,
+      "loss": 0.3317,
+      "step": 3935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544033522081716e-05,
+      "loss": 0.2699,
+      "step": 3936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5438005569981986e-05,
+      "loss": 0.2758,
+      "step": 3937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5435675500012212e-05,
+      "loss": 0.2902,
+      "step": 3938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5433345011087427e-05,
+      "loss": 0.3263,
+      "step": 3939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5431014103387263e-05,
+      "loss": 0.2625,
+      "step": 3940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542868277709136e-05,
+      "loss": 0.3054,
+      "step": 3941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426351032379418e-05,
+      "loss": 0.326,
+      "step": 3942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5424018869431144e-05,
+      "loss": 0.6332,
+      "step": 3943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5421686288426303e-05,
+      "loss": 0.3088,
+      "step": 3944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541935328954467e-05,
+      "loss": 0.2856,
+      "step": 3945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541701987296606e-05,
+      "loss": 0.3416,
+      "step": 3946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414686038870327e-05,
+      "loss": 0.2977,
+      "step": 3947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541235178743735e-05,
+      "loss": 0.2991,
+      "step": 3948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541001711884704e-05,
+      "loss": 0.2932,
+      "step": 3949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540768203327934e-05,
+      "loss": 0.2958,
+      "step": 3950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5405346530914233e-05,
+      "loss": 0.3158,
+      "step": 3951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403010611931718e-05,
+      "loss": 0.308,
+      "step": 3952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5400674276511848e-05,
+      "loss": 0.2679,
+      "step": 3953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5398337524834688e-05,
+      "loss": 0.2838,
+      "step": 3954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396000357080345e-05,
+      "loss": 0.2902,
+      "step": 3955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393662773428956e-05,
+      "loss": 0.2709,
+      "step": 3956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391324774060695e-05,
+      "loss": 0.2622,
+      "step": 3957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538898635915576e-05,
+      "loss": 0.3279,
+      "step": 3958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386647528894377e-05,
+      "loss": 0.3098,
+      "step": 3959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5384308283456824e-05,
+      "loss": 0.3135,
+      "step": 3960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538196862302339e-05,
+      "loss": 0.3124,
+      "step": 3961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5379628547774412e-05,
+      "loss": 0.2665,
+      "step": 3962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377288057890246e-05,
+      "loss": 0.2953,
+      "step": 3963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374947153551284e-05,
+      "loss": 0.2875,
+      "step": 3964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372605834937953e-05,
+      "loss": 0.2756,
+      "step": 3965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370264102230716e-05,
+      "loss": 0.3225,
+      "step": 3966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5367921955610055e-05,
+      "loss": 0.2899,
+      "step": 3967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365579395256493e-05,
+      "loss": 0.2505,
+      "step": 3968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5363236421350584e-05,
+      "loss": 0.3038,
+      "step": 3969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536089303407291e-05,
+      "loss": 0.2897,
+      "step": 3970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5358549233604093e-05,
+      "loss": 0.2704,
+      "step": 3971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535620502012478e-05,
+      "loss": 0.2425,
+      "step": 3972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5353860393815642e-05,
+      "loss": 0.3146,
+      "step": 3973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351515354857404e-05,
+      "loss": 0.2863,
+      "step": 3974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349169903430804e-05,
+      "loss": 0.2706,
+      "step": 3975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346824039716622e-05,
+      "loss": 0.2833,
+      "step": 3976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5344477763895663e-05,
+      "loss": 0.3802,
+      "step": 3977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534213107614876e-05,
+      "loss": 0.2939,
+      "step": 3978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5339783976656793e-05,
+      "loss": 0.3157,
+      "step": 3979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533743646560066e-05,
+      "loss": 0.3018,
+      "step": 3980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53350885431613e-05,
+      "loss": 0.2964,
+      "step": 3981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5332740209519674e-05,
+      "loss": 0.2876,
+      "step": 3982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5330391464856784e-05,
+      "loss": 0.3074,
+      "step": 3983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328042309353655e-05,
+      "loss": 0.3092,
+      "step": 3984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532569274319136e-05,
+      "loss": 0.59,
+      "step": 3985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5323342766550978e-05,
+      "loss": 0.2571,
+      "step": 3986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320992379613637e-05,
+      "loss": 0.2595,
+      "step": 3987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5318641582560497e-05,
+      "loss": 0.2527,
+      "step": 3988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5316290375572746e-05,
+      "loss": 0.3005,
+      "step": 3989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313938758831596e-05,
+      "loss": 0.2512,
+      "step": 3990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5311586732518307e-05,
+      "loss": 0.348,
+      "step": 3991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530923429681416e-05,
+      "loss": 0.2892,
+      "step": 3992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5306881451900462e-05,
+      "loss": 0.2681,
+      "step": 3993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304528197958565e-05,
+      "loss": 0.3127,
+      "step": 3994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530217453516985e-05,
+      "loss": 0.3745,
+      "step": 3995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299820463715716e-05,
+      "loss": 0.2645,
+      "step": 3996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5297465983777612e-05,
+      "loss": 0.3236,
+      "step": 3997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295111095536997e-05,
+      "loss": 0.2545,
+      "step": 3998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529275579917539e-05,
+      "loss": 0.3099,
+      "step": 3999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5290400094874316e-05,
+      "loss": 0.3007,
+      "step": 4000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5288043982815345e-05,
+      "loss": 0.2887,
+      "step": 4001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528568746318007e-05,
+      "loss": 0.2738,
+      "step": 4002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5283330536150126e-05,
+      "loss": 0.2722,
+      "step": 4003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528097320190717e-05,
+      "loss": 0.3093,
+      "step": 4004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278615460632892e-05,
+      "loss": 0.3059,
+      "step": 4005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527625731250901e-05,
+      "loss": 0.2981,
+      "step": 4006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273898757717295e-05,
+      "loss": 0.2946,
+      "step": 4007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527153979643952e-05,
+      "loss": 0.2905,
+      "step": 4008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5269180428857506e-05,
+      "loss": 0.2736,
+      "step": 4009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52668206551531e-05,
+      "loss": 0.3032,
+      "step": 4010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526446047550818e-05,
+      "loss": 0.2682,
+      "step": 4011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526209989010466e-05,
+      "loss": 0.2825,
+      "step": 4012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525973889912448e-05,
+      "loss": 0.2828,
+      "step": 4013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257377502749614e-05,
+      "loss": 0.3463,
+      "step": 4014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5255015701162071e-05,
+      "loss": 0.2434,
+      "step": 4015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252653494543883e-05,
+      "loss": 0.2639,
+      "step": 4016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5250290883077114e-05,
+      "loss": 0.2601,
+      "step": 4017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5247927866943869e-05,
+      "loss": 0.2635,
+      "step": 4018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5245564446326273e-05,
+      "loss": 0.2798,
+      "step": 4019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5243200621406492e-05,
+      "loss": 0.284,
+      "step": 4020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5240836392366713e-05,
+      "loss": 0.2991,
+      "step": 4021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523847175938916e-05,
+      "loss": 0.304,
+      "step": 4022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523610672265609e-05,
+      "loss": 0.3409,
+      "step": 4023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5233741282349783e-05,
+      "loss": 0.3078,
+      "step": 4024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523137543865256e-05,
+      "loss": 0.2864,
+      "step": 4025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229009191746769e-05,
+      "loss": 0.2774,
+      "step": 4026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226642541814785e-05,
+      "loss": 0.2698,
+      "step": 4027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522427548903902e-05,
+      "loss": 0.2891,
+      "step": 4028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221908033601911e-05,
+      "loss": 0.3114,
+      "step": 4029
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219540175685938e-05,
+      "loss": 0.2944,
+      "step": 4030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217171915473592e-05,
+      "loss": 0.2927,
+      "step": 4031
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214803253147421e-05,
+      "loss": 0.348,
+      "step": 4032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521243418888998e-05,
+      "loss": 0.2881,
+      "step": 4033
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5210064722883865e-05,
+      "loss": 0.2661,
+      "step": 4034
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207694855311707e-05,
+      "loss": 0.2761,
+      "step": 4035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205324586356161e-05,
+      "loss": 0.2665,
+      "step": 4036
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202953916199916e-05,
+      "loss": 0.2978,
+      "step": 4037
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5200582845025688e-05,
+      "loss": 0.2997,
+      "step": 4038
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198211373016239e-05,
+      "loss": 0.3301,
+      "step": 4039
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195839500354337e-05,
+      "loss": 0.4129,
+      "step": 4040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193467227222803e-05,
+      "loss": 0.2791,
+      "step": 4041
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5191094553804476e-05,
+      "loss": 0.2906,
+      "step": 4042
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188721480282226e-05,
+      "loss": 0.3106,
+      "step": 4043
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5186348006838966e-05,
+      "loss": 0.2775,
+      "step": 4044
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183974133657628e-05,
+      "loss": 0.288,
+      "step": 4045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181599860921182e-05,
+      "loss": 0.2826,
+      "step": 4046
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179225188812616e-05,
+      "loss": 0.6179,
+      "step": 4047
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5176850117514964e-05,
+      "loss": 0.3457,
+      "step": 4048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5174474647211291e-05,
+      "loss": 0.2855,
+      "step": 4049
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5172098778084672e-05,
+      "loss": 0.3423,
+      "step": 4050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169722510318242e-05,
+      "loss": 0.2305,
+      "step": 4051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516734584409514e-05,
+      "loss": 0.3118,
+      "step": 4052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164968779598558e-05,
+      "loss": 0.3058,
+      "step": 4053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.51625913170117e-05,
+      "loss": 0.2775,
+      "step": 4054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5160213456517812e-05,
+      "loss": 0.2543,
+      "step": 4055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5157835198300169e-05,
+      "loss": 0.2903,
+      "step": 4056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155456542542072e-05,
+      "loss": 0.2977,
+      "step": 4057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5153077489426865e-05,
+      "loss": 0.2975,
+      "step": 4058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150698039137903e-05,
+      "loss": 0.3013,
+      "step": 4059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148318191858588e-05,
+      "loss": 0.2806,
+      "step": 4060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5145937947772344e-05,
+      "loss": 0.3035,
+      "step": 4061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514355730706263e-05,
+      "loss": 0.3047,
+      "step": 4062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5141176269912931e-05,
+      "loss": 0.2446,
+      "step": 4063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138794836506772e-05,
+      "loss": 0.3136,
+      "step": 4064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136413007027699e-05,
+      "loss": 0.2704,
+      "step": 4065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5134030781659288e-05,
+      "loss": 0.3084,
+      "step": 4066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131648160585155e-05,
+      "loss": 0.2347,
+      "step": 4067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512926514398894e-05,
+      "loss": 0.278,
+      "step": 4068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126881732054305e-05,
+      "loss": 0.2468,
+      "step": 4069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124497924964966e-05,
+      "loss": 0.3073,
+      "step": 4070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5122113722904646e-05,
+      "loss": 0.2711,
+      "step": 4071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5119729126057108e-05,
+      "loss": 0.2786,
+      "step": 4072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117344134606146e-05,
+      "loss": 0.3053,
+      "step": 4073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114958748735584e-05,
+      "loss": 0.2523,
+      "step": 4074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112572968629275e-05,
+      "loss": 0.2957,
+      "step": 4075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110186794471105e-05,
+      "loss": 0.2795,
+      "step": 4076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107800226444988e-05,
+      "loss": 0.3082,
+      "step": 4077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5105413264734866e-05,
+      "loss": 0.2845,
+      "step": 4078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103025909524718e-05,
+      "loss": 0.2731,
+      "step": 4079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5100638160998544e-05,
+      "loss": 0.6196,
+      "step": 4080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098250019340385e-05,
+      "loss": 0.3105,
+      "step": 4081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095861484734307e-05,
+      "loss": 0.2779,
+      "step": 4082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093472557364404e-05,
+      "loss": 0.2938,
+      "step": 4083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091083237414806e-05,
+      "loss": 0.2758,
+      "step": 4084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5088693525069664e-05,
+      "loss": 0.2797,
+      "step": 4085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508630342051317e-05,
+      "loss": 0.3054,
+      "step": 4086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5083912923929545e-05,
+      "loss": 0.2538,
+      "step": 4087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5081522035503025e-05,
+      "loss": 0.2682,
+      "step": 4088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50791307554179e-05,
+      "loss": 0.2856,
+      "step": 4089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076739083858472e-05,
+      "loss": 0.297,
+      "step": 4090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5074347021009081e-05,
+      "loss": 0.2659,
+      "step": 4091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5071954567054094e-05,
+      "loss": 0.3081,
+      "step": 4092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506956172217791e-05,
+      "loss": 0.3233,
+      "step": 4093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067168486564959e-05,
+      "loss": 0.2972,
+      "step": 4094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064774860399699e-05,
+      "loss": 0.3282,
+      "step": 4095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5062380843866618e-05,
+      "loss": 0.2553,
+      "step": 4096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059986437150233e-05,
+      "loss": 0.3025,
+      "step": 4097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057591640435098e-05,
+      "loss": 0.2498,
+      "step": 4098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505519645390579e-05,
+      "loss": 0.2912,
+      "step": 4099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052800877746915e-05,
+      "loss": 0.2963,
+      "step": 4100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050404912143118e-05,
+      "loss": 0.3167,
+      "step": 4101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048008557279064e-05,
+      "loss": 0.3188,
+      "step": 4102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045611813339456e-05,
+      "loss": 0.3483,
+      "step": 4103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043214680509018e-05,
+      "loss": 0.2523,
+      "step": 4104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040817158972509e-05,
+      "loss": 0.2682,
+      "step": 4105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038419248914725e-05,
+      "loss": 0.3566,
+      "step": 4106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036020950520476e-05,
+      "loss": 0.3027,
+      "step": 4107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5033622263974621e-05,
+      "loss": 0.2971,
+      "step": 4108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5031223189462031e-05,
+      "loss": 0.2941,
+      "step": 4109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028823727167621e-05,
+      "loss": 0.3111,
+      "step": 4110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5026423877276322e-05,
+      "loss": 0.3297,
+      "step": 4111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024023639973109e-05,
+      "loss": 0.3239,
+      "step": 4112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021623015442976e-05,
+      "loss": 0.3541,
+      "step": 4113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019222003870954e-05,
+      "loss": 0.3144,
+      "step": 4114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016820605442105e-05,
+      "loss": 0.2922,
+      "step": 4115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501441882034151e-05,
+      "loss": 0.321,
+      "step": 4116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012016648754291e-05,
+      "loss": 0.2629,
+      "step": 4117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500961409086559e-05,
+      "loss": 0.2981,
+      "step": 4118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500721114686059e-05,
+      "loss": 0.2495,
+      "step": 4119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5004807816924494e-05,
+      "loss": 0.3297,
+      "step": 4120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002404101242543e-05,
+      "loss": 0.2517,
+      "step": 4121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.3335,
+      "step": 4122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4997595513382166e-05,
+      "loss": 0.3159,
+      "step": 4123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4995190641574361e-05,
+      "loss": 0.3029,
+      "step": 4124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4992785384761945e-05,
+      "loss": 0.2993,
+      "step": 4125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49903797431303e-05,
+      "loss": 0.2734,
+      "step": 4126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987973716864843e-05,
+      "loss": 0.3224,
+      "step": 4127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985567306151018e-05,
+      "loss": 0.3124,
+      "step": 4128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4983160511174302e-05,
+      "loss": 0.2655,
+      "step": 4129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980753332120193e-05,
+      "loss": 0.3051,
+      "step": 4130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497834576917423e-05,
+      "loss": 0.2711,
+      "step": 4131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4975937822521972e-05,
+      "loss": 0.3042,
+      "step": 4132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973529492349013e-05,
+      "loss": 0.3492,
+      "step": 4133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497112077884098e-05,
+      "loss": 0.2485,
+      "step": 4134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4968711682183515e-05,
+      "loss": 0.2875,
+      "step": 4135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966302202562308e-05,
+      "loss": 0.2903,
+      "step": 4136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963892340163067e-05,
+      "loss": 0.6328,
+      "step": 4137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961482095171529e-05,
+      "loss": 0.2867,
+      "step": 4138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4959071467773467e-05,
+      "loss": 0.329,
+      "step": 4139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4956660458154679e-05,
+      "loss": 0.2579,
+      "step": 4140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954249066501e-05,
+      "loss": 0.2817,
+      "step": 4141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951837292998277e-05,
+      "loss": 0.3085,
+      "step": 4142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4949425137832406e-05,
+      "loss": 0.2899,
+      "step": 4143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947012601189299e-05,
+      "loss": 0.2602,
+      "step": 4144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4944599683254903e-05,
+      "loss": 0.3393,
+      "step": 4145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942186384215198e-05,
+      "loss": 0.2863,
+      "step": 4146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939772704256187e-05,
+      "loss": 0.3138,
+      "step": 4147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4937358643563906e-05,
+      "loss": 0.3074,
+      "step": 4148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934944202324413e-05,
+      "loss": 0.3161,
+      "step": 4149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4932529380723806e-05,
+      "loss": 0.2722,
+      "step": 4150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930114178948207e-05,
+      "loss": 0.2604,
+      "step": 4151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927698597183768e-05,
+      "loss": 0.3197,
+      "step": 4152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4925282635616671e-05,
+      "loss": 0.2986,
+      "step": 4153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922866294433122e-05,
+      "loss": 0.2798,
+      "step": 4154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920449573819366e-05,
+      "loss": 0.2947,
+      "step": 4155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491803247396167e-05,
+      "loss": 0.2785,
+      "step": 4156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915614995046329e-05,
+      "loss": 0.3163,
+      "step": 4157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913197137259675e-05,
+      "loss": 0.3738,
+      "step": 4158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910778900788061e-05,
+      "loss": 0.2656,
+      "step": 4159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908360285817875e-05,
+      "loss": 0.3268,
+      "step": 4160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490594129253553e-05,
+      "loss": 0.344,
+      "step": 4161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4903521921127472e-05,
+      "loss": 0.5894,
+      "step": 4162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901102171780175e-05,
+      "loss": 0.3341,
+      "step": 4163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4898682044680135e-05,
+      "loss": 0.2888,
+      "step": 4164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896261540013894e-05,
+      "loss": 0.6628,
+      "step": 4165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893840657968001e-05,
+      "loss": 0.2863,
+      "step": 4166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891419398729057e-05,
+      "loss": 0.2621,
+      "step": 4167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488899776248367e-05,
+      "loss": 0.2929,
+      "step": 4168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886575749418494e-05,
+      "loss": 0.3234,
+      "step": 4169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884153359720205e-05,
+      "loss": 0.3334,
+      "step": 4170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488173059357551e-05,
+      "loss": 0.269,
+      "step": 4171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4879307451171141e-05,
+      "loss": 0.2747,
+      "step": 4172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4876883932693864e-05,
+      "loss": 0.3294,
+      "step": 4173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874460038330469e-05,
+      "loss": 0.2809,
+      "step": 4174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487203576826778e-05,
+      "loss": 0.296,
+      "step": 4175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869611122692649e-05,
+      "loss": 0.3059,
+      "step": 4176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4867186101791951e-05,
+      "loss": 0.3608,
+      "step": 4177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48647607057526e-05,
+      "loss": 0.2745,
+      "step": 4178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4862334934761533e-05,
+      "loss": 0.3418,
+      "step": 4179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485990878900571e-05,
+      "loss": 0.2679,
+      "step": 4180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857482268672136e-05,
+      "loss": 0.2737,
+      "step": 4181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4855055373947829e-05,
+      "loss": 0.2919,
+      "step": 4182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485262810501984e-05,
+      "loss": 0.3055,
+      "step": 4183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850200462075255e-05,
+      "loss": 0.3151,
+      "step": 4184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847772445301186e-05,
+      "loss": 0.2498,
+      "step": 4185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845344054884772e-05,
+      "loss": 0.2716,
+      "step": 4186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842915291013176e-05,
+      "loss": 0.3465,
+      "step": 4187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840486153873599e-05,
+      "loss": 0.2859,
+      "step": 4188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483805664365327e-05,
+      "loss": 0.2778,
+      "step": 4189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4835626760539437e-05,
+      "loss": 0.3149,
+      "step": 4190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833196504719389e-05,
+      "loss": 0.308,
+      "step": 4191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4830765876380438e-05,
+      "loss": 0.2543,
+      "step": 4192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482833487570992e-05,
+      "loss": 0.3069,
+      "step": 4193
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4825903502895207e-05,
+      "loss": 0.3235,
+      "step": 4194
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823471758123697e-05,
+      "loss": 0.2879,
+      "step": 4195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482103964158282e-05,
+      "loss": 0.2627,
+      "step": 4196
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818607153460025e-05,
+      "loss": 0.2626,
+      "step": 4197
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816174293942804e-05,
+      "loss": 0.3358,
+      "step": 4198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813741063218662e-05,
+      "loss": 0.2737,
+      "step": 4199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811307461475151e-05,
+      "loss": 0.3,
+      "step": 4200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4808873488899829e-05,
+      "loss": 0.2836,
+      "step": 4201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4806439145680298e-05,
+      "loss": 0.2899,
+      "step": 4202
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804004432004191e-05,
+      "loss": 0.2928,
+      "step": 4203
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801569348059158e-05,
+      "loss": 0.2711,
+      "step": 4204
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799133894032887e-05,
+      "loss": 0.3206,
+      "step": 4205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4796698070113084e-05,
+      "loss": 0.3285,
+      "step": 4206
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4794261876487496e-05,
+      "loss": 0.2906,
+      "step": 4207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791825313343896e-05,
+      "loss": 0.3313,
+      "step": 4208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4789388380870074e-05,
+      "loss": 0.257,
+      "step": 4209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786951079253861e-05,
+      "loss": 0.3441,
+      "step": 4210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4784513408683115e-05,
+      "loss": 0.299,
+      "step": 4211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782075369345715e-05,
+      "loss": 0.2814,
+      "step": 4212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4779636961429573e-05,
+      "loss": 0.257,
+      "step": 4213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477719818512263e-05,
+      "loss": 0.3127,
+      "step": 4214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774759040612859e-05,
+      "loss": 0.2875,
+      "step": 4215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477231952808825e-05,
+      "loss": 0.3006,
+      "step": 4216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769879647736835e-05,
+      "loss": 0.3078,
+      "step": 4217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4767439399746666e-05,
+      "loss": 0.255,
+      "step": 4218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764998784305825e-05,
+      "loss": 0.2906,
+      "step": 4219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4762557801602422e-05,
+      "loss": 0.2764,
+      "step": 4220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476011645182459e-05,
+      "loss": 0.2819,
+      "step": 4221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757674735160512e-05,
+      "loss": 0.2874,
+      "step": 4222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755232651798368e-05,
+      "loss": 0.2423,
+      "step": 4223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475279020192639e-05,
+      "loss": 0.3406,
+      "step": 4224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750347385732826e-05,
+      "loss": 0.2884,
+      "step": 4225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747904203405959e-05,
+      "loss": 0.3337,
+      "step": 4226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745460655134091e-05,
+      "loss": 0.2497,
+      "step": 4227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474301674110557e-05,
+      "loss": 0.3538,
+      "step": 4228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740572461508753e-05,
+      "loss": 0.3043,
+      "step": 4229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738127816532034e-05,
+      "loss": 0.2672,
+      "step": 4230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4735682806363834e-05,
+      "loss": 0.3094,
+      "step": 4231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733237431192604e-05,
+      "loss": 0.2589,
+      "step": 4232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730791691206818e-05,
+      "loss": 0.3208,
+      "step": 4233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4728345586594986e-05,
+      "loss": 0.3356,
+      "step": 4234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725899117545638e-05,
+      "loss": 0.2686,
+      "step": 4235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723452284247341e-05,
+      "loss": 0.3402,
+      "step": 4236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4721005086888678e-05,
+      "loss": 0.2506,
+      "step": 4237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718557525658272e-05,
+      "loss": 0.3091,
+      "step": 4238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716109600744766e-05,
+      "loss": 0.2957,
+      "step": 4239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713661312336832e-05,
+      "loss": 0.2809,
+      "step": 4240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711212660623181e-05,
+      "loss": 0.2968,
+      "step": 4241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708763645792531e-05,
+      "loss": 0.3277,
+      "step": 4242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4706314268033652e-05,
+      "loss": 0.2744,
+      "step": 4243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703864527535321e-05,
+      "loss": 0.3157,
+      "step": 4244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4701414424486353e-05,
+      "loss": 0.3535,
+      "step": 4245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698963959075592e-05,
+      "loss": 0.3309,
+      "step": 4246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696513131491907e-05,
+      "loss": 0.2908,
+      "step": 4247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694061941924199e-05,
+      "loss": 0.2975,
+      "step": 4248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4691610390561389e-05,
+      "loss": 0.286,
+      "step": 4249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4689158477592433e-05,
+      "loss": 0.3047,
+      "step": 4250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686706203206309e-05,
+      "loss": 0.2431,
+      "step": 4251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684253567592029e-05,
+      "loss": 0.3116,
+      "step": 4252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4681800570938628e-05,
+      "loss": 0.2908,
+      "step": 4253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4679347213435176e-05,
+      "loss": 0.2996,
+      "step": 4254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4676893495270762e-05,
+      "loss": 0.2726,
+      "step": 4255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4674439416634505e-05,
+      "loss": 0.3203,
+      "step": 4256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4671984977715556e-05,
+      "loss": 0.329,
+      "step": 4257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4669530178703089e-05,
+      "loss": 0.288,
+      "step": 4258
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4667075019786306e-05,
+      "loss": 0.2962,
+      "step": 4259
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4664619501154445e-05,
+      "loss": 0.2955,
+      "step": 4260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662163622996758e-05,
+      "loss": 0.296,
+      "step": 4261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465970738550254e-05,
+      "loss": 0.2678,
+      "step": 4262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657250788861099e-05,
+      "loss": 0.2631,
+      "step": 4263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4654793833261777e-05,
+      "loss": 0.2744,
+      "step": 4264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4652336518893948e-05,
+      "loss": 0.2983,
+      "step": 4265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464987884594701e-05,
+      "loss": 0.2576,
+      "step": 4266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647420814610384e-05,
+      "loss": 0.2767,
+      "step": 4267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4644962425073526e-05,
+      "loss": 0.3093,
+      "step": 4268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642503677525917e-05,
+      "loss": 0.6454,
+      "step": 4269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4640044572157062e-05,
+      "loss": 0.3005,
+      "step": 4270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637585109156498e-05,
+      "loss": 0.254,
+      "step": 4271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4635125288713789e-05,
+      "loss": 0.3408,
+      "step": 4272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632665111018525e-05,
+      "loss": 0.3262,
+      "step": 4273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630204576260328e-05,
+      "loss": 0.282,
+      "step": 4274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627743684628838e-05,
+      "loss": 0.278,
+      "step": 4275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4625282436313733e-05,
+      "loss": 0.2499,
+      "step": 4276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622820831504712e-05,
+      "loss": 0.2925,
+      "step": 4277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46203588703915e-05,
+      "loss": 0.2936,
+      "step": 4278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461789655316386e-05,
+      "loss": 0.3112,
+      "step": 4279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461543388001157e-05,
+      "loss": 0.2738,
+      "step": 4280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612970851124442e-05,
+      "loss": 0.3252,
+      "step": 4281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4610507466692312e-05,
+      "loss": 0.2949,
+      "step": 4282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460804372690505e-05,
+      "loss": 0.2961,
+      "step": 4283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4605579631952544e-05,
+      "loss": 0.2932,
+      "step": 4284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603115182024721e-05,
+      "loss": 0.2913,
+      "step": 4285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600650377311523e-05,
+      "loss": 0.323,
+      "step": 4286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598185218002925e-05,
+      "loss": 0.2891,
+      "step": 4287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595719704288932e-05,
+      "loss": 0.2695,
+      "step": 4288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593253836359573e-05,
+      "loss": 0.2529,
+      "step": 4289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4590787614404902e-05,
+      "loss": 0.3531,
+      "step": 4290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4588321038615005e-05,
+      "loss": 0.2762,
+      "step": 4291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4585854109179995e-05,
+      "loss": 0.2966,
+      "step": 4292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583386826290013e-05,
+      "loss": 0.2784,
+      "step": 4293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4580919190135219e-05,
+      "loss": 0.2909,
+      "step": 4294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457845120090581e-05,
+      "loss": 0.2937,
+      "step": 4295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575982858792002e-05,
+      "loss": 0.3417,
+      "step": 4296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573514163984044e-05,
+      "loss": 0.3081,
+      "step": 4297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571045116672219e-05,
+      "loss": 0.2868,
+      "step": 4298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568575717046819e-05,
+      "loss": 0.3148,
+      "step": 4299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566105965298179e-05,
+      "loss": 0.2949,
+      "step": 4300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4563635861616652e-05,
+      "loss": 0.3117,
+      "step": 4301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561165406192622e-05,
+      "loss": 0.3089,
+      "step": 4302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558694599216496e-05,
+      "loss": 0.2485,
+      "step": 4303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455622344087872e-05,
+      "loss": 0.3022,
+      "step": 4304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4553751931369755e-05,
+      "loss": 0.3172,
+      "step": 4305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551280070880089e-05,
+      "loss": 0.2928,
+      "step": 4306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4548807859600248e-05,
+      "loss": 0.3112,
+      "step": 4307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546335297720769e-05,
+      "loss": 0.2914,
+      "step": 4308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454386238543223e-05,
+      "loss": 0.3488,
+      "step": 4309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4541389122925229e-05,
+      "loss": 0.3497,
+      "step": 4310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538915510390397e-05,
+      "loss": 0.2661,
+      "step": 4311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4536441548018385e-05,
+      "loss": 0.2687,
+      "step": 4312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4533967235999872e-05,
+      "loss": 0.3354,
+      "step": 4313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453149257452557e-05,
+      "loss": 0.2734,
+      "step": 4314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529017563786208e-05,
+      "loss": 0.2393,
+      "step": 4315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452654220397255e-05,
+      "loss": 0.3446,
+      "step": 4316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524066495275388e-05,
+      "loss": 0.2862,
+      "step": 4317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4521590437885533e-05,
+      "loss": 0.2607,
+      "step": 4318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451911403199383e-05,
+      "loss": 0.2426,
+      "step": 4319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4516637277791149e-05,
+      "loss": 0.3161,
+      "step": 4320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514160175468379e-05,
+      "loss": 0.3187,
+      "step": 4321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451168272521645e-05,
+      "loss": 0.2853,
+      "step": 4322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4509204927226307e-05,
+      "loss": 0.3265,
+      "step": 4323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4506726781688935e-05,
+      "loss": 0.3212,
+      "step": 4324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4504248288795328e-05,
+      "loss": 0.2377,
+      "step": 4325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450176944873652e-05,
+      "loss": 0.3303,
+      "step": 4326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4499290261703565e-05,
+      "loss": 0.3043,
+      "step": 4327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4496810727887547e-05,
+      "loss": 0.2609,
+      "step": 4328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449433084747958e-05,
+      "loss": 0.2778,
+      "step": 4329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4491850620670798e-05,
+      "loss": 0.3074,
+      "step": 4330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4489370047652364e-05,
+      "loss": 0.2453,
+      "step": 4331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486889128615472e-05,
+      "loss": 0.2794,
+      "step": 4332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484407863751335e-05,
+      "loss": 0.2837,
+      "step": 4333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481926253251197e-05,
+      "loss": 0.2922,
+      "step": 4334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4479444297306326e-05,
+      "loss": 0.2958,
+      "step": 4335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4476961996108027e-05,
+      "loss": 0.2794,
+      "step": 4336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4474479349847617e-05,
+      "loss": 0.2917,
+      "step": 4337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471996358716451e-05,
+      "loss": 0.311,
+      "step": 4338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4469513022905898e-05,
+      "loss": 0.3389,
+      "step": 4339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4467029342607368e-05,
+      "loss": 0.3246,
+      "step": 4340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464545318012286e-05,
+      "loss": 0.2924,
+      "step": 4341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462060949312114e-05,
+      "loss": 0.3387,
+      "step": 4342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4459576236698331e-05,
+      "loss": 0.2733,
+      "step": 4343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4457091180362445e-05,
+      "loss": 0.2754,
+      "step": 4344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4454605780495998e-05,
+      "loss": 0.2828,
+      "step": 4345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452120037290547e-05,
+      "loss": 0.3292,
+      "step": 4346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4449633950937678e-05,
+      "loss": 0.3133,
+      "step": 4347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447147521629013e-05,
+      "loss": 0.2632,
+      "step": 4348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4444660749556192e-05,
+      "loss": 0.2484,
+      "step": 4349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442173634910881e-05,
+      "loss": 0.2896,
+      "step": 4350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4439686177884778e-05,
+      "loss": 0.2784,
+      "step": 4351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4437198378669598e-05,
+      "loss": 0.2953,
+      "step": 4352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434710237457094e-05,
+      "loss": 0.3451,
+      "step": 4353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432221754439037e-05,
+      "loss": 0.2583,
+      "step": 4354
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4429732929807227e-05,
+      "loss": 0.2822,
+      "step": 4355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427243763753488e-05,
+      "loss": 0.25,
+      "step": 4356
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424754256469681e-05,
+      "loss": 0.2886,
+      "step": 4357
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422264408147676e-05,
+      "loss": 0.2718,
+      "step": 4358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4419774218979383e-05,
+      "loss": 0.283,
+      "step": 4359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4417283689156731e-05,
+      "loss": 0.2921,
+      "step": 4360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414792818871676e-05,
+      "loss": 0.3076,
+      "step": 4361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441230160831621e-05,
+      "loss": 0.3334,
+      "step": 4362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409810057682333e-05,
+      "loss": 0.2635,
+      "step": 4363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4407318167162092e-05,
+      "loss": 0.3124,
+      "step": 4364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404825936947539e-05,
+      "loss": 0.3156,
+      "step": 4365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440233336723077e-05,
+      "loss": 0.3383,
+      "step": 4366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4399840458203896e-05,
+      "loss": 0.2731,
+      "step": 4367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397347210059059e-05,
+      "loss": 0.2467,
+      "step": 4368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439485362298843e-05,
+      "loss": 0.3231,
+      "step": 4369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392359697184197e-05,
+      "loss": 0.2736,
+      "step": 4370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4389865432838583e-05,
+      "loss": 0.3091,
+      "step": 4371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387370830143832e-05,
+      "loss": 0.2951,
+      "step": 4372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4384875889292216e-05,
+      "loss": 0.2574,
+      "step": 4373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4382380610476032e-05,
+      "loss": 0.3156,
+      "step": 4374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379884993887605e-05,
+      "loss": 0.2957,
+      "step": 4375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377389039719285e-05,
+      "loss": 0.3148,
+      "step": 4376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4374892748163447e-05,
+      "loss": 0.3152,
+      "step": 4377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4372396119412493e-05,
+      "loss": 0.2703,
+      "step": 4378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369899153658848e-05,
+      "loss": 0.2795,
+      "step": 4379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436740185109497e-05,
+      "loss": 0.3353,
+      "step": 4380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436490421191334e-05,
+      "loss": 0.3352,
+      "step": 4381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436240623630646e-05,
+      "loss": 0.3149,
+      "step": 4382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4359907924466863e-05,
+      "loss": 0.2802,
+      "step": 4383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357409276587105e-05,
+      "loss": 0.2801,
+      "step": 4384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354910292859769e-05,
+      "loss": 0.2872,
+      "step": 4385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352410973477466e-05,
+      "loss": 0.2979,
+      "step": 4386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349911318632832e-05,
+      "loss": 0.3075,
+      "step": 4387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434741132851853e-05,
+      "loss": 0.278,
+      "step": 4388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434491100332724e-05,
+      "loss": 0.2346,
+      "step": 4389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4342410343251683e-05,
+      "loss": 0.3038,
+      "step": 4390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4339909348484589e-05,
+      "loss": 0.2928,
+      "step": 4391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337408019218728e-05,
+      "loss": 0.2984,
+      "step": 4392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334906355646887e-05,
+      "loss": 0.2754,
+      "step": 4393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332404357961884e-05,
+      "loss": 0.2935,
+      "step": 4394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4329902026356564e-05,
+      "loss": 0.2723,
+      "step": 4395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327399361023785e-05,
+      "loss": 0.29,
+      "step": 4396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324896362156451e-05,
+      "loss": 0.2786,
+      "step": 4397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432239302994747e-05,
+      "loss": 0.3416,
+      "step": 4398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319889364589794e-05,
+      "loss": 0.3063,
+      "step": 4399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317385366276393e-05,
+      "loss": 0.2725,
+      "step": 4400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314881035200259e-05,
+      "loss": 0.2886,
+      "step": 4401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312376371554417e-05,
+      "loss": 0.291,
+      "step": 4402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430987137553191e-05,
+      "loss": 0.2755,
+      "step": 4403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307366047325814e-05,
+      "loss": 0.3079,
+      "step": 4404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304860387129225e-05,
+      "loss": 0.2746,
+      "step": 4405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302354395135269e-05,
+      "loss": 0.2892,
+      "step": 4406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299848071537097e-05,
+      "loss": 0.2817,
+      "step": 4407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4297341416527881e-05,
+      "loss": 0.2639,
+      "step": 4408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294834430300822e-05,
+      "loss": 0.2954,
+      "step": 4409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4292327113049145e-05,
+      "loss": 0.3021,
+      "step": 4410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4289819464966104e-05,
+      "loss": 0.3441,
+      "step": 4411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4287311486244975e-05,
+      "loss": 0.3043,
+      "step": 4412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428480317707906e-05,
+      "loss": 0.2898,
+      "step": 4413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4282294537661692e-05,
+      "loss": 0.3163,
+      "step": 4414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4279785568186217e-05,
+      "loss": 0.2556,
+      "step": 4415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4277276268846017e-05,
+      "loss": 0.3167,
+      "step": 4416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4274766639834498e-05,
+      "loss": 0.2766,
+      "step": 4417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272256681345087e-05,
+      "loss": 0.2863,
+      "step": 4418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269746393571244e-05,
+      "loss": 0.2793,
+      "step": 4419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4267235776706445e-05,
+      "loss": 0.3325,
+      "step": 4420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4264724830944198e-05,
+      "loss": 0.2736,
+      "step": 4421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262213556478033e-05,
+      "loss": 0.2925,
+      "step": 4422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259701953501509e-05,
+      "loss": 0.2755,
+      "step": 4423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257190022208203e-05,
+      "loss": 0.2681,
+      "step": 4424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254677762791727e-05,
+      "loss": 0.3132,
+      "step": 4425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425216517544571e-05,
+      "loss": 0.326,
+      "step": 4426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4249652260363815e-05,
+      "loss": 0.6044,
+      "step": 4427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247139017739722e-05,
+      "loss": 0.3224,
+      "step": 4428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244625447767138e-05,
+      "loss": 0.311,
+      "step": 4429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4242111550639797e-05,
+      "loss": 0.2689,
+      "step": 4430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239597326551459e-05,
+      "loss": 0.3358,
+      "step": 4431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237082775695907e-05,
+      "loss": 0.3406,
+      "step": 4432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4234567898266954e-05,
+      "loss": 0.2959,
+      "step": 4433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423205269445843e-05,
+      "loss": 0.3246,
+      "step": 4434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422953716446419e-05,
+      "loss": 0.6313,
+      "step": 4435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4227021308478129e-05,
+      "loss": 0.3161,
+      "step": 4436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224505126694153e-05,
+      "loss": 0.2886,
+      "step": 4437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4221988619306192e-05,
+      "loss": 0.3649,
+      "step": 4438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219471786508212e-05,
+      "loss": 0.2962,
+      "step": 4439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4216954628494195e-05,
+      "loss": 0.296,
+      "step": 4440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214437145458153e-05,
+      "loss": 0.2706,
+      "step": 4441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4211919337594118e-05,
+      "loss": 0.2719,
+      "step": 4442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420940120509615e-05,
+      "loss": 0.2982,
+      "step": 4443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206882748158341e-05,
+      "loss": 0.2864,
+      "step": 4444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4204363966974798e-05,
+      "loss": 0.29,
+      "step": 4445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420184486173965e-05,
+      "loss": 0.2858,
+      "step": 4446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4199325432647067e-05,
+      "loss": 0.2722,
+      "step": 4447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196805679891225e-05,
+      "loss": 0.2625,
+      "step": 4448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4194285603666337e-05,
+      "loss": 0.3297,
+      "step": 4449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191765204166643e-05,
+      "loss": 0.2593,
+      "step": 4450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4189244481586398e-05,
+      "loss": 0.6046,
+      "step": 4451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186723436119887e-05,
+      "loss": 0.3358,
+      "step": 4452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184202067961422e-05,
+      "loss": 0.2955,
+      "step": 4453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181680377305336e-05,
+      "loss": 0.2835,
+      "step": 4454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179158364345986e-05,
+      "loss": 0.3147,
+      "step": 4455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176636029277764e-05,
+      "loss": 0.3148,
+      "step": 4456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4174113372295071e-05,
+      "loss": 0.3585,
+      "step": 4457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171590393592346e-05,
+      "loss": 0.2871,
+      "step": 4458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4169067093364047e-05,
+      "loss": 0.2687,
+      "step": 4459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166543471804653e-05,
+      "loss": 0.2679,
+      "step": 4460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4164019529108677e-05,
+      "loss": 0.2631,
+      "step": 4461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161495265470649e-05,
+      "loss": 0.2864,
+      "step": 4462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415897068108513e-05,
+      "loss": 0.3023,
+      "step": 4463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156445776146703e-05,
+      "loss": 0.2661,
+      "step": 4464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415392055084997e-05,
+      "loss": 0.3284,
+      "step": 4465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415139500538957e-05,
+      "loss": 0.3329,
+      "step": 4466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4148869139960151e-05,
+      "loss": 0.3131,
+      "step": 4467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146342954756402e-05,
+      "loss": 0.253,
+      "step": 4468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4143816449973025e-05,
+      "loss": 0.2968,
+      "step": 4469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141289625804748e-05,
+      "loss": 0.3346,
+      "step": 4470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4138762482446335e-05,
+      "loss": 0.2781,
+      "step": 4471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4136235020092558e-05,
+      "loss": 0.2485,
+      "step": 4472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133707238938222e-05,
+      "loss": 0.2839,
+      "step": 4473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131179139178157e-05,
+      "loss": 0.313,
+      "step": 4474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412865072100722e-05,
+      "loss": 0.2728,
+      "step": 4475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126121984620283e-05,
+      "loss": 0.2833,
+      "step": 4476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123592930212251e-05,
+      "loss": 0.2365,
+      "step": 4477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121063557978051e-05,
+      "loss": 0.2928,
+      "step": 4478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118533868112637e-05,
+      "loss": 0.343,
+      "step": 4479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411600386081098e-05,
+      "loss": 0.3048,
+      "step": 4480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4113473536268083e-05,
+      "loss": 0.2992,
+      "step": 4481
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110942894678971e-05,
+      "loss": 0.3042,
+      "step": 4482
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41084119362387e-05,
+      "loss": 0.2762,
+      "step": 4483
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4105880661142331e-05,
+      "loss": 0.2833,
+      "step": 4484
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103349069584971e-05,
+      "loss": 0.2654,
+      "step": 4485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100817161761738e-05,
+      "loss": 0.3018,
+      "step": 4486
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409828493786778e-05,
+      "loss": 0.257,
+      "step": 4487
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409575239809827e-05,
+      "loss": 0.273,
+      "step": 4488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4093219542648405e-05,
+      "loss": 0.2712,
+      "step": 4489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4090686371713403e-05,
+      "loss": 0.2744,
+      "step": 4490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4088152885488504e-05,
+      "loss": 0.2911,
+      "step": 4491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085619084168983e-05,
+      "loss": 0.2342,
+      "step": 4492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083084967950131e-05,
+      "loss": 0.2644,
+      "step": 4493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4080550537027264e-05,
+      "loss": 0.2812,
+      "step": 4494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4078015791595724e-05,
+      "loss": 0.2603,
+      "step": 4495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407548073185088e-05,
+      "loss": 0.2922,
+      "step": 4496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072945357988118e-05,
+      "loss": 0.3272,
+      "step": 4497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070409670202849e-05,
+      "loss": 0.3045,
+      "step": 4498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067873668690517e-05,
+      "loss": 0.288,
+      "step": 4499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065337353646583e-05,
+      "loss": 0.3287,
+      "step": 4500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4062800725266532e-05,
+      "loss": 0.2673,
+      "step": 4501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406026378374588e-05,
+      "loss": 0.301,
+      "step": 4502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057726529280154e-05,
+      "loss": 0.2833,
+      "step": 4503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4055188962064918e-05,
+      "loss": 0.3015,
+      "step": 4504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4052651082295754e-05,
+      "loss": 0.2495,
+      "step": 4505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405011289016827e-05,
+      "loss": 0.3339,
+      "step": 4506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047574385878095e-05,
+      "loss": 0.2712,
+      "step": 4507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4045035569620886e-05,
+      "loss": 0.3035,
+      "step": 4508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042496441592323e-05,
+      "loss": 0.3135,
+      "step": 4509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039957001988112e-05,
+      "loss": 0.3093,
+      "step": 4510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4037417251003972e-05,
+      "loss": 0.2986,
+      "step": 4511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034877188835662e-05,
+      "loss": 0.3021,
+      "step": 4512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4032336815678957e-05,
+      "loss": 0.3362,
+      "step": 4513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029796131729652e-05,
+      "loss": 0.2817,
+      "step": 4514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027255137183575e-05,
+      "loss": 0.2631,
+      "step": 4515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4024713832236571e-05,
+      "loss": 0.353,
+      "step": 4516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022172217084512e-05,
+      "loss": 0.2555,
+      "step": 4517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4019630291923289e-05,
+      "loss": 0.2841,
+      "step": 4518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4017088056948826e-05,
+      "loss": 0.2895,
+      "step": 4519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4014545512357068e-05,
+      "loss": 0.2599,
+      "step": 4520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4012002658343976e-05,
+      "loss": 0.2904,
+      "step": 4521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4009459495105542e-05,
+      "loss": 0.2791,
+      "step": 4522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4006916022837784e-05,
+      "loss": 0.2493,
+      "step": 4523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004372241736736e-05,
+      "loss": 0.2982,
+      "step": 4524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4001828151998462e-05,
+      "loss": 0.289,
+      "step": 4525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999283753819047e-05,
+      "loss": 0.3278,
+      "step": 4526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3996739047394601e-05,
+      "loss": 0.2567,
+      "step": 4527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399419403292126e-05,
+      "loss": 0.2598,
+      "step": 4528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991648710595179e-05,
+      "loss": 0.2757,
+      "step": 4529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3989103080612533e-05,
+      "loss": 0.2809,
+      "step": 4530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3986557143169539e-05,
+      "loss": 0.2926,
+      "step": 4531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3984010898462417e-05,
+      "loss": 0.2939,
+      "step": 4532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981464346687419e-05,
+      "loss": 0.2754,
+      "step": 4533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3978917488040822e-05,
+      "loss": 0.2684,
+      "step": 4534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3976370322718928e-05,
+      "loss": 0.6372,
+      "step": 4535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3973822850918055e-05,
+      "loss": 0.2732,
+      "step": 4536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3971275072834552e-05,
+      "loss": 0.2924,
+      "step": 4537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3968726988664788e-05,
+      "loss": 0.277,
+      "step": 4538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396617859860516e-05,
+      "loss": 0.3034,
+      "step": 4539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3963629902852082e-05,
+      "loss": 0.2946,
+      "step": 4540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961080901601996e-05,
+      "loss": 0.2529,
+      "step": 4541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3958531595051367e-05,
+      "loss": 0.2889,
+      "step": 4542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955981983396683e-05,
+      "loss": 0.2687,
+      "step": 4543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953432066834454e-05,
+      "loss": 0.282,
+      "step": 4544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3950881845561214e-05,
+      "loss": 0.2836,
+      "step": 4545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3948331319773525e-05,
+      "loss": 0.3462,
+      "step": 4546
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3945780489667968e-05,
+      "loss": 0.2766,
+      "step": 4547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943229355441145e-05,
+      "loss": 0.3083,
+      "step": 4548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940677917289689e-05,
+      "loss": 0.3028,
+      "step": 4549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393812617541025e-05,
+      "loss": 0.2505,
+      "step": 4550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935574129999504e-05,
+      "loss": 0.2574,
+      "step": 4551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3933021781254152e-05,
+      "loss": 0.2752,
+      "step": 4552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930469129370913e-05,
+      "loss": 0.3438,
+      "step": 4553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927916174546536e-05,
+      "loss": 0.2852,
+      "step": 4554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925362916977787e-05,
+      "loss": 0.3031,
+      "step": 4555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3922809356861462e-05,
+      "loss": 0.3129,
+      "step": 4556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3920255494394373e-05,
+      "loss": 0.3022,
+      "step": 4557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917701329773364e-05,
+      "loss": 0.2629,
+      "step": 4558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915146863195292e-05,
+      "loss": 0.2819,
+      "step": 4559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3912592094857044e-05,
+      "loss": 0.2679,
+      "step": 4560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910037024955534e-05,
+      "loss": 0.2628,
+      "step": 4561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3907481653687687e-05,
+      "loss": 0.2759,
+      "step": 4562
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390492598125046e-05,
+      "loss": 0.3445,
+      "step": 4563
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3902370007840835e-05,
+      "loss": 0.2772,
+      "step": 4564
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899813733655814e-05,
+      "loss": 0.3097,
+      "step": 4565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389725715889242e-05,
+      "loss": 0.2908,
+      "step": 4566
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894700283747697e-05,
+      "loss": 0.3004,
+      "step": 4567
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3892143108418723e-05,
+      "loss": 0.6012,
+      "step": 4568
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388958563310259e-05,
+      "loss": 0.3231,
+      "step": 4569
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887027857996416e-05,
+      "loss": 0.3204,
+      "step": 4570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884469783297339e-05,
+      "loss": 0.6304,
+      "step": 4571
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881911409202525e-05,
+      "loss": 0.255,
+      "step": 4572
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3879352735909163e-05,
+      "loss": 0.3339,
+      "step": 4573
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387679376361446e-05,
+      "loss": 0.315,
+      "step": 4574
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3874234492515649e-05,
+      "loss": 0.3082,
+      "step": 4575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3871674922809985e-05,
+      "loss": 0.2715,
+      "step": 4576
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386911505469475e-05,
+      "loss": 0.2662,
+      "step": 4577
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3866554888367243e-05,
+      "loss": 0.2839,
+      "step": 4578
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863994424024792e-05,
+      "loss": 0.2803,
+      "step": 4579
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861433661864744e-05,
+      "loss": 0.2603,
+      "step": 4580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858872602084467e-05,
+      "loss": 0.2668,
+      "step": 4581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385631124488136e-05,
+      "loss": 0.2881,
+      "step": 4582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853749590452834e-05,
+      "loss": 0.28,
+      "step": 4583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3851187638996331e-05,
+      "loss": 0.2723,
+      "step": 4584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3848625390709315e-05,
+      "loss": 0.2834,
+      "step": 4585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846062845789275e-05,
+      "loss": 0.2627,
+      "step": 4586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3843500004433708e-05,
+      "loss": 0.3146,
+      "step": 4587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840936866840155e-05,
+      "loss": 0.2851,
+      "step": 4588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838373433206167e-05,
+      "loss": 0.3091,
+      "step": 4589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835809703729322e-05,
+      "loss": 0.3173,
+      "step": 4590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3833245678607215e-05,
+      "loss": 0.2961,
+      "step": 4591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3830681358037477e-05,
+      "loss": 0.2628,
+      "step": 4592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828116742217744e-05,
+      "loss": 0.3027,
+      "step": 4593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3825551831345685e-05,
+      "loss": 0.2852,
+      "step": 4594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822986625618997e-05,
+      "loss": 0.2632,
+      "step": 4595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382042112523539e-05,
+      "loss": 0.2593,
+      "step": 4596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38178553303926e-05,
+      "loss": 0.3315,
+      "step": 4597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3815289241288383e-05,
+      "loss": 0.2971,
+      "step": 4598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3812722858120528e-05,
+      "loss": 0.2962,
+      "step": 4599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810156181086832e-05,
+      "loss": 0.2778,
+      "step": 4600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807589210385123e-05,
+      "loss": 0.3069,
+      "step": 4601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805021946213251e-05,
+      "loss": 0.2755,
+      "step": 4602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3802454388769091e-05,
+      "loss": 0.3104,
+      "step": 4603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799886538250534e-05,
+      "loss": 0.3015,
+      "step": 4604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3797318394855496e-05,
+      "loss": 0.2944,
+      "step": 4605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3794749958781924e-05,
+      "loss": 0.2846,
+      "step": 4606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3792181230227773e-05,
+      "loss": 0.3088,
+      "step": 4607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3789612209391031e-05,
+      "loss": 0.3127,
+      "step": 4608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3787042896469705e-05,
+      "loss": 0.2922,
+      "step": 4609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3784473291661824e-05,
+      "loss": 0.2703,
+      "step": 4610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3781903395165441e-05,
+      "loss": 0.309,
+      "step": 4611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3779333207178632e-05,
+      "loss": 0.3019,
+      "step": 4612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3776762727899494e-05,
+      "loss": 0.2759,
+      "step": 4613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3774191957526144e-05,
+      "loss": 0.2824,
+      "step": 4614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771620896256732e-05,
+      "loss": 0.2808,
+      "step": 4615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769049544289415e-05,
+      "loss": 0.2684,
+      "step": 4616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3766477901822379e-05,
+      "loss": 0.2996,
+      "step": 4617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763905969053841e-05,
+      "loss": 0.2628,
+      "step": 4618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761333746182028e-05,
+      "loss": 0.2499,
+      "step": 4619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3758761233405195e-05,
+      "loss": 0.2328,
+      "step": 4620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756188430921618e-05,
+      "loss": 0.234,
+      "step": 4621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3753615338929598e-05,
+      "loss": 0.35,
+      "step": 4622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3751041957627456e-05,
+      "loss": 0.3224,
+      "step": 4623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374846828721353e-05,
+      "loss": 0.3289,
+      "step": 4624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745894327886192e-05,
+      "loss": 0.2624,
+      "step": 4625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3743320079843828e-05,
+      "loss": 0.2701,
+      "step": 4626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740745543284852e-05,
+      "loss": 0.2567,
+      "step": 4627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3738170718407689e-05,
+      "loss": 0.3145,
+      "step": 4628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37355956054108e-05,
+      "loss": 0.3188,
+      "step": 4629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373302020449266e-05,
+      "loss": 0.3017,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3730444515851766e-05,
+      "loss": 0.2667,
+      "step": 4631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3727868539686641e-05,
+      "loss": 0.3004,
+      "step": 4632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3725292276195832e-05,
+      "loss": 0.3397,
+      "step": 4633
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3722715725577902e-05,
+      "loss": 0.3406,
+      "step": 4634
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720138888031436e-05,
+      "loss": 0.2543,
+      "step": 4635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717561763755045e-05,
+      "loss": 0.2812,
+      "step": 4636
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714984352947365e-05,
+      "loss": 0.3057,
+      "step": 4637
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712406655807047e-05,
+      "loss": 0.3049,
+      "step": 4638
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709828672532766e-05,
+      "loss": 0.2825,
+      "step": 4639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3707250403323222e-05,
+      "loss": 0.2918,
+      "step": 4640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704671848377136e-05,
+      "loss": 0.272,
+      "step": 4641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702093007893249e-05,
+      "loss": 0.2676,
+      "step": 4642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3699513882070323e-05,
+      "loss": 0.2919,
+      "step": 4643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369693447110715e-05,
+      "loss": 0.277,
+      "step": 4644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694354775202534e-05,
+      "loss": 0.6479,
+      "step": 4645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3691774794555306e-05,
+      "loss": 0.3091,
+      "step": 4646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368919452936432e-05,
+      "loss": 0.309,
+      "step": 4647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3686613979828444e-05,
+      "loss": 0.2812,
+      "step": 4648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684033146146585e-05,
+      "loss": 0.3076,
+      "step": 4649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368145202851765e-05,
+      "loss": 0.2703,
+      "step": 4650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678870627140585e-05,
+      "loss": 0.2853,
+      "step": 4651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676288942214348e-05,
+      "loss": 0.3292,
+      "step": 4652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673706973937928e-05,
+      "loss": 0.3247,
+      "step": 4653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3671124722510325e-05,
+      "loss": 0.2856,
+      "step": 4654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3668542188130567e-05,
+      "loss": 0.253,
+      "step": 4655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665959370997706e-05,
+      "loss": 0.2986,
+      "step": 4656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663376271310809e-05,
+      "loss": 0.3129,
+      "step": 4657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3660792889268967e-05,
+      "loss": 0.2698,
+      "step": 4658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3658209225071301e-05,
+      "loss": 0.2264,
+      "step": 4659
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655625278916947e-05,
+      "loss": 0.3402,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3653041051005056e-05,
+      "loss": 0.2313,
+      "step": 4661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650456541534811e-05,
+      "loss": 0.2849,
+      "step": 4662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3647871750705412e-05,
+      "loss": 0.2709,
+      "step": 4663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645286678716084e-05,
+      "loss": 0.2778,
+      "step": 4664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642701325766073e-05,
+      "loss": 0.2761,
+      "step": 4665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364011569205464e-05,
+      "loss": 0.2565,
+      "step": 4666
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3637529777781077e-05,
+      "loss": 0.2918,
+      "step": 4667
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634943583144693e-05,
+      "loss": 0.2708,
+      "step": 4668
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632357108344819e-05,
+      "loss": 0.2958,
+      "step": 4669
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3629770353580804e-05,
+      "loss": 0.2606,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627183319052026e-05,
+      "loss": 0.2998,
+      "step": 4671
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624596004957884e-05,
+      "loss": 0.2653,
+      "step": 4672
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622008411497787e-05,
+      "loss": 0.2742,
+      "step": 4673
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361942053887118e-05,
+      "loss": 0.3118,
+      "step": 4674
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616832387277525e-05,
+      "loss": 0.239,
+      "step": 4675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614243956916297e-05,
+      "loss": 0.2817,
+      "step": 4676
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3611655247987004e-05,
+      "loss": 0.2899,
+      "step": 4677
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360906626068917e-05,
+      "loss": 0.2753,
+      "step": 4678
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606476995222344e-05,
+      "loss": 0.2676,
+      "step": 4679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603887451786088e-05,
+      "loss": 0.3101,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3601297630579996e-05,
+      "loss": 0.3201,
+      "step": 4681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359870753180368e-05,
+      "loss": 0.2611,
+      "step": 4682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596117155656763e-05,
+      "loss": 0.2714,
+      "step": 4683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593526502338909e-05,
+      "loss": 0.2728,
+      "step": 4684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3590935572049787e-05,
+      "loss": 0.3516,
+      "step": 4685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588344364989096e-05,
+      "loss": 0.2943,
+      "step": 4686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358575288135655e-05,
+      "loss": 0.2857,
+      "step": 4687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358316112135189e-05,
+      "loss": 0.3317,
+      "step": 4688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580569085174877e-05,
+      "loss": 0.2747,
+      "step": 4689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357797677302529e-05,
+      "loss": 0.2361,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575384185102933e-05,
+      "loss": 0.2667,
+      "step": 4691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357279132160763e-05,
+      "loss": 0.2779,
+      "step": 4692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570198182739222e-05,
+      "loss": 0.2538,
+      "step": 4693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567604768697585e-05,
+      "loss": 0.2997,
+      "step": 4694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3565011079682597e-05,
+      "loss": 0.2556,
+      "step": 4695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356241711589417e-05,
+      "loss": 0.2922,
+      "step": 4696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3559822877532234e-05,
+      "loss": 0.6287,
+      "step": 4697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3557228364796742e-05,
+      "loss": 0.3558,
+      "step": 4698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554633577887663e-05,
+      "loss": 0.2886,
+      "step": 4699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552038517004991e-05,
+      "loss": 0.3183,
+      "step": 4700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3549443182348743e-05,
+      "loss": 0.2421,
+      "step": 4701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546847574118951e-05,
+      "loss": 0.3222,
+      "step": 4702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3544251692515675e-05,
+      "loss": 0.3259,
+      "step": 4703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541655537738992e-05,
+      "loss": 0.2913,
+      "step": 4704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3539059109988999e-05,
+      "loss": 0.2619,
+      "step": 4705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536462409465816e-05,
+      "loss": 0.3214,
+      "step": 4706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3533865436369584e-05,
+      "loss": 0.3145,
+      "step": 4707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3531268190900467e-05,
+      "loss": 0.3072,
+      "step": 4708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3528670673258645e-05,
+      "loss": 0.269,
+      "step": 4709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526072883644326e-05,
+      "loss": 0.2829,
+      "step": 4710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3523474822257729e-05,
+      "loss": 0.3028,
+      "step": 4711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520876489299104e-05,
+      "loss": 0.2947,
+      "step": 4712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3518277884968718e-05,
+      "loss": 0.2807,
+      "step": 4713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515679009466856e-05,
+      "loss": 0.3173,
+      "step": 4714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3513079862993825e-05,
+      "loss": 0.2651,
+      "step": 4715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3510480445749958e-05,
+      "loss": 0.2939,
+      "step": 4716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3507880757935605e-05,
+      "loss": 0.3118,
+      "step": 4717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3505280799751134e-05,
+      "loss": 0.2829,
+      "step": 4718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3502680571396943e-05,
+      "loss": 0.3098,
+      "step": 4719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3500080073073436e-05,
+      "loss": 0.3193,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3497479304981053e-05,
+      "loss": 0.2904,
+      "step": 4721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349487826732025e-05,
+      "loss": 0.2912,
+      "step": 4722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492276960291495e-05,
+      "loss": 0.3058,
+      "step": 4723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3489675384095291e-05,
+      "loss": 0.2505,
+      "step": 4724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3487073538932149e-05,
+      "loss": 0.3296,
+      "step": 4725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348447142500261e-05,
+      "loss": 0.5804,
+      "step": 4726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348186904250723e-05,
+      "loss": 0.3107,
+      "step": 4727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3479266391646588e-05,
+      "loss": 0.2668,
+      "step": 4728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476663472621286e-05,
+      "loss": 0.2903,
+      "step": 4729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347406028563194e-05,
+      "loss": 0.2913,
+      "step": 4730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3471456830879195e-05,
+      "loss": 0.3231,
+      "step": 4731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3468853108563709e-05,
+      "loss": 0.2511,
+      "step": 4732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3466249118886165e-05,
+      "loss": 0.2686,
+      "step": 4733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463644862047267e-05,
+      "loss": 0.2515,
+      "step": 4734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461040338247737e-05,
+      "loss": 0.3395,
+      "step": 4735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458435547688323e-05,
+      "loss": 0.2753,
+      "step": 4736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455830490569782e-05,
+      "loss": 0.2785,
+      "step": 4737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3453225167092902e-05,
+      "loss": 0.2432,
+      "step": 4738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450619577458488e-05,
+      "loss": 0.2924,
+      "step": 4739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344801372186737e-05,
+      "loss": 0.3099,
+      "step": 4740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344540760052039e-05,
+      "loss": 0.299,
+      "step": 4741
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442801213618417e-05,
+      "loss": 0.3189,
+      "step": 4742
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344019456136234e-05,
+      "loss": 0.2726,
+      "step": 4743
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437587643953062e-05,
+      "loss": 0.3393,
+      "step": 4744
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434980461591514e-05,
+      "loss": 0.2943,
+      "step": 4745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432373014478644e-05,
+      "loss": 0.3002,
+      "step": 4746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3429765302815423e-05,
+      "loss": 0.2913,
+      "step": 4747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427157326802843e-05,
+      "loss": 0.2818,
+      "step": 4748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3424549086641905e-05,
+      "loss": 0.2925,
+      "step": 4749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421940582533645e-05,
+      "loss": 0.3036,
+      "step": 4750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419331814679115e-05,
+      "loss": 0.3173,
+      "step": 4751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416722783279386e-05,
+      "loss": 0.2527,
+      "step": 4752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3414113488535542e-05,
+      "loss": 0.343,
+      "step": 4753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411503930648704e-05,
+      "loss": 0.2633,
+      "step": 4754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408894109820002e-05,
+      "loss": 0.2786,
+      "step": 4755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340628402625058e-05,
+      "loss": 0.2847,
+      "step": 4756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340367368014162e-05,
+      "loss": 0.2754,
+      "step": 4757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3401063071694309e-05,
+      "loss": 0.3082,
+      "step": 4758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339845220110986e-05,
+      "loss": 0.3127,
+      "step": 4759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3395841068589513e-05,
+      "loss": 0.2743,
+      "step": 4760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393229674334512e-05,
+      "loss": 0.2956,
+      "step": 4761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3390618018546135e-05,
+      "loss": 0.3106,
+      "step": 4762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388006101425674e-05,
+      "loss": 0.2979,
+      "step": 4763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3385393923174443e-05,
+      "loss": 0.3387,
+      "step": 4764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382781483993777e-05,
+      "loss": 0.3207,
+      "step": 4765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3380168784085028e-05,
+      "loss": 0.2396,
+      "step": 4766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377555823649573e-05,
+      "loss": 0.3193,
+      "step": 4767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374942602888803e-05,
+      "loss": 0.3192,
+      "step": 4768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3372329122004135e-05,
+      "loss": 0.2903,
+      "step": 4769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369715381197e-05,
+      "loss": 0.3156,
+      "step": 4770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367101380668852e-05,
+      "loss": 0.3001,
+      "step": 4771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3364487120621168e-05,
+      "loss": 0.2541,
+      "step": 4772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3361872601255441e-05,
+      "loss": 0.3005,
+      "step": 4773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359257822773187e-05,
+      "loss": 0.2616,
+      "step": 4774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3356642785375937e-05,
+      "loss": 0.2515,
+      "step": 4775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3354027489265246e-05,
+      "loss": 0.2652,
+      "step": 4776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335141193464269e-05,
+      "loss": 0.2809,
+      "step": 4777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3348796121709862e-05,
+      "loss": 0.2805,
+      "step": 4778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3346180050668376e-05,
+      "loss": 0.2776,
+      "step": 4779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343563721719865e-05,
+      "loss": 0.2867,
+      "step": 4780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3340947135065986e-05,
+      "loss": 0.3054,
+      "step": 4781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3338330290908408e-05,
+      "loss": 0.2798,
+      "step": 4782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335713189448824e-05,
+      "loss": 0.3165,
+      "step": 4783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3333095830888954e-05,
+      "loss": 0.2949,
+      "step": 4784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330478215430523e-05,
+      "loss": 0.2692,
+      "step": 4785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332786034327529e-05,
+      "loss": 0.317,
+      "step": 4786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3325242214625022e-05,
+      "loss": 0.2678,
+      "step": 4787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332262382968152e-05,
+      "loss": 0.3068,
+      "step": 4788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320005188646587e-05,
+      "loss": 0.3077,
+      "step": 4789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331738629172206e-05,
+      "loss": 0.2463,
+      "step": 4790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314767139109786e-05,
+      "loss": 0.2664,
+      "step": 4791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3312147731011642e-05,
+      "loss": 0.3241,
+      "step": 4792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3309528067629518e-05,
+      "loss": 0.2742,
+      "step": 4793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330690814916532e-05,
+      "loss": 0.2962,
+      "step": 4794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304287975820985e-05,
+      "loss": 0.3509,
+      "step": 4795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3301667547798458e-05,
+      "loss": 0.2891,
+      "step": 4796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3299046865299713e-05,
+      "loss": 0.2622,
+      "step": 4797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296425928526735e-05,
+      "loss": 0.2841,
+      "step": 4798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3293804737681533e-05,
+      "loss": 0.3107,
+      "step": 4799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291183292966141e-05,
+      "loss": 0.3146,
+      "step": 4800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288561594582599e-05,
+      "loss": 0.2556,
+      "step": 4801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285939642732979e-05,
+      "loss": 0.3171,
+      "step": 4802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283317437619369e-05,
+      "loss": 0.27,
+      "step": 4803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280694979443873e-05,
+      "loss": 0.3049,
+      "step": 4804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3278072268408621e-05,
+      "loss": 0.2706,
+      "step": 4805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3275449304715753e-05,
+      "loss": 0.2852,
+      "step": 4806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3272826088567441e-05,
+      "loss": 0.2865,
+      "step": 4807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270202620165861e-05,
+      "loss": 0.3076,
+      "step": 4808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3267578899713223e-05,
+      "loss": 0.3184,
+      "step": 4809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3264954927411751e-05,
+      "loss": 0.2737,
+      "step": 4810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262330703463682e-05,
+      "loss": 0.291,
+      "step": 4811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259706228071286e-05,
+      "loss": 0.2676,
+      "step": 4812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3257081501436839e-05,
+      "loss": 0.313,
+      "step": 4813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3254456523762643e-05,
+      "loss": 0.2609,
+      "step": 4814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251831295251019e-05,
+      "loss": 0.2407,
+      "step": 4815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3249205816104307e-05,
+      "loss": 0.2679,
+      "step": 4816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246580086524868e-05,
+      "loss": 0.3162,
+      "step": 4817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3243954106715074e-05,
+      "loss": 0.3313,
+      "step": 4818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3241327876877328e-05,
+      "loss": 0.2509,
+      "step": 4819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3238701397214044e-05,
+      "loss": 0.6091,
+      "step": 4820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236074667927659e-05,
+      "loss": 0.3187,
+      "step": 4821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233447689220629e-05,
+      "loss": 0.2988,
+      "step": 4822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230820461295429e-05,
+      "loss": 0.2989,
+      "step": 4823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228192984354552e-05,
+      "loss": 0.2506,
+      "step": 4824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225565258600507e-05,
+      "loss": 0.2597,
+      "step": 4825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3222937284235835e-05,
+      "loss": 0.2708,
+      "step": 4826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220309061463081e-05,
+      "loss": 0.3536,
+      "step": 4827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217680590484813e-05,
+      "loss": 0.2786,
+      "step": 4828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3215051871503628e-05,
+      "loss": 0.2586,
+      "step": 4829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321242290472213e-05,
+      "loss": 0.2776,
+      "step": 4830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209793690342947e-05,
+      "loss": 0.2984,
+      "step": 4831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207164228568725e-05,
+      "loss": 0.2921,
+      "step": 4832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3204534519602133e-05,
+      "loss": 0.6091,
+      "step": 4833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3201904563645853e-05,
+      "loss": 0.2847,
+      "step": 4834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319927436090259e-05,
+      "loss": 0.2813,
+      "step": 4835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3196643911575072e-05,
+      "loss": 0.2755,
+      "step": 4836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319401321586603e-05,
+      "loss": 0.5737,
+      "step": 4837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191382273978237e-05,
+      "loss": 0.3258,
+      "step": 4838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3188751086114464e-05,
+      "loss": 0.2497,
+      "step": 4839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3186119652477514e-05,
+      "loss": 0.3023,
+      "step": 4840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3183487973270204e-05,
+      "loss": 0.2723,
+      "step": 4841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.318085604869537e-05,
+      "loss": 0.2432,
+      "step": 4842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3178223878955874e-05,
+      "loss": 0.254,
+      "step": 4843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3175591464254581e-05,
+      "loss": 0.2863,
+      "step": 4844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317295880479439e-05,
+      "loss": 0.336,
+      "step": 4845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170325900778211e-05,
+      "loss": 0.3009,
+      "step": 4846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167692752408978e-05,
+      "loss": 0.3116,
+      "step": 4847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165059359889639e-05,
+      "loss": 0.3154,
+      "step": 4848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3162425723423162e-05,
+      "loss": 0.31,
+      "step": 4849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3159791843212542e-05,
+      "loss": 0.2816,
+      "step": 4850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157157719460774e-05,
+      "loss": 0.3444,
+      "step": 4851
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3154523352370894e-05,
+      "loss": 0.6122,
+      "step": 4852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151888742145932e-05,
+      "loss": 0.306,
+      "step": 4853
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3149253888988967e-05,
+      "loss": 0.2873,
+      "step": 4854
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146618793103074e-05,
+      "loss": 0.2795,
+      "step": 4855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3143983454691348e-05,
+      "loss": 0.3004,
+      "step": 4856
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141347873956915e-05,
+      "loss": 0.2885,
+      "step": 4857
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3138712051102908e-05,
+      "loss": 0.2515,
+      "step": 4858
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136075986332485e-05,
+      "loss": 0.3378,
+      "step": 4859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3133439679848824e-05,
+      "loss": 0.277,
+      "step": 4860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3130803131855113e-05,
+      "loss": 0.3376,
+      "step": 4861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3128166342554567e-05,
+      "loss": 0.297,
+      "step": 4862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3125529312150414e-05,
+      "loss": 0.3082,
+      "step": 4863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312289204084591e-05,
+      "loss": 0.3073,
+      "step": 4864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120254528844312e-05,
+      "loss": 0.3297,
+      "step": 4865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3117616776348915e-05,
+      "loss": 0.2859,
+      "step": 4866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3114978783563022e-05,
+      "loss": 0.3062,
+      "step": 4867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3112340550689955e-05,
+      "loss": 0.2731,
+      "step": 4868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310970207793306e-05,
+      "loss": 0.2795,
+      "step": 4869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3107063365495692e-05,
+      "loss": 0.2925,
+      "step": 4870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104424413581231e-05,
+      "loss": 0.2915,
+      "step": 4871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3101785222393075e-05,
+      "loss": 0.2886,
+      "step": 4872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3099145792134642e-05,
+      "loss": 0.3122,
+      "step": 4873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3096506123009368e-05,
+      "loss": 0.2537,
+      "step": 4874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093866215220698e-05,
+      "loss": 0.2968,
+      "step": 4875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309122606897211e-05,
+      "loss": 0.3199,
+      "step": 4876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3088585684467088e-05,
+      "loss": 0.2534,
+      "step": 4877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3085945061909144e-05,
+      "loss": 0.2666,
+      "step": 4878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3083304201501803e-05,
+      "loss": 0.2632,
+      "step": 4879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3080663103448607e-05,
+      "loss": 0.2697,
+      "step": 4880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078021767953125e-05,
+      "loss": 0.256,
+      "step": 4881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075380195218931e-05,
+      "loss": 0.2749,
+      "step": 4882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307273838544963e-05,
+      "loss": 0.2484,
+      "step": 4883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070096338848835e-05,
+      "loss": 0.316,
+      "step": 4884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3067454055620184e-05,
+      "loss": 0.2657,
+      "step": 4885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306481153596733e-05,
+      "loss": 0.2841,
+      "step": 4886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3062168780093949e-05,
+      "loss": 0.2512,
+      "step": 4887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059525788203728e-05,
+      "loss": 0.2742,
+      "step": 4888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3056882560500378e-05,
+      "loss": 0.2917,
+      "step": 4889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3054239097187625e-05,
+      "loss": 0.2507,
+      "step": 4890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305159539846921e-05,
+      "loss": 0.3232,
+      "step": 4891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048951464548902e-05,
+      "loss": 0.6477,
+      "step": 4892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046307295630482e-05,
+      "loss": 0.295,
+      "step": 4893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043662891917748e-05,
+      "loss": 0.2845,
+      "step": 4894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3041018253614518e-05,
+      "loss": 0.3442,
+      "step": 4895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3038373380924623e-05,
+      "loss": 0.3137,
+      "step": 4896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3035728274051924e-05,
+      "loss": 0.2985,
+      "step": 4897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3033082933200287e-05,
+      "loss": 0.2945,
+      "step": 4898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3030437358573606e-05,
+      "loss": 0.2918,
+      "step": 4899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302779155037579e-05,
+      "loss": 0.3008,
+      "step": 4900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302514550881076e-05,
+      "loss": 0.2714,
+      "step": 4901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022499234082463e-05,
+      "loss": 0.3484,
+      "step": 4902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3019852726394857e-05,
+      "loss": 0.3029,
+      "step": 4903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017205985951926e-05,
+      "loss": 0.2746,
+      "step": 4904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3014559012957665e-05,
+      "loss": 0.2798,
+      "step": 4905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011911807616091e-05,
+      "loss": 0.2724,
+      "step": 4906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3009264370131239e-05,
+      "loss": 0.3561,
+      "step": 4907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006616700707156e-05,
+      "loss": 0.2535,
+      "step": 4908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003968799547915e-05,
+      "loss": 0.291,
+      "step": 4909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.30013206668576e-05,
+      "loss": 0.3071,
+      "step": 4910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998672302840318e-05,
+      "loss": 0.2776,
+      "step": 4911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996023707700197e-05,
+      "loss": 0.2747,
+      "step": 4912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993374881641367e-05,
+      "loss": 0.2949,
+      "step": 4913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990725824867995e-05,
+      "loss": 0.3315,
+      "step": 4914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988076537584254e-05,
+      "loss": 0.2838,
+      "step": 4915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985427019994335e-05,
+      "loss": 0.2912,
+      "step": 4916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982777272302454e-05,
+      "loss": 0.3068,
+      "step": 4917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980127294712839e-05,
+      "loss": 0.6093,
+      "step": 4918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977477087429739e-05,
+      "loss": 0.2781,
+      "step": 4919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974826650657418e-05,
+      "loss": 0.2736,
+      "step": 4920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972175984600157e-05,
+      "loss": 0.3271,
+      "step": 4921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969525089462253e-05,
+      "loss": 0.3005,
+      "step": 4922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966873965448032e-05,
+      "loss": 0.2636,
+      "step": 4923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964222612761825e-05,
+      "loss": 0.2827,
+      "step": 4924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2961571031607985e-05,
+      "loss": 0.3504,
+      "step": 4925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2958919222190885e-05,
+      "loss": 0.3324,
+      "step": 4926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295626718471491e-05,
+      "loss": 0.3079,
+      "step": 4927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295361491938447e-05,
+      "loss": 0.2971,
+      "step": 4928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2950962426403981e-05,
+      "loss": 0.3065,
+      "step": 4929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948309705977893e-05,
+      "loss": 0.6127,
+      "step": 4930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2945656758310663e-05,
+      "loss": 0.2986,
+      "step": 4931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294300358360676e-05,
+      "loss": 0.3096,
+      "step": 4932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294035018207069e-05,
+      "loss": 0.3211,
+      "step": 4933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2937696553906949e-05,
+      "loss": 0.3181,
+      "step": 4934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935042699320075e-05,
+      "loss": 0.2803,
+      "step": 4935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2932388618514616e-05,
+      "loss": 0.2709,
+      "step": 4936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2929734311695125e-05,
+      "loss": 0.2866,
+      "step": 4937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927079779066196e-05,
+      "loss": 0.3434,
+      "step": 4938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2924425020832419e-05,
+      "loss": 0.3034,
+      "step": 4939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292177003719841e-05,
+      "loss": 0.2884,
+      "step": 4940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919114828368806e-05,
+      "loss": 0.2755,
+      "step": 4941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291645939454825e-05,
+      "loss": 0.2856,
+      "step": 4942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2913803735941417e-05,
+      "loss": 0.2612,
+      "step": 4943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291114785275299e-05,
+      "loss": 0.3066,
+      "step": 4944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908491745187673e-05,
+      "loss": 0.269,
+      "step": 4945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290583541345018e-05,
+      "loss": 0.3151,
+      "step": 4946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903178857745256e-05,
+      "loss": 0.3021,
+      "step": 4947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900522078277645e-05,
+      "loss": 0.3053,
+      "step": 4948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897865075252125e-05,
+      "loss": 0.2658,
+      "step": 4949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2895207848873488e-05,
+      "loss": 0.2978,
+      "step": 4950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2892550399346533e-05,
+      "loss": 0.2809,
+      "step": 4951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2889892726876085e-05,
+      "loss": 0.2707,
+      "step": 4952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887234831666987e-05,
+      "loss": 0.299,
+      "step": 4953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2884576713924093e-05,
+      "loss": 0.3063,
+      "step": 4954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2881918373852278e-05,
+      "loss": 0.2911,
+      "step": 4955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879259811656435e-05,
+      "loss": 0.2944,
+      "step": 4956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2876601027541475e-05,
+      "loss": 0.3289,
+      "step": 4957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287394202171232e-05,
+      "loss": 0.2614,
+      "step": 4958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2871282794373916e-05,
+      "loss": 0.2828,
+      "step": 4959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286862334573122e-05,
+      "loss": 0.3165,
+      "step": 4960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286596367598921e-05,
+      "loss": 0.238,
+      "step": 4961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2863303785352883e-05,
+      "loss": 0.2675,
+      "step": 4962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2860643674027246e-05,
+      "loss": 0.2759,
+      "step": 4963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2857983342217333e-05,
+      "loss": 0.2704,
+      "step": 4964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2855322790128182e-05,
+      "loss": 0.3199,
+      "step": 4965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852662017964863e-05,
+      "loss": 0.2951,
+      "step": 4966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2850001025932444e-05,
+      "loss": 0.2819,
+      "step": 4967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2847339814236033e-05,
+      "loss": 0.2407,
+      "step": 4968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284467838308074e-05,
+      "loss": 0.2733,
+      "step": 4969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2842016732671689e-05,
+      "loss": 0.2523,
+      "step": 4970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2839354863214035e-05,
+      "loss": 0.2656,
+      "step": 4971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836692774912935e-05,
+      "loss": 0.2934,
+      "step": 4972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2834030467973572e-05,
+      "loss": 0.3209,
+      "step": 4973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831367942601146e-05,
+      "loss": 0.2899,
+      "step": 4974
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828705199000869e-05,
+      "loss": 0.2903,
+      "step": 4975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2826042237377972e-05,
+      "loss": 0.3071,
+      "step": 4976
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2823379057937702e-05,
+      "loss": 0.3071,
+      "step": 4977
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2820715660885328e-05,
+      "loss": 0.2733,
+      "step": 4978
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818052046426125e-05,
+      "loss": 0.3013,
+      "step": 4979
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2815388214765397e-05,
+      "loss": 0.2896,
+      "step": 4980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2812724166108454e-05,
+      "loss": 0.6068,
+      "step": 4981
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281005990066063e-05,
+      "loss": 0.2501,
+      "step": 4982
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2807395418627278e-05,
+      "loss": 0.3109,
+      "step": 4983
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804730720213756e-05,
+      "loss": 0.3362,
+      "step": 4984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280206580562545e-05,
+      "loss": 0.2627,
+      "step": 4985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799400675067754e-05,
+      "loss": 0.3087,
+      "step": 4986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796735328746089e-05,
+      "loss": 0.2274,
+      "step": 4987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2794069766865884e-05,
+      "loss": 0.2737,
+      "step": 4988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2791403989632586e-05,
+      "loss": 0.3082,
+      "step": 4989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788737997251665e-05,
+      "loss": 0.2284,
+      "step": 4990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2786071789928593e-05,
+      "loss": 0.3211,
+      "step": 4991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2783405367868878e-05,
+      "loss": 0.3515,
+      "step": 4992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780738731278028e-05,
+      "loss": 0.2344,
+      "step": 4993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778071880361577e-05,
+      "loss": 0.2803,
+      "step": 4994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2775404815325074e-05,
+      "loss": 0.2524,
+      "step": 4995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277273753637408e-05,
+      "loss": 0.3319,
+      "step": 4996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277007004371418e-05,
+      "loss": 0.2653,
+      "step": 4997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767402337550966e-05,
+      "loss": 0.3116,
+      "step": 4998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2764734418090052e-05,
+      "loss": 0.3344,
+      "step": 4999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2762066285537071e-05,
+      "loss": 0.3002,
+      "step": 5000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275939794009767e-05,
+      "loss": 0.267,
+      "step": 5001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275672938197751e-05,
+      "loss": 0.2889,
+      "step": 5002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275406061138227e-05,
+      "loss": 0.2973,
+      "step": 5003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2751391628517649e-05,
+      "loss": 0.3354,
+      "step": 5004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748722433589351e-05,
+      "loss": 0.3162,
+      "step": 5005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2746053026803114e-05,
+      "loss": 0.2952,
+      "step": 5006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743383408364678e-05,
+      "loss": 0.3312,
+      "step": 5007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740713578479802e-05,
+      "loss": 0.2976,
+      "step": 5008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738043537354268e-05,
+      "loss": 0.2833,
+      "step": 5009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2735373285193867e-05,
+      "loss": 0.2736,
+      "step": 5010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2732702822204406e-05,
+      "loss": 0.2896,
+      "step": 5011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730032148591716e-05,
+      "loss": 0.6,
+      "step": 5012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727361264561637e-05,
+      "loss": 0.3231,
+      "step": 5013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2724690170320031e-05,
+      "loss": 0.2505,
+      "step": 5014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2722018866072768e-05,
+      "loss": 0.2782,
+      "step": 5015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719347352025741e-05,
+      "loss": 0.2721,
+      "step": 5016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2716675628384856e-05,
+      "loss": 0.3256,
+      "step": 5017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714003695356037e-05,
+      "loss": 0.3231,
+      "step": 5018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711331553145223e-05,
+      "loss": 0.2431,
+      "step": 5019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270865920195837e-05,
+      "loss": 0.3104,
+      "step": 5020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705986642001451e-05,
+      "loss": 0.2838,
+      "step": 5021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2703313873480451e-05,
+      "loss": 0.3663,
+      "step": 5022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2700640896601377e-05,
+      "loss": 0.3014,
+      "step": 5023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697967711570243e-05,
+      "loss": 0.3015,
+      "step": 5024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2695294318593092e-05,
+      "loss": 0.2964,
+      "step": 5025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692620717875972e-05,
+      "loss": 0.2776,
+      "step": 5026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689946909624951e-05,
+      "loss": 0.3105,
+      "step": 5027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687272894046116e-05,
+      "loss": 0.251,
+      "step": 5028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684598671345563e-05,
+      "loss": 0.2552,
+      "step": 5029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681924241729409e-05,
+      "loss": 0.2737,
+      "step": 5030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679249605403786e-05,
+      "loss": 0.2887,
+      "step": 5031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2676574762574842e-05,
+      "loss": 0.2668,
+      "step": 5032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673899713448743e-05,
+      "loss": 0.288,
+      "step": 5033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2671224458231664e-05,
+      "loss": 0.3062,
+      "step": 5034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2668548997129807e-05,
+      "loss": 0.2898,
+      "step": 5035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665873330349371e-05,
+      "loss": 0.2919,
+      "step": 5036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2663197458096597e-05,
+      "loss": 0.2697,
+      "step": 5037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266052138057772e-05,
+      "loss": 0.2736,
+      "step": 5038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657845097999e-05,
+      "loss": 0.2621,
+      "step": 5039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655168610566715e-05,
+      "loss": 0.287,
+      "step": 5040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265249191848715e-05,
+      "loss": 0.2751,
+      "step": 5041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264981502196662e-05,
+      "loss": 0.311,
+      "step": 5042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2647137921211435e-05,
+      "loss": 0.2762,
+      "step": 5043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264446061642794e-05,
+      "loss": 0.3188,
+      "step": 5044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641783107822491e-05,
+      "loss": 0.2769,
+      "step": 5045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2639105395601452e-05,
+      "loss": 0.297,
+      "step": 5046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263642747997121e-05,
+      "loss": 0.2922,
+      "step": 5047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2633749361138162e-05,
+      "loss": 0.2924,
+      "step": 5048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2631071039308728e-05,
+      "loss": 0.2505,
+      "step": 5049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2628392514689339e-05,
+      "loss": 0.317,
+      "step": 5050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2625713787486442e-05,
+      "loss": 0.2849,
+      "step": 5051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2623034857906501e-05,
+      "loss": 0.2952,
+      "step": 5052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620355726155995e-05,
+      "loss": 0.5823,
+      "step": 5053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2617676392441419e-05,
+      "loss": 0.3296,
+      "step": 5054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614996856969275e-05,
+      "loss": 0.2885,
+      "step": 5055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2612317119946099e-05,
+      "loss": 0.2552,
+      "step": 5056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609637181578424e-05,
+      "loss": 0.3136,
+      "step": 5057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260695704207281e-05,
+      "loss": 0.3046,
+      "step": 5058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2604276701635832e-05,
+      "loss": 0.2936,
+      "step": 5059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260159616047407e-05,
+      "loss": 0.325,
+      "step": 5060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598915418794136e-05,
+      "loss": 0.3215,
+      "step": 5061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2596234476802636e-05,
+      "loss": 0.2955,
+      "step": 5062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2593553334706212e-05,
+      "loss": 0.2825,
+      "step": 5063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2590871992711517e-05,
+      "loss": 0.3372,
+      "step": 5064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.2682,
+      "step": 5065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2585508709853971e-05,
+      "loss": 0.2754,
+      "step": 5066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2582826769404492e-05,
+      "loss": 0.3116,
+      "step": 5067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2580144629883494e-05,
+      "loss": 0.3187,
+      "step": 5068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257746229149769e-05,
+      "loss": 0.3624,
+      "step": 5069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2574779754453831e-05,
+      "loss": 0.3005,
+      "step": 5070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2572097018958674e-05,
+      "loss": 0.3066,
+      "step": 5071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2569414085218986e-05,
+      "loss": 0.2709,
+      "step": 5072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2566730953441554e-05,
+      "loss": 0.2873,
+      "step": 5073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2564047623833186e-05,
+      "loss": 0.5846,
+      "step": 5074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561364096600694e-05,
+      "loss": 0.2693,
+      "step": 5075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2558680371950913e-05,
+      "loss": 0.2677,
+      "step": 5076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555996450090693e-05,
+      "loss": 0.2746,
+      "step": 5077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553312331226896e-05,
+      "loss": 0.5933,
+      "step": 5078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550628015566402e-05,
+      "loss": 0.2944,
+      "step": 5079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547943503316105e-05,
+      "loss": 0.3016,
+      "step": 5080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2545258794682906e-05,
+      "loss": 0.3448,
+      "step": 5081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542573889873741e-05,
+      "loss": 0.2834,
+      "step": 5082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539888789095542e-05,
+      "loss": 0.2861,
+      "step": 5083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537203492555265e-05,
+      "loss": 0.3148,
+      "step": 5084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2534518000459884e-05,
+      "loss": 0.319,
+      "step": 5085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531832313016374e-05,
+      "loss": 0.2494,
+      "step": 5086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252914643043174e-05,
+      "loss": 0.6106,
+      "step": 5087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2526460352912994e-05,
+      "loss": 0.3207,
+      "step": 5088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252377408066717e-05,
+      "loss": 0.3076,
+      "step": 5089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521087613901313e-05,
+      "loss": 0.3112,
+      "step": 5090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518400952822475e-05,
+      "loss": 0.3275,
+      "step": 5091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251571409763774e-05,
+      "loss": 0.3105,
+      "step": 5092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251302704855419e-05,
+      "loss": 0.3038,
+      "step": 5093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510339805778932e-05,
+      "loss": 0.3319,
+      "step": 5094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507652369519085e-05,
+      "loss": 0.3226,
+      "step": 5095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2504964739981787e-05,
+      "loss": 0.2773,
+      "step": 5096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502276917374183e-05,
+      "loss": 0.2769,
+      "step": 5097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499588901903437e-05,
+      "loss": 0.2853,
+      "step": 5098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2496900693776732e-05,
+      "loss": 0.2924,
+      "step": 5099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494212293201255e-05,
+      "loss": 0.257,
+      "step": 5100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2491523700384222e-05,
+      "loss": 0.3021,
+      "step": 5101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2488834915532852e-05,
+      "loss": 0.3013,
+      "step": 5102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486145938854384e-05,
+      "loss": 0.2578,
+      "step": 5103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2483456770556073e-05,
+      "loss": 0.2798,
+      "step": 5104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480767410845185e-05,
+      "loss": 0.326,
+      "step": 5105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2478077859929e-05,
+      "loss": 0.3029,
+      "step": 5106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247538811801482e-05,
+      "loss": 0.2978,
+      "step": 5107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2472698185309954e-05,
+      "loss": 0.3493,
+      "step": 5108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2470008062021732e-05,
+      "loss": 0.3455,
+      "step": 5109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2467317748357493e-05,
+      "loss": 0.2995,
+      "step": 5110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2464627244524595e-05,
+      "loss": 0.3005,
+      "step": 5111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2461936550730402e-05,
+      "loss": 0.3138,
+      "step": 5112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2459245667182307e-05,
+      "loss": 0.3345,
+      "step": 5113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2456554594087709e-05,
+      "loss": 0.2744,
+      "step": 5114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2453863331654019e-05,
+      "loss": 0.276,
+      "step": 5115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245117188008867e-05,
+      "loss": 0.288,
+      "step": 5116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2448480239599103e-05,
+      "loss": 0.2554,
+      "step": 5117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445788410392778e-05,
+      "loss": 0.2917,
+      "step": 5118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443096392677165e-05,
+      "loss": 0.3129,
+      "step": 5119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440404186659757e-05,
+      "loss": 0.2881,
+      "step": 5120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.243771179254805e-05,
+      "loss": 0.2708,
+      "step": 5121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435019210549564e-05,
+      "loss": 0.5685,
+      "step": 5122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432326440871832e-05,
+      "loss": 0.2533,
+      "step": 5123
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429633483722392e-05,
+      "loss": 0.2368,
+      "step": 5124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242694033930881e-05,
+      "loss": 0.2086,
+      "step": 5125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2424247007838659e-05,
+      "loss": 0.2679,
+      "step": 5126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2421553489519527e-05,
+      "loss": 0.2884,
+      "step": 5127
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2418859784559016e-05,
+      "loss": 0.2932,
+      "step": 5128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2416165893164746e-05,
+      "loss": 0.3159,
+      "step": 5129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241347181554435e-05,
+      "loss": 0.2684,
+      "step": 5130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410777551905469e-05,
+      "loss": 0.3151,
+      "step": 5131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2408083102455766e-05,
+      "loss": 0.3011,
+      "step": 5132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2405388467402915e-05,
+      "loss": 0.2405,
+      "step": 5133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2402693646954607e-05,
+      "loss": 0.2706,
+      "step": 5134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2399998641318547e-05,
+      "loss": 0.3016,
+      "step": 5135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2397303450702449e-05,
+      "loss": 0.3594,
+      "step": 5136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2394608075314048e-05,
+      "loss": 0.3301,
+      "step": 5137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2391912515361085e-05,
+      "loss": 0.2711,
+      "step": 5138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238921677105133e-05,
+      "loss": 0.2408,
+      "step": 5139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386520842592544e-05,
+      "loss": 0.2906,
+      "step": 5140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383824730192529e-05,
+      "loss": 0.3082,
+      "step": 5141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2381128434059082e-05,
+      "loss": 0.2953,
+      "step": 5142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2378431954400021e-05,
+      "loss": 0.3179,
+      "step": 5143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2375735291423174e-05,
+      "loss": 0.3124,
+      "step": 5144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2373038445336392e-05,
+      "loss": 0.256,
+      "step": 5145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237034141634753e-05,
+      "loss": 0.2609,
+      "step": 5146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367644204664468e-05,
+      "loss": 0.2864,
+      "step": 5147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364946810495088e-05,
+      "loss": 0.2834,
+      "step": 5148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362249234047295e-05,
+      "loss": 0.2791,
+      "step": 5149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2359551475529e-05,
+      "loss": 0.2488,
+      "step": 5150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356853535148137e-05,
+      "loss": 0.2633,
+      "step": 5151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2354155413112647e-05,
+      "loss": 0.2738,
+      "step": 5152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351457109630493e-05,
+      "loss": 0.2678,
+      "step": 5153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348758624909644e-05,
+      "loss": 0.3131,
+      "step": 5154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346059959158085e-05,
+      "loss": 0.2427,
+      "step": 5155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2343361112583819e-05,
+      "loss": 0.297,
+      "step": 5156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234066208539485e-05,
+      "loss": 0.2605,
+      "step": 5157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233796287779922e-05,
+      "loss": 0.265,
+      "step": 5158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335263490004961e-05,
+      "loss": 0.5809,
+      "step": 5159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2332563922220132e-05,
+      "loss": 0.3103,
+      "step": 5160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329864174652802e-05,
+      "loss": 0.3275,
+      "step": 5161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327164247511051e-05,
+      "loss": 0.2773,
+      "step": 5162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2324464141002981e-05,
+      "loss": 0.3287,
+      "step": 5163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23217638553367e-05,
+      "loss": 0.2722,
+      "step": 5164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2319063390720331e-05,
+      "loss": 0.2862,
+      "step": 5165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2316362747362019e-05,
+      "loss": 0.2706,
+      "step": 5166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2313661925469908e-05,
+      "loss": 0.27,
+      "step": 5167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310960925252171e-05,
+      "loss": 0.2825,
+      "step": 5168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2308259746916982e-05,
+      "loss": 0.2767,
+      "step": 5169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305558390672539e-05,
+      "loss": 0.27,
+      "step": 5170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302856856727043e-05,
+      "loss": 0.2974,
+      "step": 5171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300155145288724e-05,
+      "loss": 0.2477,
+      "step": 5172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2297453256565812e-05,
+      "loss": 0.2212,
+      "step": 5173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2294751190766552e-05,
+      "loss": 0.2523,
+      "step": 5174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292048948099214e-05,
+      "loss": 0.2648,
+      "step": 5175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228934652877206e-05,
+      "loss": 0.2862,
+      "step": 5176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286643932993396e-05,
+      "loss": 0.2982,
+      "step": 5177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283941160971512e-05,
+      "loss": 0.2551,
+      "step": 5178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2281238212914727e-05,
+      "loss": 0.2878,
+      "step": 5179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2278535089031377e-05,
+      "loss": 0.2994,
+      "step": 5180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.22758317895298e-05,
+      "loss": 0.2975,
+      "step": 5181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2273128314618353e-05,
+      "loss": 0.2916,
+      "step": 5182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2270424664505405e-05,
+      "loss": 0.2406,
+      "step": 5183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2267720839399347e-05,
+      "loss": 0.2346,
+      "step": 5184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2265016839508568e-05,
+      "loss": 0.3267,
+      "step": 5185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262312665041482e-05,
+      "loss": 0.3231,
+      "step": 5186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2259608316206519e-05,
+      "loss": 0.2839,
+      "step": 5187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2256903793212107e-05,
+      "loss": 0.2876,
+      "step": 5188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2254199096266705e-05,
+      "loss": 0.2913,
+      "step": 5189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2251494225578775e-05,
+      "loss": 0.29,
+      "step": 5190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2248789181356793e-05,
+      "loss": 0.2804,
+      "step": 5191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2246083963809256e-05,
+      "loss": 0.339,
+      "step": 5192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243378573144663e-05,
+      "loss": 0.3062,
+      "step": 5193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2240673009571536e-05,
+      "loss": 0.2581,
+      "step": 5194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2237967273298401e-05,
+      "loss": 0.2989,
+      "step": 5195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223526136453381e-05,
+      "loss": 0.282,
+      "step": 5196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2232555283486319e-05,
+      "loss": 0.2779,
+      "step": 5197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2229849030364496e-05,
+      "loss": 0.3223,
+      "step": 5198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227142605376928e-05,
+      "loss": 0.3434,
+      "step": 5199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2224436008732213e-05,
+      "loss": 0.2767,
+      "step": 5200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222172924063896e-05,
+      "loss": 0.3041,
+      "step": 5201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2219022301305796e-05,
+      "loss": 0.6394,
+      "step": 5202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221631519094136e-05,
+      "loss": 0.2842,
+      "step": 5203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2213607909754297e-05,
+      "loss": 0.2719,
+      "step": 5204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210900457953274e-05,
+      "loss": 0.2858,
+      "step": 5205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208192835746973e-05,
+      "loss": 0.3089,
+      "step": 5206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2205485043344074e-05,
+      "loss": 0.2872,
+      "step": 5207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2202777080953285e-05,
+      "loss": 0.3708,
+      "step": 5208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200068948783325e-05,
+      "loss": 0.2668,
+      "step": 5209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2197360647042922e-05,
+      "loss": 0.2662,
+      "step": 5210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194652175940817e-05,
+      "loss": 0.264,
+      "step": 5211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2191943535685766e-05,
+      "loss": 0.3219,
+      "step": 5212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218923472648654e-05,
+      "loss": 0.3112,
+      "step": 5213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186525748551914e-05,
+      "loss": 0.2975,
+      "step": 5214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183816602090693e-05,
+      "loss": 0.3105,
+      "step": 5215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181107287311675e-05,
+      "loss": 0.2872,
+      "step": 5216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2178397804423685e-05,
+      "loss": 0.284,
+      "step": 5217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217568815363556e-05,
+      "loss": 0.2841,
+      "step": 5218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2172978335156136e-05,
+      "loss": 0.2945,
+      "step": 5219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170268349194281e-05,
+      "loss": 0.3187,
+      "step": 5220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167558195958867e-05,
+      "loss": 0.2859,
+      "step": 5221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2164847875658776e-05,
+      "loss": 0.3312,
+      "step": 5222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2162137388502908e-05,
+      "loss": 0.2891,
+      "step": 5223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159426734700175e-05,
+      "loss": 0.2958,
+      "step": 5224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156715914459498e-05,
+      "loss": 0.2496,
+      "step": 5225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2154004927989815e-05,
+      "loss": 0.2876,
+      "step": 5226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2151293775500076e-05,
+      "loss": 0.337,
+      "step": 5227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214858245719924e-05,
+      "loss": 0.2776,
+      "step": 5228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145870973296288e-05,
+      "loss": 0.2784,
+      "step": 5229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143159324000204e-05,
+      "loss": 0.2466,
+      "step": 5230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2140447509519988e-05,
+      "loss": 0.272,
+      "step": 5231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2137735530064653e-05,
+      "loss": 0.3425,
+      "step": 5232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2135023385843228e-05,
+      "loss": 0.3345,
+      "step": 5233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2132311077064749e-05,
+      "loss": 0.2472,
+      "step": 5234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2129598603938269e-05,
+      "loss": 0.2385,
+      "step": 5235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212688596667285e-05,
+      "loss": 0.2809,
+      "step": 5236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2124173165477572e-05,
+      "loss": 0.3051,
+      "step": 5237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2121460200561521e-05,
+      "loss": 0.3031,
+      "step": 5238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21187470721338e-05,
+      "loss": 0.3035,
+      "step": 5239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2116033780403524e-05,
+      "loss": 0.2922,
+      "step": 5240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211332032557982e-05,
+      "loss": 0.311,
+      "step": 5241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110606707871828e-05,
+      "loss": 0.2844,
+      "step": 5242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107892927488698e-05,
+      "loss": 0.2643,
+      "step": 5243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2105178984639601e-05,
+      "loss": 0.2689,
+      "step": 5244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102464879533704e-05,
+      "loss": 0.3397,
+      "step": 5245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099750612380205e-05,
+      "loss": 0.3127,
+      "step": 5246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2097036183388305e-05,
+      "loss": 0.3026,
+      "step": 5247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2094321592767217e-05,
+      "loss": 0.3033,
+      "step": 5248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209160684072617e-05,
+      "loss": 0.2864,
+      "step": 5249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20888919274744e-05,
+      "loss": 0.2635,
+      "step": 5250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086176853221166e-05,
+      "loss": 0.6339,
+      "step": 5251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083461618175723e-05,
+      "loss": 0.2958,
+      "step": 5252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2080746222547356e-05,
+      "loss": 0.5901,
+      "step": 5253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2078030666545351e-05,
+      "loss": 0.2417,
+      "step": 5254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207531495037901e-05,
+      "loss": 0.281,
+      "step": 5255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207259907425765e-05,
+      "loss": 0.276,
+      "step": 5256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206988303839059e-05,
+      "loss": 0.2995,
+      "step": 5257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067166842987175e-05,
+      "loss": 0.2656,
+      "step": 5258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064450488256751e-05,
+      "loss": 0.2841,
+      "step": 5259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206173397440869e-05,
+      "loss": 0.3264,
+      "step": 5260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059017301652359e-05,
+      "loss": 0.3846,
+      "step": 5261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2056300470197144e-05,
+      "loss": 0.2714,
+      "step": 5262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2053583480252456e-05,
+      "loss": 0.3253,
+      "step": 5263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050866332027695e-05,
+      "loss": 0.2585,
+      "step": 5264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048149025732295e-05,
+      "loss": 0.2858,
+      "step": 5265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045431561575685e-05,
+      "loss": 0.3005,
+      "step": 5266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042713939767318e-05,
+      "loss": 0.2812,
+      "step": 5267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2039996160516654e-05,
+      "loss": 0.2631,
+      "step": 5268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037278224033166e-05,
+      "loss": 0.2798,
+      "step": 5269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2034560130526341e-05,
+      "loss": 0.239,
+      "step": 5270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203184188020567e-05,
+      "loss": 0.3223,
+      "step": 5271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029123473280668e-05,
+      "loss": 0.2745,
+      "step": 5272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026404909960856e-05,
+      "loss": 0.2731,
+      "step": 5273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023686190455766e-05,
+      "loss": 0.2752,
+      "step": 5274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020967314974945e-05,
+      "loss": 0.2837,
+      "step": 5275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018248283727947e-05,
+      "loss": 0.2931,
+      "step": 5276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015529096924343e-05,
+      "loss": 0.2755,
+      "step": 5277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012809754773718e-05,
+      "loss": 0.2768,
+      "step": 5278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010090257485663e-05,
+      "loss": 0.3058,
+      "step": 5279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007370605269782e-05,
+      "loss": 0.2674,
+      "step": 5280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200465079833569e-05,
+      "loss": 0.2902,
+      "step": 5281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001930836893026e-05,
+      "loss": 0.3199,
+      "step": 5282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1999210721151421e-05,
+      "loss": 0.3159,
+      "step": 5283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1996490451320532e-05,
+      "loss": 0.3105,
+      "step": 5284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1993770027610023e-05,
+      "loss": 0.61,
+      "step": 5285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991049450229577e-05,
+      "loss": 0.3182,
+      "step": 5286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1988328719388873e-05,
+      "loss": 0.274,
+      "step": 5287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1985607835297618e-05,
+      "loss": 0.3,
+      "step": 5288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982886798165521e-05,
+      "loss": 0.288,
+      "step": 5289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1980165608202303e-05,
+      "loss": 0.2626,
+      "step": 5290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977444265617713e-05,
+      "loss": 0.3077,
+      "step": 5291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1974722770621483e-05,
+      "loss": 0.26,
+      "step": 5292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1972001123423382e-05,
+      "loss": 0.2897,
+      "step": 5293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969279324233179e-05,
+      "loss": 0.2702,
+      "step": 5294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966557373260654e-05,
+      "loss": 0.2867,
+      "step": 5295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1963835270715604e-05,
+      "loss": 0.2562,
+      "step": 5296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1961113016807838e-05,
+      "loss": 0.2557,
+      "step": 5297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1958390611747167e-05,
+      "loss": 0.2822,
+      "step": 5298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1955668055743429e-05,
+      "loss": 0.2919,
+      "step": 5299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952945349006455e-05,
+      "loss": 0.2864,
+      "step": 5300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1950222491746109e-05,
+      "loss": 0.3005,
+      "step": 5301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1947499484172245e-05,
+      "loss": 0.2961,
+      "step": 5302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1944776326494745e-05,
+      "loss": 0.3578,
+      "step": 5303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942053018923494e-05,
+      "loss": 0.2657,
+      "step": 5304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1939329561668396e-05,
+      "loss": 0.3127,
+      "step": 5305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1936605954939355e-05,
+      "loss": 0.2883,
+      "step": 5306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1933882198946296e-05,
+      "loss": 0.3423,
+      "step": 5307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1931158293899154e-05,
+      "loss": 0.2751,
+      "step": 5308
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1928434240007869e-05,
+      "loss": 0.3028,
+      "step": 5309
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1925710037482405e-05,
+      "loss": 0.3023,
+      "step": 5310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1922985686532726e-05,
+      "loss": 0.6001,
+      "step": 5311
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192026118736881e-05,
+      "loss": 0.28,
+      "step": 5312
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917536540200655e-05,
+      "loss": 0.3248,
+      "step": 5313
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914811745238256e-05,
+      "loss": 0.2647,
+      "step": 5314
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1912086802691627e-05,
+      "loss": 0.2952,
+      "step": 5315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909361712770796e-05,
+      "loss": 0.268,
+      "step": 5316
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.19066364756858e-05,
+      "loss": 0.2814,
+      "step": 5317
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1903911091646684e-05,
+      "loss": 0.2876,
+      "step": 5318
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190118556086351e-05,
+      "loss": 0.3043,
+      "step": 5319
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1898459883546346e-05,
+      "loss": 0.3002,
+      "step": 5320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1895734059905275e-05,
+      "loss": 0.2681,
+      "step": 5321
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1893008090150389e-05,
+      "loss": 0.2776,
+      "step": 5322
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1890281974491794e-05,
+      "loss": 0.2712,
+      "step": 5323
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887555713139605e-05,
+      "loss": 0.3186,
+      "step": 5324
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884829306303947e-05,
+      "loss": 0.2463,
+      "step": 5325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188210275419496e-05,
+      "loss": 0.2753,
+      "step": 5326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1879376057022793e-05,
+      "loss": 0.2542,
+      "step": 5327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1876649214997602e-05,
+      "loss": 0.2787,
+      "step": 5328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187392222832957e-05,
+      "loss": 0.2897,
+      "step": 5329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1871195097228864e-05,
+      "loss": 0.2876,
+      "step": 5330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186846782190569e-05,
+      "loss": 0.2892,
+      "step": 5331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186574040257025e-05,
+      "loss": 0.3063,
+      "step": 5332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863012839432755e-05,
+      "loss": 0.2781,
+      "step": 5333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1860285132703435e-05,
+      "loss": 0.256,
+      "step": 5334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857557282592531e-05,
+      "loss": 0.2599,
+      "step": 5335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1854829289310293e-05,
+      "loss": 0.2654,
+      "step": 5336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852101153066976e-05,
+      "loss": 0.3451,
+      "step": 5337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849372874072852e-05,
+      "loss": 0.6012,
+      "step": 5338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846644452538207e-05,
+      "loss": 0.289,
+      "step": 5339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184391588867333e-05,
+      "loss": 0.2855,
+      "step": 5340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1841187182688527e-05,
+      "loss": 0.2746,
+      "step": 5341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838458334794116e-05,
+      "loss": 0.291,
+      "step": 5342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835729345200422e-05,
+      "loss": 0.3144,
+      "step": 5343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1833000214117776e-05,
+      "loss": 0.3074,
+      "step": 5344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830270941756532e-05,
+      "loss": 0.2659,
+      "step": 5345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827541528327052e-05,
+      "loss": 0.2766,
+      "step": 5346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1824811974039694e-05,
+      "loss": 0.2963,
+      "step": 5347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182208227910485e-05,
+      "loss": 0.2395,
+      "step": 5348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1819352443732908e-05,
+      "loss": 0.2582,
+      "step": 5349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181662246813427e-05,
+      "loss": 0.2554,
+      "step": 5350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813892352519343e-05,
+      "loss": 0.2722,
+      "step": 5351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811162097098559e-05,
+      "loss": 0.2477,
+      "step": 5352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180843170208235e-05,
+      "loss": 0.2471,
+      "step": 5353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180570116768116e-05,
+      "loss": 0.2558,
+      "step": 5354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180297049410545e-05,
+      "loss": 0.3052,
+      "step": 5355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800239681565679e-05,
+      "loss": 0.2889,
+      "step": 5356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797508730272329e-05,
+      "loss": 0.2632,
+      "step": 5357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179477764043589e-05,
+      "loss": 0.2872,
+      "step": 5358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1792046412266857e-05,
+      "loss": 0.2975,
+      "step": 5359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789315045975742e-05,
+      "loss": 0.2983,
+      "step": 5360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786583541773064e-05,
+      "loss": 0.2526,
+      "step": 5361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783851899869357e-05,
+      "loss": 0.3323,
+      "step": 5362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1781120120475156e-05,
+      "loss": 0.2587,
+      "step": 5363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778388203801019e-05,
+      "loss": 0.3342,
+      "step": 5364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775656150057507e-05,
+      "loss": 0.3036,
+      "step": 5365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1772923959455188e-05,
+      "loss": 0.266,
+      "step": 5366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770191632204656e-05,
+      "loss": 0.2959,
+      "step": 5367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767459168516497e-05,
+      "loss": 0.2777,
+      "step": 5368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176472656860132e-05,
+      "loss": 0.338,
+      "step": 5369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176199383266974e-05,
+      "loss": 0.3168,
+      "step": 5370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175926096093238e-05,
+      "loss": 0.2654,
+      "step": 5371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1756527953599877e-05,
+      "loss": 0.5797,
+      "step": 5372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175379481088288e-05,
+      "loss": 0.2917,
+      "step": 5373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1751061532992045e-05,
+      "loss": 0.6058,
+      "step": 5374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1748328120138038e-05,
+      "loss": 0.3195,
+      "step": 5375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745594572531538e-05,
+      "loss": 0.3198,
+      "step": 5376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1742860890383234e-05,
+      "loss": 0.2981,
+      "step": 5377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1740127073903826e-05,
+      "loss": 0.3001,
+      "step": 5378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737393123304019e-05,
+      "loss": 0.343,
+      "step": 5379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1734659038794535e-05,
+      "loss": 0.2911,
+      "step": 5380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1731924820586106e-05,
+      "loss": 0.2665,
+      "step": 5381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729190468889466e-05,
+      "loss": 0.3011,
+      "step": 5382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172645598391537e-05,
+      "loss": 0.3096,
+      "step": 5383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172372136587458e-05,
+      "loss": 0.3055,
+      "step": 5384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720986614977861e-05,
+      "loss": 0.28,
+      "step": 5385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1718251731436001e-05,
+      "loss": 0.2758,
+      "step": 5386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715516715459784e-05,
+      "loss": 0.2798,
+      "step": 5387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1712781567260018e-05,
+      "loss": 0.2996,
+      "step": 5388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171004628704751e-05,
+      "loss": 0.2979,
+      "step": 5389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707310875033085e-05,
+      "loss": 0.2941,
+      "step": 5390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170457533142757e-05,
+      "loss": 0.2889,
+      "step": 5391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1701839656441813e-05,
+      "loss": 0.2754,
+      "step": 5392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1699103850286668e-05,
+      "loss": 0.274,
+      "step": 5393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169636791317299e-05,
+      "loss": 0.3447,
+      "step": 5394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1693631845311657e-05,
+      "loss": 0.2928,
+      "step": 5395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1690895646913551e-05,
+      "loss": 0.293,
+      "step": 5396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168815931818956e-05,
+      "loss": 0.2974,
+      "step": 5397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685422859350592e-05,
+      "loss": 0.3529,
+      "step": 5398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1682686270607558e-05,
+      "loss": 0.2847,
+      "step": 5399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1679949552171382e-05,
+      "loss": 0.3314,
+      "step": 5400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1677212704252994e-05,
+      "loss": 0.285,
+      "step": 5401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1674475727063337e-05,
+      "loss": 0.3358,
+      "step": 5402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1671738620813367e-05,
+      "loss": 0.325,
+      "step": 5403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669001385714041e-05,
+      "loss": 0.2756,
+      "step": 5404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166626402197634e-05,
+      "loss": 0.3076,
+      "step": 5405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663526529811235e-05,
+      "loss": 0.2558,
+      "step": 5406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166078890942973e-05,
+      "loss": 0.2808,
+      "step": 5407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165805116104282e-05,
+      "loss": 0.2755,
+      "step": 5408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655313284861521e-05,
+      "loss": 0.3037,
+      "step": 5409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165257528109685e-05,
+      "loss": 0.298,
+      "step": 5410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1649837149959844e-05,
+      "loss": 0.2918,
+      "step": 5411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1647098891661543e-05,
+      "loss": 0.3099,
+      "step": 5412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1644360506412997e-05,
+      "loss": 0.2764,
+      "step": 5413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164162199442527e-05,
+      "loss": 0.2953,
+      "step": 5414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1638883355909429e-05,
+      "loss": 0.2708,
+      "step": 5415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1636144591076557e-05,
+      "loss": 0.288,
+      "step": 5416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1633405700137744e-05,
+      "loss": 0.2725,
+      "step": 5417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163066668330409e-05,
+      "loss": 0.2901,
+      "step": 5418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162792754078671e-05,
+      "loss": 0.5836,
+      "step": 5419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1625188272796714e-05,
+      "loss": 0.289,
+      "step": 5420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1622448879545238e-05,
+      "loss": 0.2429,
+      "step": 5421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161970936124342e-05,
+      "loss": 0.2787,
+      "step": 5422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616969718102404e-05,
+      "loss": 0.2667,
+      "step": 5423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1614229950333358e-05,
+      "loss": 0.2722,
+      "step": 5424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.161149005814744e-05,
+      "loss": 0.2934,
+      "step": 5425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608750041755832e-05,
+      "loss": 0.249,
+      "step": 5426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1606009901369718e-05,
+      "loss": 0.2656,
+      "step": 5427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1603269637200296e-05,
+      "loss": 0.6384,
+      "step": 5428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1600529249458773e-05,
+      "loss": 0.3127,
+      "step": 5429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597788738356365e-05,
+      "loss": 0.2834,
+      "step": 5430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1595048104104296e-05,
+      "loss": 0.3002,
+      "step": 5431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15923073469138e-05,
+      "loss": 0.2895,
+      "step": 5432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1589566466996124e-05,
+      "loss": 0.6115,
+      "step": 5433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586825464562515e-05,
+      "loss": 0.311,
+      "step": 5434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584084339824242e-05,
+      "loss": 0.2893,
+      "step": 5435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581343092992574e-05,
+      "loss": 0.3311,
+      "step": 5436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1578601724278794e-05,
+      "loss": 0.291,
+      "step": 5437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1575860233894195e-05,
+      "loss": 0.2982,
+      "step": 5438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573118622050075e-05,
+      "loss": 0.3025,
+      "step": 5439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1570376888957742e-05,
+      "loss": 0.3111,
+      "step": 5440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1567635034828521e-05,
+      "loss": 0.2822,
+      "step": 5441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1564893059873734e-05,
+      "loss": 0.2585,
+      "step": 5442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562150964304727e-05,
+      "loss": 0.2985,
+      "step": 5443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1559408748332841e-05,
+      "loss": 0.3049,
+      "step": 5444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1556666412169435e-05,
+      "loss": 0.3305,
+      "step": 5445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1553923956025871e-05,
+      "loss": 0.2461,
+      "step": 5446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1551181380113528e-05,
+      "loss": 0.287,
+      "step": 5447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1548438684643789e-05,
+      "loss": 0.282,
+      "step": 5448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545695869828044e-05,
+      "loss": 0.3109,
+      "step": 5449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1542952935877703e-05,
+      "loss": 0.2466,
+      "step": 5450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1540209883004171e-05,
+      "loss": 0.2564,
+      "step": 5451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537466711418874e-05,
+      "loss": 0.2736,
+      "step": 5452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1534723421333239e-05,
+      "loss": 0.2795,
+      "step": 5453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1531980012958706e-05,
+      "loss": 0.2607,
+      "step": 5454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152923648650672e-05,
+      "loss": 0.2598,
+      "step": 5455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1526492842188746e-05,
+      "loss": 0.2688,
+      "step": 5456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523749080216246e-05,
+      "loss": 0.2988,
+      "step": 5457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1521005200800694e-05,
+      "loss": 0.2971,
+      "step": 5458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1518261204153578e-05,
+      "loss": 0.2606,
+      "step": 5459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151551709048639e-05,
+      "loss": 0.2984,
+      "step": 5460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1512772860010633e-05,
+      "loss": 0.3318,
+      "step": 5461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510028512937818e-05,
+      "loss": 0.3378,
+      "step": 5462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1507284049479467e-05,
+      "loss": 0.2794,
+      "step": 5463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150453946984711e-05,
+      "loss": 0.3212,
+      "step": 5464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501794774252284e-05,
+      "loss": 0.2617,
+      "step": 5465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149904996290654e-05,
+      "loss": 0.6039,
+      "step": 5466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1496305036021427e-05,
+      "loss": 0.2834,
+      "step": 5467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1493559993808518e-05,
+      "loss": 0.3163,
+      "step": 5468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490814836479384e-05,
+      "loss": 0.2886,
+      "step": 5469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148806956424561e-05,
+      "loss": 0.2926,
+      "step": 5470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148532417731879e-05,
+      "loss": 0.3063,
+      "step": 5471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1482578675910514e-05,
+      "loss": 0.2313,
+      "step": 5472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479833060232401e-05,
+      "loss": 0.2841,
+      "step": 5473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477087330496071e-05,
+      "loss": 0.2817,
+      "step": 5474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1474341486913146e-05,
+      "loss": 0.2673,
+      "step": 5475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471595529695266e-05,
+      "loss": 0.2932,
+      "step": 5476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1468849459054073e-05,
+      "loss": 0.2584,
+      "step": 5477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146610327520122e-05,
+      "loss": 0.2826,
+      "step": 5478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1463356978348373e-05,
+      "loss": 0.3217,
+      "step": 5479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14606105687072e-05,
+      "loss": 0.2513,
+      "step": 5480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145786404648938e-05,
+      "loss": 0.3086,
+      "step": 5481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455117411906604e-05,
+      "loss": 0.3024,
+      "step": 5482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145237066517057e-05,
+      "loss": 0.3093,
+      "step": 5483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449623806492977e-05,
+      "loss": 0.296,
+      "step": 5484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1446876836085546e-05,
+      "loss": 0.2818,
+      "step": 5485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1444129754159998e-05,
+      "loss": 0.2956,
+      "step": 5486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441382560928063e-05,
+      "loss": 0.2988,
+      "step": 5487
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1438635256601484e-05,
+      "loss": 0.3042,
+      "step": 5488
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1435887841392009e-05,
+      "loss": 0.3234,
+      "step": 5489
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1433140315511392e-05,
+      "loss": 0.2617,
+      "step": 5490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14303926791714e-05,
+      "loss": 0.3085,
+      "step": 5491
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142764493258381e-05,
+      "loss": 0.2785,
+      "step": 5492
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1424897075960402e-05,
+      "loss": 0.2516,
+      "step": 5493
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422149109512967e-05,
+      "loss": 0.3103,
+      "step": 5494
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1419401033453308e-05,
+      "loss": 0.2817,
+      "step": 5495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1416652847993231e-05,
+      "loss": 0.2701,
+      "step": 5496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413904553344551e-05,
+      "loss": 0.2699,
+      "step": 5497
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411156149719094e-05,
+      "loss": 0.6191,
+      "step": 5498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408407637328694e-05,
+      "loss": 0.2776,
+      "step": 5499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405659016385191e-05,
+      "loss": 0.2715,
+      "step": 5500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402910287100436e-05,
+      "loss": 0.2797,
+      "step": 5501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1400161449686293e-05,
+      "loss": 0.2726,
+      "step": 5502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1397412504354621e-05,
+      "loss": 0.3041,
+      "step": 5503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394663451317296e-05,
+      "loss": 0.3093,
+      "step": 5504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391914290786206e-05,
+      "loss": 0.2958,
+      "step": 5505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1389165022973238e-05,
+      "loss": 0.2703,
+      "step": 5506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386415648090296e-05,
+      "loss": 0.2881,
+      "step": 5507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383666166349286e-05,
+      "loss": 0.3035,
+      "step": 5508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1380916577962127e-05,
+      "loss": 0.3101,
+      "step": 5509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1378166883140738e-05,
+      "loss": 0.3297,
+      "step": 5510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1375417082097055e-05,
+      "loss": 0.2722,
+      "step": 5511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1372667175043021e-05,
+      "loss": 0.3062,
+      "step": 5512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1369917162190585e-05,
+      "loss": 0.2759,
+      "step": 5513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367167043751701e-05,
+      "loss": 0.2617,
+      "step": 5514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1364416819938338e-05,
+      "loss": 0.2924,
+      "step": 5515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361666490962468e-05,
+      "loss": 0.2826,
+      "step": 5516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1358916057036074e-05,
+      "loss": 0.2865,
+      "step": 5517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1356165518371142e-05,
+      "loss": 0.2792,
+      "step": 5518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1353414875179673e-05,
+      "loss": 0.2476,
+      "step": 5519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1350664127673675e-05,
+      "loss": 0.3047,
+      "step": 5520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1347913276065162e-05,
+      "loss": 0.2776,
+      "step": 5521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134516232056615e-05,
+      "loss": 0.261,
+      "step": 5522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1342411261388672e-05,
+      "loss": 0.3355,
+      "step": 5523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133966009874477e-05,
+      "loss": 0.2893,
+      "step": 5524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1336908832846485e-05,
+      "loss": 0.2431,
+      "step": 5525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1334157463905876e-05,
+      "loss": 0.3019,
+      "step": 5526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1331405992134999e-05,
+      "loss": 0.2971,
+      "step": 5527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1328654417745931e-05,
+      "loss": 0.2949,
+      "step": 5528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132590274095074e-05,
+      "loss": 0.5835,
+      "step": 5529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132315096196152e-05,
+      "loss": 0.27,
+      "step": 5530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320399080990361e-05,
+      "loss": 0.3137,
+      "step": 5531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1317647098249364e-05,
+      "loss": 0.2603,
+      "step": 5532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1314895013950645e-05,
+      "loss": 0.2945,
+      "step": 5533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1312142828306309e-05,
+      "loss": 0.2939,
+      "step": 5534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1309390541528492e-05,
+      "loss": 0.3442,
+      "step": 5535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130663815382932e-05,
+      "loss": 0.3185,
+      "step": 5536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303885665420932e-05,
+      "loss": 0.2907,
+      "step": 5537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301133076515482e-05,
+      "loss": 0.2894,
+      "step": 5538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1298380387325124e-05,
+      "loss": 0.257,
+      "step": 5539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1295627598062025e-05,
+      "loss": 0.2973,
+      "step": 5540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1292874708938346e-05,
+      "loss": 0.2645,
+      "step": 5541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290121720166277e-05,
+      "loss": 0.2813,
+      "step": 5542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287368631957998e-05,
+      "loss": 0.2355,
+      "step": 5543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1284615444525706e-05,
+      "loss": 0.2487,
+      "step": 5544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281862158081605e-05,
+      "loss": 0.2957,
+      "step": 5545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1279108772837901e-05,
+      "loss": 0.3004,
+      "step": 5546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1276355289006818e-05,
+      "loss": 0.3286,
+      "step": 5547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127360170680057e-05,
+      "loss": 0.2992,
+      "step": 5548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1270848026431396e-05,
+      "loss": 0.5975,
+      "step": 5549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268094248111536e-05,
+      "loss": 0.2502,
+      "step": 5550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1265340372053237e-05,
+      "loss": 0.3278,
+      "step": 5551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1262586398468759e-05,
+      "loss": 0.3055,
+      "step": 5552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1259832327570354e-05,
+      "loss": 0.2346,
+      "step": 5553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257078159570303e-05,
+      "loss": 0.3109,
+      "step": 5554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254323894680876e-05,
+      "loss": 0.2964,
+      "step": 5555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125156953311436e-05,
+      "loss": 0.2648,
+      "step": 5556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248815075083051e-05,
+      "loss": 0.269,
+      "step": 5557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1246060520799244e-05,
+      "loss": 0.2689,
+      "step": 5558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1243305870475255e-05,
+      "loss": 0.305,
+      "step": 5559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1240551124323386e-05,
+      "loss": 0.2968,
+      "step": 5560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237796282555968e-05,
+      "loss": 0.2881,
+      "step": 5561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1235041345385328e-05,
+      "loss": 0.2873,
+      "step": 5562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1232286313023807e-05,
+      "loss": 0.3195,
+      "step": 5563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229531185683743e-05,
+      "loss": 0.3179,
+      "step": 5564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1226775963577492e-05,
+      "loss": 0.2891,
+      "step": 5565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1224020646917413e-05,
+      "loss": 0.3074,
+      "step": 5566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221265235915865e-05,
+      "loss": 0.2572,
+      "step": 5567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1218509730785231e-05,
+      "loss": 0.2892,
+      "step": 5568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215754131737889e-05,
+      "loss": 0.303,
+      "step": 5569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212998438986223e-05,
+      "loss": 0.2982,
+      "step": 5570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1210242652742632e-05,
+      "loss": 0.2475,
+      "step": 5571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207486773219515e-05,
+      "loss": 0.2832,
+      "step": 5572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1204730800629289e-05,
+      "loss": 0.2927,
+      "step": 5573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201974735184362e-05,
+      "loss": 0.2658,
+      "step": 5574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1199218577097163e-05,
+      "loss": 0.3198,
+      "step": 5575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119646232658012e-05,
+      "loss": 0.2575,
+      "step": 5576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193705983845673e-05,
+      "loss": 0.3147,
+      "step": 5577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119094954910627e-05,
+      "loss": 0.3093,
+      "step": 5578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188193022574356e-05,
+      "loss": 0.2626,
+      "step": 5579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1185436404462398e-05,
+      "loss": 0.3241,
+      "step": 5580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182679694982857e-05,
+      "loss": 0.2479,
+      "step": 5581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179922894348207e-05,
+      "loss": 0.2788,
+      "step": 5582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1177166002770937e-05,
+      "loss": 0.3054,
+      "step": 5583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1174409020463524e-05,
+      "loss": 0.5701,
+      "step": 5584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1171651947638468e-05,
+      "loss": 0.2875,
+      "step": 5585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1168894784508268e-05,
+      "loss": 0.2573,
+      "step": 5586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1166137531285435e-05,
+      "loss": 0.2822,
+      "step": 5587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163380188182482e-05,
+      "loss": 0.2219,
+      "step": 5588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160622755411932e-05,
+      "loss": 0.2692,
+      "step": 5589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157865233186315e-05,
+      "loss": 0.3212,
+      "step": 5590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1155107621718168e-05,
+      "loss": 0.304,
+      "step": 5591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152349921220036e-05,
+      "loss": 0.302,
+      "step": 5592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114959213190446e-05,
+      "loss": 0.2794,
+      "step": 5593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1146834253984008e-05,
+      "loss": 0.2825,
+      "step": 5594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1144076287671232e-05,
+      "loss": 0.2881,
+      "step": 5595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1141318233178713e-05,
+      "loss": 0.6,
+      "step": 5596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138560090719025e-05,
+      "loss": 0.3169,
+      "step": 5597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113580186050475e-05,
+      "loss": 0.3161,
+      "step": 5598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133043542748481e-05,
+      "loss": 0.3279,
+      "step": 5599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1130285137662811e-05,
+      "loss": 0.2771,
+      "step": 5600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127526645460347e-05,
+      "loss": 0.3382,
+      "step": 5601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124768066353705e-05,
+      "loss": 0.3408,
+      "step": 5602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122009400555495e-05,
+      "loss": 0.3117,
+      "step": 5603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119250648278345e-05,
+      "loss": 0.2899,
+      "step": 5604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1116491809734886e-05,
+      "loss": 0.3174,
+      "step": 5605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113732885137755e-05,
+      "loss": 0.2377,
+      "step": 5606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1110973874699595e-05,
+      "loss": 0.2486,
+      "step": 5607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110821477863306e-05,
+      "loss": 0.2451,
+      "step": 5608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1105455597150805e-05,
+      "loss": 0.3058,
+      "step": 5609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102696330465495e-05,
+      "loss": 0.3296,
+      "step": 5610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10999369787898e-05,
+      "loss": 0.2969,
+      "step": 5611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097177542336399e-05,
+      "loss": 0.2857,
+      "step": 5612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1094418021317972e-05,
+      "loss": 0.2979,
+      "step": 5613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109165841594721e-05,
+      "loss": 0.2902,
+      "step": 5614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088898726436814e-05,
+      "loss": 0.2621,
+      "step": 5615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086138952999487e-05,
+      "loss": 0.2905,
+      "step": 5616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083379095847933e-05,
+      "loss": 0.2766,
+      "step": 5617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1080619155194873e-05,
+      "loss": 0.2778,
+      "step": 5618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077859131253026e-05,
+      "loss": 0.3027,
+      "step": 5619
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1075099024235123e-05,
+      "loss": 0.2944,
+      "step": 5620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1072338834353902e-05,
+      "loss": 0.298,
+      "step": 5621
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10695785618221e-05,
+      "loss": 0.3042,
+      "step": 5622
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066818206852472e-05,
+      "loss": 0.35,
+      "step": 5623
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1064057769657764e-05,
+      "loss": 0.2962,
+      "step": 5624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106129725045074e-05,
+      "loss": 0.3158,
+      "step": 5625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058536649444167e-05,
+      "loss": 0.2773,
+      "step": 5626
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055775966850821e-05,
+      "loss": 0.3075,
+      "step": 5627
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1053015202883483e-05,
+      "loss": 0.2625,
+      "step": 5628
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050254357754932e-05,
+      "loss": 0.2972,
+      "step": 5629
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104749343167797e-05,
+      "loss": 0.2762,
+      "step": 5630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1044732424865386e-05,
+      "loss": 0.2307,
+      "step": 5631
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104197133752999e-05,
+      "loss": 0.3394,
+      "step": 5632
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1039210169884589e-05,
+      "loss": 0.3195,
+      "step": 5633
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1036448922142004e-05,
+      "loss": 0.3192,
+      "step": 5634
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033687594515062e-05,
+      "loss": 0.3105,
+      "step": 5635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030926187216581e-05,
+      "loss": 0.2896,
+      "step": 5636
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1028164700459409e-05,
+      "loss": 0.2466,
+      "step": 5637
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1025403134456378e-05,
+      "loss": 0.2784,
+      "step": 5638
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1022641489420342e-05,
+      "loss": 0.2577,
+      "step": 5639
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1019879765564155e-05,
+      "loss": 0.2577,
+      "step": 5640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017117963100672e-05,
+      "loss": 0.2607,
+      "step": 5641
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1014356082242766e-05,
+      "loss": 0.3171,
+      "step": 5642
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011594123203302e-05,
+      "loss": 0.2711,
+      "step": 5643
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100883208619516e-05,
+      "loss": 0.2964,
+      "step": 5644
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1006069971431228e-05,
+      "loss": 0.282,
+      "step": 5645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003307779124392e-05,
+      "loss": 0.2993,
+      "step": 5646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1000545509487555e-05,
+      "loss": 0.3268,
+      "step": 5647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997783162733608e-05,
+      "loss": 0.296,
+      "step": 5648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0995020739075468e-05,
+      "loss": 0.2925,
+      "step": 5649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0992258238726046e-05,
+      "loss": 0.2502,
+      "step": 5650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0989495661898259e-05,
+      "loss": 0.3224,
+      "step": 5651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0986733008805038e-05,
+      "loss": 0.27,
+      "step": 5652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0983970279659311e-05,
+      "loss": 0.3111,
+      "step": 5653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0981207474674021e-05,
+      "loss": 0.2794,
+      "step": 5654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0978444594062104e-05,
+      "loss": 0.2609,
+      "step": 5655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975681638036513e-05,
+      "loss": 0.2902,
+      "step": 5656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0972918606810198e-05,
+      "loss": 0.2939,
+      "step": 5657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0970155500596127e-05,
+      "loss": 0.3184,
+      "step": 5658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0967392319607267e-05,
+      "loss": 0.2737,
+      "step": 5659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964629064056583e-05,
+      "loss": 0.3092,
+      "step": 5660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096186573415706e-05,
+      "loss": 0.3104,
+      "step": 5661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0959102330121676e-05,
+      "loss": 0.2681,
+      "step": 5662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956338852163424e-05,
+      "loss": 0.2479,
+      "step": 5663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953575300495299e-05,
+      "loss": 0.3253,
+      "step": 5664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950811675330303e-05,
+      "loss": 0.3229,
+      "step": 5665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0948047976881439e-05,
+      "loss": 0.2815,
+      "step": 5666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0945284205361723e-05,
+      "loss": 0.2849,
+      "step": 5667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0942520360984172e-05,
+      "loss": 0.2917,
+      "step": 5668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939756443961809e-05,
+      "loss": 0.3042,
+      "step": 5669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093699245450766e-05,
+      "loss": 0.2601,
+      "step": 5670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934228392834763e-05,
+      "loss": 0.2871,
+      "step": 5671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931464259156158e-05,
+      "loss": 0.2992,
+      "step": 5672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928700053684893e-05,
+      "loss": 0.2832,
+      "step": 5673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925935776634014e-05,
+      "loss": 0.2737,
+      "step": 5674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923171428216581e-05,
+      "loss": 0.2444,
+      "step": 5675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0920407008645656e-05,
+      "loss": 0.3256,
+      "step": 5676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917642518134304e-05,
+      "loss": 0.225,
+      "step": 5677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0914877956895604e-05,
+      "loss": 0.2691,
+      "step": 5678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0912113325142632e-05,
+      "loss": 0.2818,
+      "step": 5679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909348623088472e-05,
+      "loss": 0.3021,
+      "step": 5680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090658385094621e-05,
+      "loss": 0.311,
+      "step": 5681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903819008928948e-05,
+      "loss": 0.2844,
+      "step": 5682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090105409724978e-05,
+      "loss": 0.2791,
+      "step": 5683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898289116121817e-05,
+      "loss": 0.2706,
+      "step": 5684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0895524065758164e-05,
+      "loss": 0.3007,
+      "step": 5685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892758946371943e-05,
+      "loss": 0.2937,
+      "step": 5686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0889993758176276e-05,
+      "loss": 0.307,
+      "step": 5687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0887228501384287e-05,
+      "loss": 0.3426,
+      "step": 5688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0884463176209105e-05,
+      "loss": 0.2866,
+      "step": 5689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881697782863874e-05,
+      "loss": 0.2861,
+      "step": 5690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0878932321561734e-05,
+      "loss": 0.2752,
+      "step": 5691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0876166792515836e-05,
+      "loss": 0.2538,
+      "step": 5692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873401195939328e-05,
+      "loss": 0.2941,
+      "step": 5693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870635532045375e-05,
+      "loss": 0.3277,
+      "step": 5694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086786980104713e-05,
+      "loss": 0.2917,
+      "step": 5695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0865104003157774e-05,
+      "loss": 0.278,
+      "step": 5696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862338138590479e-05,
+      "loss": 0.2872,
+      "step": 5697
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0859572207558416e-05,
+      "loss": 0.3156,
+      "step": 5698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085680621027478e-05,
+      "loss": 0.2991,
+      "step": 5699
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085404014695275e-05,
+      "loss": 0.2949,
+      "step": 5700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0851274017805525e-05,
+      "loss": 0.2551,
+      "step": 5701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0848507823046306e-05,
+      "loss": 0.3223,
+      "step": 5702
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845741562888297e-05,
+      "loss": 0.292,
+      "step": 5703
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084297523754471e-05,
+      "loss": 0.2451,
+      "step": 5704
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840208847228753e-05,
+      "loss": 0.3699,
+      "step": 5705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0837442392153651e-05,
+      "loss": 0.2649,
+      "step": 5706
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083467587253263e-05,
+      "loss": 0.2656,
+      "step": 5707
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831909288578913e-05,
+      "loss": 0.2858,
+      "step": 5708
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082914264050574e-05,
+      "loss": 0.314,
+      "step": 5709
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082637592852635e-05,
+      "loss": 0.2957,
+      "step": 5710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0823609152853987e-05,
+      "loss": 0.2497,
+      "step": 5711
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820842313701898e-05,
+      "loss": 0.3034,
+      "step": 5712
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818075411283341e-05,
+      "loss": 0.2827,
+      "step": 5713
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081530844581157e-05,
+      "loss": 0.2714,
+      "step": 5714
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812541417499855e-05,
+      "loss": 0.2974,
+      "step": 5715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809774326561464e-05,
+      "loss": 0.2956,
+      "step": 5716
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080700717320967e-05,
+      "loss": 0.2864,
+      "step": 5717
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080423995765775e-05,
+      "loss": 0.3052,
+      "step": 5718
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801472680118984e-05,
+      "loss": 0.3207,
+      "step": 5719
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798705340806668e-05,
+      "loss": 0.2517,
+      "step": 5720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0795937939934088e-05,
+      "loss": 0.28,
+      "step": 5721
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793170477714546e-05,
+      "loss": 0.2748,
+      "step": 5722
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0790402954361344e-05,
+      "loss": 0.2958,
+      "step": 5723
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0787635370087786e-05,
+      "loss": 0.2582,
+      "step": 5724
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0784867725107187e-05,
+      "loss": 0.3481,
+      "step": 5725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078210001963286e-05,
+      "loss": 0.2842,
+      "step": 5726
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0779332253878127e-05,
+      "loss": 0.2576,
+      "step": 5727
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776564428056317e-05,
+      "loss": 0.2343,
+      "step": 5728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0773796542380757e-05,
+      "loss": 0.2874,
+      "step": 5729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771028597064785e-05,
+      "loss": 0.2637,
+      "step": 5730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768260592321735e-05,
+      "loss": 0.2996,
+      "step": 5731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076549252836496e-05,
+      "loss": 0.3318,
+      "step": 5732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0762724405407795e-05,
+      "loss": 0.5643,
+      "step": 5733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759956223663608e-05,
+      "loss": 0.2958,
+      "step": 5734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075718798334575e-05,
+      "loss": 0.2444,
+      "step": 5735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075441968466758e-05,
+      "loss": 0.6011,
+      "step": 5736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751651327842474e-05,
+      "loss": 0.2748,
+      "step": 5737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0748882913083794e-05,
+      "loss": 0.2813,
+      "step": 5738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074611444060492e-05,
+      "loss": 0.3068,
+      "step": 5739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074334591061923e-05,
+      "loss": 0.2938,
+      "step": 5740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740577323340112e-05,
+      "loss": 0.2624,
+      "step": 5741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0737808678980954e-05,
+      "loss": 0.3124,
+      "step": 5742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0735039977755147e-05,
+      "loss": 0.2632,
+      "step": 5743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0732271219876092e-05,
+      "loss": 0.2575,
+      "step": 5744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0729502405557188e-05,
+      "loss": 0.3065,
+      "step": 5745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0726733535011844e-05,
+      "loss": 0.2885,
+      "step": 5746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723964608453467e-05,
+      "loss": 0.2744,
+      "step": 5747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0721195626095477e-05,
+      "loss": 0.2945,
+      "step": 5748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718426588151296e-05,
+      "loss": 0.2759,
+      "step": 5749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071565749483434e-05,
+      "loss": 0.3277,
+      "step": 5750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0712888346358041e-05,
+      "loss": 0.2669,
+      "step": 5751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0710119142935829e-05,
+      "loss": 0.2228,
+      "step": 5752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707349884781142e-05,
+      "loss": 0.2715,
+      "step": 5753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704580572107424e-05,
+      "loss": 0.3269,
+      "step": 5754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0701811205128115e-05,
+      "loss": 0.2429,
+      "step": 5755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0699041784056667e-05,
+      "loss": 0.311,
+      "step": 5756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696272309106532e-05,
+      "loss": 0.2752,
+      "step": 5757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0693502780491168e-05,
+      "loss": 0.2734,
+      "step": 5758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0690733198424035e-05,
+      "loss": 0.3068,
+      "step": 5759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06879635631186e-05,
+      "loss": 0.3011,
+      "step": 5760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685193874788335e-05,
+      "loss": 0.2622,
+      "step": 5761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682424133646712e-05,
+      "loss": 0.2632,
+      "step": 5762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679654339907208e-05,
+      "loss": 0.2817,
+      "step": 5763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0676884493783304e-05,
+      "loss": 0.62,
+      "step": 5764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674114595488489e-05,
+      "loss": 0.2929,
+      "step": 5765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0671344645236253e-05,
+      "loss": 0.3076,
+      "step": 5766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0668574643240087e-05,
+      "loss": 0.2897,
+      "step": 5767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0665804589713494e-05,
+      "loss": 0.3101,
+      "step": 5768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066303448486997e-05,
+      "loss": 0.3101,
+      "step": 5769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0660264328923024e-05,
+      "loss": 0.2764,
+      "step": 5770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0657494122086165e-05,
+      "loss": 0.2588,
+      "step": 5771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0654723864572909e-05,
+      "loss": 0.2718,
+      "step": 5772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0651953556596777e-05,
+      "loss": 0.3024,
+      "step": 5773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.064918319837128e-05,
+      "loss": 0.2729,
+      "step": 5774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646412790109954e-05,
+      "loss": 0.2581,
+      "step": 5775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643642332026323e-05,
+      "loss": 0.3369,
+      "step": 5776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064087182433392e-05,
+      "loss": 0.2566,
+      "step": 5777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0638101267246283e-05,
+      "loss": 0.2369,
+      "step": 5778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635330660976955e-05,
+      "loss": 0.2759,
+      "step": 5779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632560005739481e-05,
+      "loss": 0.3203,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0629789301747404e-05,
+      "loss": 0.2769,
+      "step": 5781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0627018549214284e-05,
+      "loss": 0.2582,
+      "step": 5782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0624247748353666e-05,
+      "loss": 0.2404,
+      "step": 5783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062147689937912e-05,
+      "loss": 0.3202,
+      "step": 5784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618706002504202e-05,
+      "loss": 0.2665,
+      "step": 5785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0615935057942485e-05,
+      "loss": 0.3236,
+      "step": 5786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613164065907539e-05,
+      "loss": 0.2955,
+      "step": 5787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610393026612933e-05,
+      "loss": 0.6254,
+      "step": 5788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0607621940272253e-05,
+      "loss": 0.2924,
+      "step": 5789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060485080709907e-05,
+      "loss": 0.3092,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602079627306979e-05,
+      "loss": 0.265,
+      "step": 5791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0599308401109564e-05,
+      "loss": 0.2678,
+      "step": 5792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0596537128720421e-05,
+      "loss": 0.3041,
+      "step": 5793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0593765810353142e-05,
+      "loss": 0.2837,
+      "step": 5794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059099444622133e-05,
+      "loss": 0.2853,
+      "step": 5795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0588223036538583e-05,
+      "loss": 0.2825,
+      "step": 5796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585451581518513e-05,
+      "loss": 0.2611,
+      "step": 5797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0582680081374728e-05,
+      "loss": 0.3231,
+      "step": 5798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0579908536320842e-05,
+      "loss": 0.3279,
+      "step": 5799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057713694657047e-05,
+      "loss": 0.2585,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0574365312337235e-05,
+      "loss": 0.2878,
+      "step": 5801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571593633834758e-05,
+      "loss": 0.2378,
+      "step": 5802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056882191127667e-05,
+      "loss": 0.294,
+      "step": 5803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0566050144876599e-05,
+      "loss": 0.296,
+      "step": 5804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0563278334848178e-05,
+      "loss": 0.2853,
+      "step": 5805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560506481405048e-05,
+      "loss": 0.3073,
+      "step": 5806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0557734584760849e-05,
+      "loss": 0.2751,
+      "step": 5807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0554962645129223e-05,
+      "loss": 0.2862,
+      "step": 5808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552190662723816e-05,
+      "loss": 0.287,
+      "step": 5809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549418637758284e-05,
+      "loss": 0.2589,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0546646570446277e-05,
+      "loss": 0.3345,
+      "step": 5811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0543874461001456e-05,
+      "loss": 0.36,
+      "step": 5812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0541102309637477e-05,
+      "loss": 0.2893,
+      "step": 5813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0538330116568006e-05,
+      "loss": 0.2719,
+      "step": 5814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0535557882006708e-05,
+      "loss": 0.2906,
+      "step": 5815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0532785606167256e-05,
+      "loss": 0.3034,
+      "step": 5816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530013289263318e-05,
+      "loss": 0.266,
+      "step": 5817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0527240931508582e-05,
+      "loss": 0.278,
+      "step": 5818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0524468533116716e-05,
+      "loss": 0.3,
+      "step": 5819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052169609430141e-05,
+      "loss": 0.2811,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0518923615276342e-05,
+      "loss": 0.2853,
+      "step": 5821
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051615109625521e-05,
+      "loss": 0.2915,
+      "step": 5822
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0513378537451697e-05,
+      "loss": 0.3004,
+      "step": 5823
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0510605939079505e-05,
+      "loss": 0.2691,
+      "step": 5824
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0507833301352335e-05,
+      "loss": 0.2947,
+      "step": 5825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0505060624483878e-05,
+      "loss": 0.2592,
+      "step": 5826
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0502287908687847e-05,
+      "loss": 0.2948,
+      "step": 5827
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499515154177941e-05,
+      "loss": 0.2931,
+      "step": 5828
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049674236116788e-05,
+      "loss": 0.288,
+      "step": 5829
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049396952987137e-05,
+      "loss": 0.6166,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049119666050213e-05,
+      "loss": 0.31,
+      "step": 5831
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048842375327388e-05,
+      "loss": 0.277,
+      "step": 5832
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0485650808400339e-05,
+      "loss": 0.29,
+      "step": 5833
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482877826095233e-05,
+      "loss": 0.2904,
+      "step": 5834
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0480104806572288e-05,
+      "loss": 0.337,
+      "step": 5835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0477331750045239e-05,
+      "loss": 0.2444,
+      "step": 5836
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047455865672782e-05,
+      "loss": 0.2681,
+      "step": 5837
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0471785526833762e-05,
+      "loss": 0.2759,
+      "step": 5838
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0469012360576807e-05,
+      "loss": 0.2717,
+      "step": 5839
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0466239158170697e-05,
+      "loss": 0.3179,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0463465919829175e-05,
+      "loss": 0.3065,
+      "step": 5841
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046069264576599e-05,
+      "loss": 0.3027,
+      "step": 5842
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457919336194892e-05,
+      "loss": 0.2836,
+      "step": 5843
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455145991329639e-05,
+      "loss": 0.3084,
+      "step": 5844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0452372611383977e-05,
+      "loss": 0.2818,
+      "step": 5845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0449599196571671e-05,
+      "loss": 0.2429,
+      "step": 5846
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044682574710648e-05,
+      "loss": 0.2764,
+      "step": 5847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444052263202169e-05,
+      "loss": 0.2596,
+      "step": 5848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04412787450725e-05,
+      "loss": 0.2999,
+      "step": 5849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043850519293125e-05,
+      "loss": 0.3206,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0435731606992188e-05,
+      "loss": 0.2603,
+      "step": 5851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0432957987469081e-05,
+      "loss": 0.2875,
+      "step": 5852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0430184334575715e-05,
+      "loss": 0.2653,
+      "step": 5853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427410648525863e-05,
+      "loss": 0.2701,
+      "step": 5854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0424636929533315e-05,
+      "loss": 0.3293,
+      "step": 5855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0421863177811848e-05,
+      "loss": 0.2586,
+      "step": 5856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0419089393575253e-05,
+      "loss": 0.2738,
+      "step": 5857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041631557703732e-05,
+      "loss": 0.2709,
+      "step": 5858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0413541728411836e-05,
+      "loss": 0.2313,
+      "step": 5859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04107678479126e-05,
+      "loss": 0.2802,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407993935753406e-05,
+      "loss": 0.2375,
+      "step": 5861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405219992148057e-05,
+      "loss": 0.3206,
+      "step": 5862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402446017310355e-05,
+      "loss": 0.2903,
+      "step": 5863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0399672011454101e-05,
+      "loss": 0.2972,
+      "step": 5864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0396897974793102e-05,
+      "loss": 0.2703,
+      "step": 5865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039412390754117e-05,
+      "loss": 0.3087,
+      "step": 5866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0391349809912115e-05,
+      "loss": 0.2697,
+      "step": 5867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388575682119748e-05,
+      "loss": 0.2578,
+      "step": 5868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0385801524377888e-05,
+      "loss": 0.3052,
+      "step": 5869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0383027336900356e-05,
+      "loss": 0.3104,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380253119900967e-05,
+      "loss": 0.2667,
+      "step": 5871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0377478873593546e-05,
+      "loss": 0.2794,
+      "step": 5872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374704598191918e-05,
+      "loss": 0.3147,
+      "step": 5873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371930293909911e-05,
+      "loss": 0.2961,
+      "step": 5874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0369155960961356e-05,
+      "loss": 0.2938,
+      "step": 5875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366381599560086e-05,
+      "loss": 0.2693,
+      "step": 5876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036360720991993e-05,
+      "loss": 0.2653,
+      "step": 5877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0360832792254727e-05,
+      "loss": 0.3221,
+      "step": 5878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0358058346778314e-05,
+      "loss": 0.2729,
+      "step": 5879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355283873704533e-05,
+      "loss": 0.2859,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0352509373247227e-05,
+      "loss": 0.2946,
+      "step": 5881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0349734845620244e-05,
+      "loss": 0.2281,
+      "step": 5882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0346960291037424e-05,
+      "loss": 0.3042,
+      "step": 5883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034418570971262e-05,
+      "loss": 0.5502,
+      "step": 5884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034141110185968e-05,
+      "loss": 0.2851,
+      "step": 5885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0338636467692462e-05,
+      "loss": 0.328,
+      "step": 5886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335861807424816e-05,
+      "loss": 0.2556,
+      "step": 5887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0333087121270602e-05,
+      "loss": 0.3177,
+      "step": 5888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330312409443681e-05,
+      "loss": 0.293,
+      "step": 5889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0327537672157908e-05,
+      "loss": 0.2901,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0324762909627151e-05,
+      "loss": 0.2851,
+      "step": 5891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321988122065274e-05,
+      "loss": 0.2742,
+      "step": 5892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0319213309686145e-05,
+      "loss": 0.3129,
+      "step": 5893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031643847270363e-05,
+      "loss": 0.2683,
+      "step": 5894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.03136636113316e-05,
+      "loss": 0.2502,
+      "step": 5895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031088872578393e-05,
+      "loss": 0.264,
+      "step": 5896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308113816274492e-05,
+      "loss": 0.2546,
+      "step": 5897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0305338883017163e-05,
+      "loss": 0.2793,
+      "step": 5898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302563926225824e-05,
+      "loss": 0.2556,
+      "step": 5899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0299788946114352e-05,
+      "loss": 0.3304,
+      "step": 5900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0297013942896628e-05,
+      "loss": 0.2787,
+      "step": 5901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294238916786537e-05,
+      "loss": 0.3022,
+      "step": 5902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291463867997967e-05,
+      "loss": 0.2983,
+      "step": 5903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0288688796744797e-05,
+      "loss": 0.3044,
+      "step": 5904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0285913703240927e-05,
+      "loss": 0.2782,
+      "step": 5905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283138587700236e-05,
+      "loss": 0.2772,
+      "step": 5906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0280363450336623e-05,
+      "loss": 0.2653,
+      "step": 5907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0277588291363984e-05,
+      "loss": 0.2966,
+      "step": 5908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027481311099621e-05,
+      "loss": 0.2208,
+      "step": 5909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0272037909447197e-05,
+      "loss": 0.3234,
+      "step": 5910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0269262686930845e-05,
+      "loss": 0.2675,
+      "step": 5911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026648744366106e-05,
+      "loss": 0.248,
+      "step": 5912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0263712179851736e-05,
+      "loss": 0.2637,
+      "step": 5913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0260936895716781e-05,
+      "loss": 0.2807,
+      "step": 5914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0258161591470105e-05,
+      "loss": 0.2745,
+      "step": 5915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0255386267325602e-05,
+      "loss": 0.2509,
+      "step": 5916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0252610923497188e-05,
+      "loss": 0.2764,
+      "step": 5917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249835560198772e-05,
+      "loss": 0.29,
+      "step": 5918
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247060177644267e-05,
+      "loss": 0.2548,
+      "step": 5919
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0244284776047589e-05,
+      "loss": 0.2979,
+      "step": 5920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024150935562264e-05,
+      "loss": 0.2521,
+      "step": 5921
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023873391658335e-05,
+      "loss": 0.2816,
+      "step": 5922
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0235958459143623e-05,
+      "loss": 0.2781,
+      "step": 5923
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233182983517387e-05,
+      "loss": 0.2908,
+      "step": 5924
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0230407489918556e-05,
+      "loss": 0.2952,
+      "step": 5925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0227631978561057e-05,
+      "loss": 0.2786,
+      "step": 5926
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224856449658811e-05,
+      "loss": 0.2825,
+      "step": 5927
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022208090342574e-05,
+      "loss": 0.2765,
+      "step": 5928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219305340075767e-05,
+      "loss": 0.2783,
+      "step": 5929
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0216529759822823e-05,
+      "loss": 0.2986,
+      "step": 5930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021375416288084e-05,
+      "loss": 0.3079,
+      "step": 5931
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021097854946374e-05,
+      "loss": 0.291,
+      "step": 5932
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0208202919785453e-05,
+      "loss": 0.2977,
+      "step": 5933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205427274059915e-05,
+      "loss": 0.2658,
+      "step": 5934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020265161250106e-05,
+      "loss": 0.2443,
+      "step": 5935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199875935322815e-05,
+      "loss": 0.2748,
+      "step": 5936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197100242739123e-05,
+      "loss": 0.316,
+      "step": 5937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019432453496392e-05,
+      "loss": 0.2752,
+      "step": 5938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0191548812211143e-05,
+      "loss": 0.2738,
+      "step": 5939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188773074694727e-05,
+      "loss": 0.3325,
+      "step": 5940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0185997322628618e-05,
+      "loss": 0.599,
+      "step": 5941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018322155622675e-05,
+      "loss": 0.2682,
+      "step": 5942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0180445775703074e-05,
+      "loss": 0.3058,
+      "step": 5943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177669981271528e-05,
+      "loss": 0.2665,
+      "step": 5944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174894173146055e-05,
+      "loss": 0.3224,
+      "step": 5945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172118351540608e-05,
+      "loss": 0.2926,
+      "step": 5946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0169342516669125e-05,
+      "loss": 0.2706,
+      "step": 5947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166566668745558e-05,
+      "loss": 0.2663,
+      "step": 5948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0163790807983858e-05,
+      "loss": 0.2739,
+      "step": 5949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016101493459797e-05,
+      "loss": 0.254,
+      "step": 5950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0158239048801848e-05,
+      "loss": 0.29,
+      "step": 5951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155463150809439e-05,
+      "loss": 0.2756,
+      "step": 5952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0152687240834702e-05,
+      "loss": 0.3024,
+      "step": 5953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149911319091583e-05,
+      "loss": 0.2458,
+      "step": 5954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0147135385794043e-05,
+      "loss": 0.2426,
+      "step": 5955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144359441156033e-05,
+      "loss": 0.2494,
+      "step": 5956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014158348539151e-05,
+      "loss": 0.303,
+      "step": 5957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0138807518714435e-05,
+      "loss": 0.2877,
+      "step": 5958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013603154133876e-05,
+      "loss": 0.2545,
+      "step": 5959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0133255553478447e-05,
+      "loss": 0.3298,
+      "step": 5960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013047955534745e-05,
+      "loss": 0.2941,
+      "step": 5961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012770354715974e-05,
+      "loss": 0.2681,
+      "step": 5962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0124927529129267e-05,
+      "loss": 0.277,
+      "step": 5963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0122151501469999e-05,
+      "loss": 0.3475,
+      "step": 5964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01193754643959e-05,
+      "loss": 0.2628,
+      "step": 5965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116599418120924e-05,
+      "loss": 0.2421,
+      "step": 5966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0113823362859042e-05,
+      "loss": 0.2729,
+      "step": 5967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0111047298824222e-05,
+      "loss": 0.2359,
+      "step": 5968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0108271226230423e-05,
+      "loss": 0.2687,
+      "step": 5969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105495145291612e-05,
+      "loss": 0.267,
+      "step": 5970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0102719056221757e-05,
+      "loss": 0.2715,
+      "step": 5971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099942959234826e-05,
+      "loss": 0.3027,
+      "step": 5972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0097166854544782e-05,
+      "loss": 0.3155,
+      "step": 5973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094390742365598e-05,
+      "loss": 0.2945,
+      "step": 5974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0091614622911243e-05,
+      "loss": 0.2979,
+      "step": 5975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0088838496395688e-05,
+      "loss": 0.3051,
+      "step": 5976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0086062363032896e-05,
+      "loss": 0.2727,
+      "step": 5977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0083286223036845e-05,
+      "loss": 0.2344,
+      "step": 5978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080510076621503e-05,
+      "loss": 0.3022,
+      "step": 5979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077733924000841e-05,
+      "loss": 0.2423,
+      "step": 5980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074957765388832e-05,
+      "loss": 0.2502,
+      "step": 5981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007218160099945e-05,
+      "loss": 0.3272,
+      "step": 5982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069405431046669e-05,
+      "loss": 0.2727,
+      "step": 5983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0066629255744458e-05,
+      "loss": 0.2729,
+      "step": 5984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063853075306792e-05,
+      "loss": 0.247,
+      "step": 5985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006107688994765e-05,
+      "loss": 0.2372,
+      "step": 5986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058300699880998e-05,
+      "loss": 0.2842,
+      "step": 5987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055524505320821e-05,
+      "loss": 0.2956,
+      "step": 5988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052748306481088e-05,
+      "loss": 0.2763,
+      "step": 5989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049972103575775e-05,
+      "loss": 0.356,
+      "step": 5990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0047195896818863e-05,
+      "loss": 0.259,
+      "step": 5991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004441968642432e-05,
+      "loss": 0.2883,
+      "step": 5992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0041643472606129e-05,
+      "loss": 0.2598,
+      "step": 5993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038867255578261e-05,
+      "loss": 0.2783,
+      "step": 5994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036091035554703e-05,
+      "loss": 0.2498,
+      "step": 5995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033314812749423e-05,
+      "loss": 0.324,
+      "step": 5996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030538587376402e-05,
+      "loss": 0.2609,
+      "step": 5997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002776235964962e-05,
+      "loss": 0.2521,
+      "step": 5998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024986129783047e-05,
+      "loss": 0.2919,
+      "step": 5999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0022209897990673e-05,
+      "loss": 0.2731,
+      "step": 6000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0019433664486466e-05,
+      "loss": 0.2726,
+      "step": 6001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001665742948441e-05,
+      "loss": 0.29,
+      "step": 6002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001388119319848e-05,
+      "loss": 0.3542,
+      "step": 6003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0011104955842656e-05,
+      "loss": 0.2651,
+      "step": 6004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000832871763092e-05,
+      "loss": 0.2708,
+      "step": 6005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0005552478777244e-05,
+      "loss": 0.2897,
+      "step": 6006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002776239495613e-05,
+      "loss": 0.2759,
+      "step": 6007
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.2853,
+      "step": 6008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997223760504392e-06,
+      "loss": 0.3304,
+      "step": 6009
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994447521222758e-06,
+      "loss": 0.2999,
+      "step": 6010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991671282369084e-06,
+      "loss": 0.2718,
+      "step": 6011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988895044157345e-06,
+      "loss": 0.2897,
+      "step": 6012
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986118806801525e-06,
+      "loss": 0.2991,
+      "step": 6013
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983342570515592e-06,
+      "loss": 0.2459,
+      "step": 6014
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.980566335513538e-06,
+      "loss": 0.2974,
+      "step": 6015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.977790102009332e-06,
+      "loss": 0.3033,
+      "step": 6016
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975013870216953e-06,
+      "loss": 0.3109,
+      "step": 6017
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.972237640350383e-06,
+      "loss": 0.2469,
+      "step": 6018
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969461412623601e-06,
+      "loss": 0.2933,
+      "step": 6019
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966685187250582e-06,
+      "loss": 0.302,
+      "step": 6020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963908964445298e-06,
+      "loss": 0.3113,
+      "step": 6021
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96113274442174e-06,
+      "loss": 0.2988,
+      "step": 6022
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.958356527393876e-06,
+      "loss": 0.2715,
+      "step": 6023
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.955580313575683e-06,
+      "loss": 0.3322,
+      "step": 6024
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95280410318114e-06,
+      "loss": 0.2742,
+      "step": 6025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950027896424228e-06,
+      "loss": 0.2628,
+      "step": 6026
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.947251693518917e-06,
+      "loss": 0.2476,
+      "step": 6027
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94447549467918e-06,
+      "loss": 0.2836,
+      "step": 6028
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941699300119004e-06,
+      "loss": 0.2685,
+      "step": 6029
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.938923110052353e-06,
+      "loss": 0.3553,
+      "step": 6030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93614692469321e-06,
+      "loss": 0.2778,
+      "step": 6031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.933370744255543e-06,
+      "loss": 0.2977,
+      "step": 6032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930594568953335e-06,
+      "loss": 0.3034,
+      "step": 6033
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92781839900055e-06,
+      "loss": 0.2576,
+      "step": 6034
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92504223461117e-06,
+      "loss": 0.2761,
+      "step": 6035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922266075999162e-06,
+      "loss": 0.2675,
+      "step": 6036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9194899233785e-06,
+      "loss": 0.284,
+      "step": 6037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.916713776963156e-06,
+      "loss": 0.5876,
+      "step": 6038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.913937636967106e-06,
+      "loss": 0.3221,
+      "step": 6039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.911161503604317e-06,
+      "loss": 0.2642,
+      "step": 6040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.908385377088757e-06,
+      "loss": 0.2617,
+      "step": 6041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.905609257634404e-06,
+      "loss": 0.2643,
+      "step": 6042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.902833145455221e-06,
+      "loss": 0.2636,
+      "step": 6043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90005704076518e-06,
+      "loss": 0.3041,
+      "step": 6044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897280943778245e-06,
+      "loss": 0.2446,
+      "step": 6045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894504854708391e-06,
+      "loss": 0.3168,
+      "step": 6046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.891728773769582e-06,
+      "loss": 0.2867,
+      "step": 6047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88895270117578e-06,
+      "loss": 0.2738,
+      "step": 6048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.886176637140959e-06,
+      "loss": 0.2551,
+      "step": 6049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883400581879077e-06,
+      "loss": 0.2531,
+      "step": 6050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880624535604107e-06,
+      "loss": 0.2666,
+      "step": 6051
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877848498530001e-06,
+      "loss": 0.318,
+      "step": 6052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875072470870735e-06,
+      "loss": 0.2891,
+      "step": 6053
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872296452840266e-06,
+      "loss": 0.3179,
+      "step": 6054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86952044465255e-06,
+      "loss": 0.2405,
+      "step": 6055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.866744446521555e-06,
+      "loss": 0.2767,
+      "step": 6056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863968458661244e-06,
+      "loss": 0.3212,
+      "step": 6057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86119248128557e-06,
+      "loss": 0.2786,
+      "step": 6058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85841651460849e-06,
+      "loss": 0.2606,
+      "step": 6059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85564055884397e-06,
+      "loss": 0.2619,
+      "step": 6060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85286461420596e-06,
+      "loss": 0.2863,
+      "step": 6061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85008868090842e-06,
+      "loss": 0.3206,
+      "step": 6062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847312759165301e-06,
+      "loss": 0.2991,
+      "step": 6063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844536849190564e-06,
+      "loss": 0.2839,
+      "step": 6064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.841760951198159e-06,
+      "loss": 0.3005,
+      "step": 6065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838985065402032e-06,
+      "loss": 0.2558,
+      "step": 6066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.836209192016146e-06,
+      "loss": 0.2752,
+      "step": 6067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833433331254443e-06,
+      "loss": 0.2789,
+      "step": 6068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830657483330877e-06,
+      "loss": 0.2729,
+      "step": 6069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827881648459396e-06,
+      "loss": 0.2756,
+      "step": 6070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825105826853946e-06,
+      "loss": 0.28,
+      "step": 6071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822330018728474e-06,
+      "loss": 0.2957,
+      "step": 6072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819554224296929e-06,
+      "loss": 0.251,
+      "step": 6073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.816778443773253e-06,
+      "loss": 0.2593,
+      "step": 6074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.814002677371387e-06,
+      "loss": 0.2833,
+      "step": 6075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811226925305273e-06,
+      "loss": 0.2727,
+      "step": 6076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80845118778886e-06,
+      "loss": 0.252,
+      "step": 6077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805675465036084e-06,
+      "loss": 0.2828,
+      "step": 6078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802899757260875e-06,
+      "loss": 0.2953,
+      "step": 6079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800124064677186e-06,
+      "loss": 0.3062,
+      "step": 6080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797348387498944e-06,
+      "loss": 0.2557,
+      "step": 6081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794572725940088e-06,
+      "loss": 0.2673,
+      "step": 6082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791797080214547e-06,
+      "loss": 0.2929,
+      "step": 6083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789021450536265e-06,
+      "loss": 0.2668,
+      "step": 6084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.786245837119165e-06,
+      "loss": 0.2434,
+      "step": 6085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783470240177175e-06,
+      "loss": 0.3055,
+      "step": 6086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780694659924235e-06,
+      "loss": 0.3118,
+      "step": 6087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.777919096574264e-06,
+      "loss": 0.2756,
+      "step": 6088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775143550341194e-06,
+      "loss": 0.3181,
+      "step": 6089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.772368021438943e-06,
+      "loss": 0.2761,
+      "step": 6090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769592510081445e-06,
+      "loss": 0.2732,
+      "step": 6091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766817016482618e-06,
+      "loss": 0.3576,
+      "step": 6092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.764041540856379e-06,
+      "loss": 0.3076,
+      "step": 6093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761266083416655e-06,
+      "loss": 0.27,
+      "step": 6094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758490644377363e-06,
+      "loss": 0.2689,
+      "step": 6095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.755715223952418e-06,
+      "loss": 0.2906,
+      "step": 6096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752939822355733e-06,
+      "loss": 0.308,
+      "step": 6097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75016443980123e-06,
+      "loss": 0.256,
+      "step": 6098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747389076502814e-06,
+      "loss": 0.2881,
+      "step": 6099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.744613732674401e-06,
+      "loss": 0.2219,
+      "step": 6100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.7418384085299e-06,
+      "loss": 0.2842,
+      "step": 6101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73906310428322e-06,
+      "loss": 0.2794,
+      "step": 6102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.736287820148269e-06,
+      "loss": 0.2455,
+      "step": 6103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.733512556338941e-06,
+      "loss": 0.278,
+      "step": 6104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.730737313069157e-06,
+      "loss": 0.2773,
+      "step": 6105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.727962090552808e-06,
+      "loss": 0.3433,
+      "step": 6106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725186889003795e-06,
+      "loss": 0.2629,
+      "step": 6107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.722411708636018e-06,
+      "loss": 0.2755,
+      "step": 6108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719636549663379e-06,
+      "loss": 0.6019,
+      "step": 6109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716861412299769e-06,
+      "loss": 0.3077,
+      "step": 6110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.714086296759078e-06,
+      "loss": 0.263,
+      "step": 6111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.711311203255207e-06,
+      "loss": 0.3013,
+      "step": 6112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.708536132002038e-06,
+      "loss": 0.2523,
+      "step": 6113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705761083213463e-06,
+      "loss": 0.2492,
+      "step": 6114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.702986057103375e-06,
+      "loss": 0.6278,
+      "step": 6115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700211053885653e-06,
+      "loss": 0.3215,
+      "step": 6116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697436073774178e-06,
+      "loss": 0.2739,
+      "step": 6117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694661116982838e-06,
+      "loss": 0.2698,
+      "step": 6118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691886183725512e-06,
+      "loss": 0.2839,
+      "step": 6119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689111274216075e-06,
+      "loss": 0.3051,
+      "step": 6120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.6863363886684e-06,
+      "loss": 0.2514,
+      "step": 6121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.683561527296375e-06,
+      "loss": 0.2735,
+      "step": 6122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68078669031386e-06,
+      "loss": 0.2988,
+      "step": 6123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678011877934728e-06,
+      "loss": 0.2815,
+      "step": 6124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675237090372852e-06,
+      "loss": 0.296,
+      "step": 6125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672462327842095e-06,
+      "loss": 0.2869,
+      "step": 6126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669687590556325e-06,
+      "loss": 0.2947,
+      "step": 6127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666912878729398e-06,
+      "loss": 0.2599,
+      "step": 6128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664138192575187e-06,
+      "loss": 0.264,
+      "step": 6129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.661363532307543e-06,
+      "loss": 0.2852,
+      "step": 6130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658588898140322e-06,
+      "loss": 0.2942,
+      "step": 6131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.655814290287381e-06,
+      "loss": 0.3217,
+      "step": 6132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65303970896258e-06,
+      "loss": 0.3466,
+      "step": 6133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650265154379761e-06,
+      "loss": 0.3008,
+      "step": 6134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.647490626752773e-06,
+      "loss": 0.2971,
+      "step": 6135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644716126295468e-06,
+      "loss": 0.3315,
+      "step": 6136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.641941653221687e-06,
+      "loss": 0.3229,
+      "step": 6137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.639167207745276e-06,
+      "loss": 0.3083,
+      "step": 6138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.636392790080073e-06,
+      "loss": 0.3123,
+      "step": 6139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633618400439918e-06,
+      "loss": 0.2985,
+      "step": 6140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630844039038647e-06,
+      "loss": 0.3315,
+      "step": 6141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628069706090089e-06,
+      "loss": 0.2172,
+      "step": 6142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625295401808085e-06,
+      "loss": 0.2834,
+      "step": 6143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622521126406458e-06,
+      "loss": 0.2465,
+      "step": 6144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619746880099036e-06,
+      "loss": 0.2546,
+      "step": 6145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.616972663099648e-06,
+      "loss": 0.2654,
+      "step": 6146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614198475622113e-06,
+      "loss": 0.2982,
+      "step": 6147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611424317880257e-06,
+      "loss": 0.2773,
+      "step": 6148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.608650190087888e-06,
+      "loss": 0.2869,
+      "step": 6149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.605876092458835e-06,
+      "loss": 0.3239,
+      "step": 6150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.603102025206901e-06,
+      "loss": 0.2952,
+      "step": 6151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6003279885459e-06,
+      "loss": 0.28,
+      "step": 6152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597553982689649e-06,
+      "loss": 0.3323,
+      "step": 6153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594780007851947e-06,
+      "loss": 0.2784,
+      "step": 6154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592006064246596e-06,
+      "loss": 0.2544,
+      "step": 6155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.589232152087404e-06,
+      "loss": 0.315,
+      "step": 6156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586458271588167e-06,
+      "loss": 0.2506,
+      "step": 6157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.583684422962686e-06,
+      "loss": 0.2751,
+      "step": 6158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.580910606424747e-06,
+      "loss": 0.289,
+      "step": 6159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578136822188154e-06,
+      "loss": 0.2582,
+      "step": 6160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.575363070466689e-06,
+      "loss": 0.2971,
+      "step": 6161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572589351474135e-06,
+      "loss": 0.2603,
+      "step": 6162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.569815665424288e-06,
+      "loss": 0.2667,
+      "step": 6163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56704201253092e-06,
+      "loss": 0.2909,
+      "step": 6164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.564268393007819e-06,
+      "loss": 0.28,
+      "step": 6165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56149480706875e-06,
+      "loss": 0.2992,
+      "step": 6166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.558721254927501e-06,
+      "loss": 0.6023,
+      "step": 6167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555947736797836e-06,
+      "loss": 0.2901,
+      "step": 6168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553174252893522e-06,
+      "loss": 0.2755,
+      "step": 6169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55040080342833e-06,
+      "loss": 0.3389,
+      "step": 6170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547627388616026e-06,
+      "loss": 0.2678,
+      "step": 6171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.544854008670366e-06,
+      "loss": 0.2711,
+      "step": 6172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542080663805108e-06,
+      "loss": 0.2574,
+      "step": 6173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.539307354234013e-06,
+      "loss": 0.2719,
+      "step": 6174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.536534080170827e-06,
+      "loss": 0.2614,
+      "step": 6175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.533760841829306e-06,
+      "loss": 0.2811,
+      "step": 6176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.530987639423195e-06,
+      "loss": 0.3169,
+      "step": 6177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.528214473166241e-06,
+      "loss": 0.2661,
+      "step": 6178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525441343272185e-06,
+      "loss": 0.6213,
+      "step": 6179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522668249954761e-06,
+      "loss": 0.3129,
+      "step": 6180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.519895193427713e-06,
+      "loss": 0.3321,
+      "step": 6181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51712217390477e-06,
+      "loss": 0.5764,
+      "step": 6182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.514349191599665e-06,
+      "loss": 0.3439,
+      "step": 6183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511576246726123e-06,
+      "loss": 0.2642,
+      "step": 6184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.508803339497872e-06,
+      "loss": 0.2975,
+      "step": 6185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.506030470128635e-06,
+      "loss": 0.3195,
+      "step": 6186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.503257638832122e-06,
+      "loss": 0.3178,
+      "step": 6187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50048484582206e-06,
+      "loss": 0.2945,
+      "step": 6188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.497712091312158e-06,
+      "loss": 0.3174,
+      "step": 6189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494939375516122e-06,
+      "loss": 0.293,
+      "step": 6190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49216669864767e-06,
+      "loss": 0.2802,
+      "step": 6191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489394060920496e-06,
+      "loss": 0.3081,
+      "step": 6192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.486621462548307e-06,
+      "loss": 0.308,
+      "step": 6193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.483848903744795e-06,
+      "loss": 0.2646,
+      "step": 6194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48107638472366e-06,
+      "loss": 0.2584,
+      "step": 6195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.478303905698595e-06,
+      "loss": 0.2626,
+      "step": 6196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.475531466883284e-06,
+      "loss": 0.2735,
+      "step": 6197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472759068491421e-06,
+      "loss": 0.2536,
+      "step": 6198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469986710736683e-06,
+      "loss": 0.2912,
+      "step": 6199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.467214393832746e-06,
+      "loss": 0.2922,
+      "step": 6200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.464442117993296e-06,
+      "loss": 0.2674,
+      "step": 6201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.461669883431997e-06,
+      "loss": 0.2382,
+      "step": 6202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458897690362528e-06,
+      "loss": 0.2496,
+      "step": 6203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.456125538998546e-06,
+      "loss": 0.2491,
+      "step": 6204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.453353429553724e-06,
+      "loss": 0.281,
+      "step": 6205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45058136224172e-06,
+      "loss": 0.2861,
+      "step": 6206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447809337276184e-06,
+      "loss": 0.2622,
+      "step": 6207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44503735487078e-06,
+      "loss": 0.255,
+      "step": 6208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442265415239154e-06,
+      "loss": 0.2674,
+      "step": 6209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439493518594957e-06,
+      "loss": 0.2699,
+      "step": 6210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436721665151824e-06,
+      "loss": 0.2976,
+      "step": 6211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433949855123405e-06,
+      "loss": 0.3428,
+      "step": 6212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.431178088723334e-06,
+      "loss": 0.2438,
+      "step": 6213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.428406366165244e-06,
+      "loss": 0.2827,
+      "step": 6214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425634687662768e-06,
+      "loss": 0.2614,
+      "step": 6215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422863053429534e-06,
+      "loss": 0.2676,
+      "step": 6216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.420091463679164e-06,
+      "loss": 0.3492,
+      "step": 6217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.417319918625274e-06,
+      "loss": 0.2786,
+      "step": 6218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41454841848149e-06,
+      "loss": 0.3083,
+      "step": 6219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41177696346142e-06,
+      "loss": 0.3035,
+      "step": 6220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409005553778673e-06,
+      "loss": 0.27,
+      "step": 6221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40623418964686e-06,
+      "loss": 0.2797,
+      "step": 6222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.403462871279582e-06,
+      "loss": 0.2883,
+      "step": 6223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40069159889044e-06,
+      "loss": 0.2992,
+      "step": 6224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397920372693023e-06,
+      "loss": 0.2802,
+      "step": 6225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395149192900934e-06,
+      "loss": 0.6013,
+      "step": 6226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392378059727752e-06,
+      "loss": 0.2268,
+      "step": 6227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389606973387067e-06,
+      "loss": 0.3,
+      "step": 6228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386835934092464e-06,
+      "loss": 0.2955,
+      "step": 6229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.384064942057518e-06,
+      "loss": 0.3259,
+      "step": 6230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.381293997495801e-06,
+      "loss": 0.3099,
+      "step": 6231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378523100620883e-06,
+      "loss": 0.2547,
+      "step": 6232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.375752251646336e-06,
+      "loss": 0.2751,
+      "step": 6233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372981450785723e-06,
+      "loss": 0.2932,
+      "step": 6234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370210698252597e-06,
+      "loss": 0.3113,
+      "step": 6235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367439994260522e-06,
+      "loss": 0.3331,
+      "step": 6236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.364669339023047e-06,
+      "loss": 0.2291,
+      "step": 6237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361898732753715e-06,
+      "loss": 0.2784,
+      "step": 6238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359128175666083e-06,
+      "loss": 0.2957,
+      "step": 6239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356357667973679e-06,
+      "loss": 0.262,
+      "step": 6240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.353587209890049e-06,
+      "loss": 0.3329,
+      "step": 6241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.35081680162872e-06,
+      "loss": 0.2634,
+      "step": 6242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.348046443403227e-06,
+      "loss": 0.3326,
+      "step": 6243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345276135427093e-06,
+      "loss": 0.6267,
+      "step": 6244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.342505877913835e-06,
+      "loss": 0.2528,
+      "step": 6245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339735671076978e-06,
+      "loss": 0.3056,
+      "step": 6246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336965515130034e-06,
+      "loss": 0.2305,
+      "step": 6247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.334195410286513e-06,
+      "loss": 0.2986,
+      "step": 6248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.331425356759915e-06,
+      "loss": 0.2968,
+      "step": 6249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32865535476375e-06,
+      "loss": 0.231,
+      "step": 6250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325885404511513e-06,
+      "loss": 0.313,
+      "step": 6251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.323115506216698e-06,
+      "loss": 0.2585,
+      "step": 6252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320345660092794e-06,
+      "loss": 0.2469,
+      "step": 6253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317575866353293e-06,
+      "loss": 0.2513,
+      "step": 6254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314806125211669e-06,
+      "loss": 0.2708,
+      "step": 6255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.312036436881402e-06,
+      "loss": 0.3019,
+      "step": 6256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.309266801575968e-06,
+      "loss": 0.2678,
+      "step": 6257
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306497219508835e-06,
+      "loss": 0.2935,
+      "step": 6258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30372769089347e-06,
+      "loss": 0.2889,
+      "step": 6259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.300958215943335e-06,
+      "loss": 0.2858,
+      "step": 6260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298188794871888e-06,
+      "loss": 0.3174,
+      "step": 6261
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29541942789258e-06,
+      "loss": 0.288,
+      "step": 6262
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29265011521886e-06,
+      "loss": 0.2868,
+      "step": 6263
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289880857064175e-06,
+      "loss": 0.2936,
+      "step": 6264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.287111653641964e-06,
+      "loss": 0.2488,
+      "step": 6265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28434250516566e-06,
+      "loss": 0.322,
+      "step": 6266
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281573411848707e-06,
+      "loss": 0.2982,
+      "step": 6267
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.278804373904525e-06,
+      "loss": 0.2462,
+      "step": 6268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276035391546538e-06,
+      "loss": 0.2831,
+      "step": 6269
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27326646498816e-06,
+      "loss": 0.2899,
+      "step": 6270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270497594442815e-06,
+      "loss": 0.2493,
+      "step": 6271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.267728780123913e-06,
+      "loss": 0.2828,
+      "step": 6272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264960022244855e-06,
+      "loss": 0.248,
+      "step": 6273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.262191321019049e-06,
+      "loss": 0.2837,
+      "step": 6274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259422676659892e-06,
+      "loss": 0.2541,
+      "step": 6275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.256654089380773e-06,
+      "loss": 0.2232,
+      "step": 6276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253885559395084e-06,
+      "loss": 0.2437,
+      "step": 6277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.251117086916209e-06,
+      "loss": 0.3179,
+      "step": 6278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24834867215753e-06,
+      "loss": 0.2921,
+      "step": 6279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.245580315332418e-06,
+      "loss": 0.3156,
+      "step": 6280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.242812016654252e-06,
+      "loss": 0.2936,
+      "step": 6281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240043776336397e-06,
+      "loss": 0.3064,
+      "step": 6282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.237275594592205e-06,
+      "loss": 0.2792,
+      "step": 6283
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234507471635043e-06,
+      "loss": 0.2784,
+      "step": 6284
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231739407678267e-06,
+      "loss": 0.2673,
+      "step": 6285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22897140293522e-06,
+      "loss": 0.2992,
+      "step": 6286
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.226203457619245e-06,
+      "loss": 0.5927,
+      "step": 6287
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223435571943685e-06,
+      "loss": 0.2769,
+      "step": 6288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220667746121875e-06,
+      "loss": 0.2463,
+      "step": 6289
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217899980367142e-06,
+      "loss": 0.3136,
+      "step": 6290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.215132274892817e-06,
+      "loss": 0.2569,
+      "step": 6291
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212364629912218e-06,
+      "loss": 0.2844,
+      "step": 6292
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209597045638661e-06,
+      "loss": 0.2471,
+      "step": 6293
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206829522285456e-06,
+      "loss": 0.3096,
+      "step": 6294
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.204062060065915e-06,
+      "loss": 0.2534,
+      "step": 6295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201294659193337e-06,
+      "loss": 0.2936,
+      "step": 6296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198527319881017e-06,
+      "loss": 0.2944,
+      "step": 6297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195760042342254e-06,
+      "loss": 0.2858,
+      "step": 6298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192992826790335e-06,
+      "loss": 0.2599,
+      "step": 6299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19022567343854e-06,
+      "loss": 0.2722,
+      "step": 6300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187458582500145e-06,
+      "loss": 0.2665,
+      "step": 6301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184691554188432e-06,
+      "loss": 0.2485,
+      "step": 6302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181924588716664e-06,
+      "loss": 0.2971,
+      "step": 6303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179157686298104e-06,
+      "loss": 0.2433,
+      "step": 6304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.176390847146018e-06,
+      "loss": 0.2442,
+      "step": 6305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173624071473655e-06,
+      "loss": 0.2957,
+      "step": 6306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170857359494265e-06,
+      "loss": 0.2916,
+      "step": 6307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.168090711421089e-06,
+      "loss": 0.2991,
+      "step": 6308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165324127467375e-06,
+      "loss": 0.2511,
+      "step": 6309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.162557607846352e-06,
+      "loss": 0.2816,
+      "step": 6310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159791152771247e-06,
+      "loss": 0.2364,
+      "step": 6311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.157024762455292e-06,
+      "loss": 0.2662,
+      "step": 6312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.154258437111706e-06,
+      "loss": 0.2716,
+      "step": 6313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.151492176953697e-06,
+      "loss": 0.267,
+      "step": 6314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148725982194477e-06,
+      "loss": 0.2935,
+      "step": 6315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145959853047254e-06,
+      "loss": 0.6002,
+      "step": 6316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.143193789725227e-06,
+      "loss": 0.2593,
+      "step": 6317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140427792441584e-06,
+      "loss": 0.324,
+      "step": 6318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137661861409525e-06,
+      "loss": 0.2525,
+      "step": 6319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134895996842228e-06,
+      "loss": 0.3224,
+      "step": 6320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13213019895287e-06,
+      "loss": 0.3298,
+      "step": 6321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129364467954628e-06,
+      "loss": 0.2899,
+      "step": 6322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126598804060675e-06,
+      "loss": 0.3036,
+      "step": 6323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.123833207484169e-06,
+      "loss": 0.2993,
+      "step": 6324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121067678438267e-06,
+      "loss": 0.3029,
+      "step": 6325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11830221713613e-06,
+      "loss": 0.2406,
+      "step": 6326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115536823790896e-06,
+      "loss": 0.2467,
+      "step": 6327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112771498615717e-06,
+      "loss": 0.3025,
+      "step": 6328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.110006241823726e-06,
+      "loss": 0.2587,
+      "step": 6329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107241053628058e-06,
+      "loss": 0.348,
+      "step": 6330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104475934241839e-06,
+      "loss": 0.2704,
+      "step": 6331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101710883878185e-06,
+      "loss": 0.2437,
+      "step": 6332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098945902750222e-06,
+      "loss": 0.2874,
+      "step": 6333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096180991071055e-06,
+      "loss": 0.2757,
+      "step": 6334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.093416149053791e-06,
+      "loss": 0.3019,
+      "step": 6335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090651376911532e-06,
+      "loss": 0.2715,
+      "step": 6336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087886674857371e-06,
+      "loss": 0.2633,
+      "step": 6337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0851220431044e-06,
+      "loss": 0.3062,
+      "step": 6338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.082357481865697e-06,
+      "loss": 0.2756,
+      "step": 6339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07959299135435e-06,
+      "loss": 0.2933,
+      "step": 6340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076828571783422e-06,
+      "loss": 0.2599,
+      "step": 6341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074064223365986e-06,
+      "loss": 0.2508,
+      "step": 6342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07129994631511e-06,
+      "loss": 0.2957,
+      "step": 6343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068535740843844e-06,
+      "loss": 0.2809,
+      "step": 6344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065771607165242e-06,
+      "loss": 0.2898,
+      "step": 6345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.063007545492342e-06,
+      "loss": 0.3552,
+      "step": 6346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060243556038195e-06,
+      "loss": 0.2702,
+      "step": 6347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.057479639015832e-06,
+      "loss": 0.2979,
+      "step": 6348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054715794638277e-06,
+      "loss": 0.3071,
+      "step": 6349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.051952023118563e-06,
+      "loss": 0.2598,
+      "step": 6350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0491883246697e-06,
+      "loss": 0.2349,
+      "step": 6351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046424699504703e-06,
+      "loss": 0.278,
+      "step": 6352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043661147836578e-06,
+      "loss": 0.2801,
+      "step": 6353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.040897669878327e-06,
+      "loss": 0.2824,
+      "step": 6354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.038134265842946e-06,
+      "loss": 0.3063,
+      "step": 6355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035370935943419e-06,
+      "loss": 0.2936,
+      "step": 6356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.032607680392738e-06,
+      "loss": 0.2664,
+      "step": 6357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.029844499403876e-06,
+      "loss": 0.3663,
+      "step": 6358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027081393189804e-06,
+      "loss": 0.2704,
+      "step": 6359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02431836196349e-06,
+      "loss": 0.2867,
+      "step": 6360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.021555405937901e-06,
+      "loss": 0.2859,
+      "step": 6361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.018792525325986e-06,
+      "loss": 0.2525,
+      "step": 6362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016029720340688e-06,
+      "loss": 0.2542,
+      "step": 6363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.013266991194964e-06,
+      "loss": 0.3002,
+      "step": 6364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010504338101743e-06,
+      "loss": 0.304,
+      "step": 6365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007741761273957e-06,
+      "loss": 0.287,
+      "step": 6366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.004979260924534e-06,
+      "loss": 0.2683,
+      "step": 6367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002216837266394e-06,
+      "loss": 0.2769,
+      "step": 6368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999454490512451e-06,
+      "loss": 0.3268,
+      "step": 6369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996692220875608e-06,
+      "loss": 0.2604,
+      "step": 6370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.993930028568775e-06,
+      "loss": 0.2933,
+      "step": 6371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991167913804842e-06,
+      "loss": 0.2385,
+      "step": 6372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988405876796701e-06,
+      "loss": 0.2835,
+      "step": 6373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.985643917757237e-06,
+      "loss": 0.2943,
+      "step": 6374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98288203689933e-06,
+      "loss": 0.2914,
+      "step": 6375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98012023443585e-06,
+      "loss": 0.2491,
+      "step": 6376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.977358510579658e-06,
+      "loss": 0.2793,
+      "step": 6377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.974596865543624e-06,
+      "loss": 0.267,
+      "step": 6378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.971835299540595e-06,
+      "loss": 0.3075,
+      "step": 6379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969073812783417e-06,
+      "loss": 0.2628,
+      "step": 6380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966312405484942e-06,
+      "loss": 0.3061,
+      "step": 6381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963551077857999e-06,
+      "loss": 0.6271,
+      "step": 6382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.960789830115416e-06,
+      "loss": 0.3176,
+      "step": 6383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.958028662470014e-06,
+      "loss": 0.2692,
+      "step": 6384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.955267575134618e-06,
+      "loss": 0.2994,
+      "step": 6385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.952506568322036e-06,
+      "loss": 0.2919,
+      "step": 6386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949745642245066e-06,
+      "loss": 0.262,
+      "step": 6387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94698479711652e-06,
+      "loss": 0.2734,
+      "step": 6388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944224033149182e-06,
+      "loss": 0.3241,
+      "step": 6389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941463350555835e-06,
+      "loss": 0.3006,
+      "step": 6390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.938702749549264e-06,
+      "loss": 0.2688,
+      "step": 6391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93594223034224e-06,
+      "loss": 0.2596,
+      "step": 6392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933181793147534e-06,
+      "loss": 0.2806,
+      "step": 6393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9304214381779e-06,
+      "loss": 0.2568,
+      "step": 6394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9276611656461e-06,
+      "loss": 0.2725,
+      "step": 6395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924900975764879e-06,
+      "loss": 0.269,
+      "step": 6396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922140868746977e-06,
+      "loss": 0.2949,
+      "step": 6397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.919380844805129e-06,
+      "loss": 0.338,
+      "step": 6398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916620904152069e-06,
+      "loss": 0.2419,
+      "step": 6399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913861047000518e-06,
+      "loss": 0.2715,
+      "step": 6400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911101273563185e-06,
+      "loss": 0.3061,
+      "step": 6401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908341584052791e-06,
+      "loss": 0.2795,
+      "step": 6402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.905581978682031e-06,
+      "loss": 0.2607,
+      "step": 6403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902822457663605e-06,
+      "loss": 0.2841,
+      "step": 6404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.900063021210203e-06,
+      "loss": 0.3083,
+      "step": 6405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897303669534508e-06,
+      "loss": 0.2467,
+      "step": 6406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8945444028492e-06,
+      "loss": 0.2706,
+      "step": 6407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891785221366942e-06,
+      "loss": 0.2778,
+      "step": 6408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889026125300407e-06,
+      "loss": 0.259,
+      "step": 6409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886267114862248e-06,
+      "loss": 0.2808,
+      "step": 6410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883508190265117e-06,
+      "loss": 0.277,
+      "step": 6411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880749351721656e-06,
+      "loss": 0.6112,
+      "step": 6412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877990599444508e-06,
+      "loss": 0.2902,
+      "step": 6413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8752319336463e-06,
+      "loss": 0.3396,
+      "step": 6414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872473354539653e-06,
+      "loss": 0.2559,
+      "step": 6415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869714862337194e-06,
+      "loss": 0.3209,
+      "step": 6416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866956457251524e-06,
+      "loss": 0.2571,
+      "step": 6417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86419813949525e-06,
+      "loss": 0.311,
+      "step": 6418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861439909280976e-06,
+      "loss": 0.265,
+      "step": 6419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.858681766821289e-06,
+      "loss": 0.2789,
+      "step": 6420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855923712328771e-06,
+      "loss": 0.2816,
+      "step": 6421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853165746015997e-06,
+      "loss": 0.25,
+      "step": 6422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.850407868095541e-06,
+      "loss": 0.2814,
+      "step": 6423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84765007877997e-06,
+      "loss": 0.2615,
+      "step": 6424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.844892378281831e-06,
+      "loss": 0.3049,
+      "step": 6425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842134766813687e-06,
+      "loss": 0.3261,
+      "step": 6426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.839377244588073e-06,
+      "loss": 0.2766,
+      "step": 6427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836619811817522e-06,
+      "loss": 0.2922,
+      "step": 6428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83386246871457e-06,
+      "loss": 0.3375,
+      "step": 6429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831105215491734e-06,
+      "loss": 0.314,
+      "step": 6430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.828348052361537e-06,
+      "loss": 0.2966,
+      "step": 6431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825590979536476e-06,
+      "loss": 0.2787,
+      "step": 6432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822833997229065e-06,
+      "loss": 0.2784,
+      "step": 6433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820077105651794e-06,
+      "loss": 0.3019,
+      "step": 6434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817320305017148e-06,
+      "loss": 0.2717,
+      "step": 6435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.814563595537604e-06,
+      "loss": 0.2824,
+      "step": 6436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811806977425647e-06,
+      "loss": 0.2773,
+      "step": 6437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.809050450893737e-06,
+      "loss": 0.2692,
+      "step": 6438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806294016154329e-06,
+      "loss": 0.2727,
+      "step": 6439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.803537673419885e-06,
+      "loss": 0.3005,
+      "step": 6440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80078142290284e-06,
+      "loss": 0.3063,
+      "step": 6441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.798025264815643e-06,
+      "loss": 0.2875,
+      "step": 6442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.795269199370714e-06,
+      "loss": 0.2943,
+      "step": 6443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.792513226780486e-06,
+      "loss": 0.2686,
+      "step": 6444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789757347257373e-06,
+      "loss": 0.3112,
+      "step": 6445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787001561013779e-06,
+      "loss": 0.2678,
+      "step": 6446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784245868262117e-06,
+      "loss": 0.2623,
+      "step": 6447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.781490269214772e-06,
+      "loss": 0.3001,
+      "step": 6448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778734764084136e-06,
+      "loss": 0.2631,
+      "step": 6449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77597935308259e-06,
+      "loss": 0.3328,
+      "step": 6450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773224036422512e-06,
+      "loss": 0.2712,
+      "step": 6451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770468814316262e-06,
+      "loss": 0.2569,
+      "step": 6452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.767713686976195e-06,
+      "loss": 0.2609,
+      "step": 6453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.764958654614673e-06,
+      "loss": 0.2792,
+      "step": 6454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.762203717444034e-06,
+      "loss": 0.2847,
+      "step": 6455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.759448875676616e-06,
+      "loss": 0.2769,
+      "step": 6456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75669412952475e-06,
+      "loss": 0.2563,
+      "step": 6457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.753939479200758e-06,
+      "loss": 0.2811,
+      "step": 6458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.751184924916954e-06,
+      "loss": 0.2651,
+      "step": 6459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.748430466885643e-06,
+      "loss": 0.2924,
+      "step": 6460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745676105319126e-06,
+      "loss": 0.3098,
+      "step": 6461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.742921840429702e-06,
+      "loss": 0.3028,
+      "step": 6462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.740167672429646e-06,
+      "loss": 0.2536,
+      "step": 6463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737413601531245e-06,
+      "loss": 0.2317,
+      "step": 6464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734659627946765e-06,
+      "loss": 0.2729,
+      "step": 6465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731905751888466e-06,
+      "loss": 0.292,
+      "step": 6466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.729151973568608e-06,
+      "loss": 0.2749,
+      "step": 6467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.726398293199434e-06,
+      "loss": 0.2852,
+      "step": 6468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723644710993189e-06,
+      "loss": 0.5852,
+      "step": 6469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.720891227162099e-06,
+      "loss": 0.2871,
+      "step": 6470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718137841918396e-06,
+      "loss": 0.2972,
+      "step": 6471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715384555474297e-06,
+      "loss": 0.2758,
+      "step": 6472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712631368042006e-06,
+      "loss": 0.2944,
+      "step": 6473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709878279833725e-06,
+      "loss": 0.2717,
+      "step": 6474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.707125291061656e-06,
+      "loss": 0.278,
+      "step": 6475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.704372401937982e-06,
+      "loss": 0.289,
+      "step": 6476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701619612674877e-06,
+      "loss": 0.28,
+      "step": 6477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698866923484521e-06,
+      "loss": 0.3257,
+      "step": 6478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69611433457907e-06,
+      "loss": 0.2626,
+      "step": 6479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693361846170687e-06,
+      "loss": 0.2358,
+      "step": 6480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.690609458471512e-06,
+      "loss": 0.3358,
+      "step": 6481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687857171693693e-06,
+      "loss": 0.5938,
+      "step": 6482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.685104986049362e-06,
+      "loss": 0.2931,
+      "step": 6483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682352901750636e-06,
+      "loss": 0.2775,
+      "step": 6484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679600919009642e-06,
+      "loss": 0.2616,
+      "step": 6485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676849038038483e-06,
+      "loss": 0.2341,
+      "step": 6486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674097259049263e-06,
+      "loss": 0.2853,
+      "step": 6487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.671345582254072e-06,
+      "loss": 0.2344,
+      "step": 6488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.668594007865003e-06,
+      "loss": 0.2463,
+      "step": 6489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66584253609413e-06,
+      "loss": 0.5797,
+      "step": 6490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663091167153516e-06,
+      "loss": 0.2997,
+      "step": 6491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660339901255234e-06,
+      "loss": 0.3104,
+      "step": 6492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65758873861133e-06,
+      "loss": 0.2821,
+      "step": 6493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654837679433852e-06,
+      "loss": 0.2617,
+      "step": 6494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652086723934841e-06,
+      "loss": 0.291,
+      "step": 6495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649335872326328e-06,
+      "loss": 0.275,
+      "step": 6496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64658512482033e-06,
+      "loss": 0.2801,
+      "step": 6497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643834481628861e-06,
+      "loss": 0.2419,
+      "step": 6498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641083942963929e-06,
+      "loss": 0.2874,
+      "step": 6499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638333509037537e-06,
+      "loss": 0.2811,
+      "step": 6500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.635583180061664e-06,
+      "loss": 0.2966,
+      "step": 6501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6328329562483e-06,
+      "loss": 0.3436,
+      "step": 6502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.630082837809419e-06,
+      "loss": 0.3058,
+      "step": 6503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62733282495698e-06,
+      "loss": 0.236,
+      "step": 6504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624582917902947e-06,
+      "loss": 0.2879,
+      "step": 6505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.621833116859264e-06,
+      "loss": 0.2872,
+      "step": 6506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.619083422037878e-06,
+      "loss": 0.3298,
+      "step": 6507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616333833650714e-06,
+      "loss": 0.2851,
+      "step": 6508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613584351909705e-06,
+      "loss": 0.3307,
+      "step": 6509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.610834977026765e-06,
+      "loss": 0.2833,
+      "step": 6510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608085709213797e-06,
+      "loss": 0.2859,
+      "step": 6511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.605336548682704e-06,
+      "loss": 0.3217,
+      "step": 6512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602587495645382e-06,
+      "loss": 0.322,
+      "step": 6513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599838550313714e-06,
+      "loss": 0.2433,
+      "step": 6514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597089712899564e-06,
+      "loss": 0.3004,
+      "step": 6515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.594340983614812e-06,
+      "loss": 0.3048,
+      "step": 6516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591592362671311e-06,
+      "loss": 0.2905,
+      "step": 6517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588843850280911e-06,
+      "loss": 0.2921,
+      "step": 6518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586095446655452e-06,
+      "loss": 0.2819,
+      "step": 6519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583347152006774e-06,
+      "loss": 0.2658,
+      "step": 6520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.580598966546697e-06,
+      "loss": 0.2552,
+      "step": 6521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577850890487035e-06,
+      "loss": 0.2912,
+      "step": 6522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.575102924039602e-06,
+      "loss": 0.2716,
+      "step": 6523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.572355067416194e-06,
+      "loss": 0.6274,
+      "step": 6524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.569607320828604e-06,
+      "loss": 0.2807,
+      "step": 6525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.566859684488611e-06,
+      "loss": 0.2546,
+      "step": 6526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.564112158607996e-06,
+      "loss": 0.2858,
+      "step": 6527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.561364743398521e-06,
+      "loss": 0.2894,
+      "step": 6528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558617439071938e-06,
+      "loss": 0.2871,
+      "step": 6529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.555870245840005e-06,
+      "loss": 0.3452,
+      "step": 6530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.553123163914456e-06,
+      "loss": 0.3228,
+      "step": 6531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550376193507023e-06,
+      "loss": 0.2663,
+      "step": 6532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547629334829434e-06,
+      "loss": 0.2761,
+      "step": 6533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.544882588093399e-06,
+      "loss": 0.2941,
+      "step": 6534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542135953510625e-06,
+      "loss": 0.2724,
+      "step": 6535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539389431292803e-06,
+      "loss": 0.2615,
+      "step": 6536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.536643021651629e-06,
+      "loss": 0.2693,
+      "step": 6537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533896724798784e-06,
+      "loss": 0.6083,
+      "step": 6538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.531150540945929e-06,
+      "loss": 0.2809,
+      "step": 6539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528404470304736e-06,
+      "loss": 0.2509,
+      "step": 6540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525658513086857e-06,
+      "loss": 0.3068,
+      "step": 6541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522912669503932e-06,
+      "loss": 0.2273,
+      "step": 6542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5201669397676e-06,
+      "loss": 0.2704,
+      "step": 6543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.517421324089488e-06,
+      "loss": 0.301,
+      "step": 6544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514675822681218e-06,
+      "loss": 0.314,
+      "step": 6545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511930435754391e-06,
+      "loss": 0.2621,
+      "step": 6546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.509185163520617e-06,
+      "loss": 0.2892,
+      "step": 6547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506440006191485e-06,
+      "loss": 0.2894,
+      "step": 6548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503694963978576e-06,
+      "loss": 0.3623,
+      "step": 6549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500950037093462e-06,
+      "loss": 0.3357,
+      "step": 6550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498205225747717e-06,
+      "loss": 0.6342,
+      "step": 6551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.495460530152893e-06,
+      "loss": 0.2982,
+      "step": 6552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.492715950520534e-06,
+      "loss": 0.2757,
+      "step": 6553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489971487062184e-06,
+      "loss": 0.2718,
+      "step": 6554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.487227139989372e-06,
+      "loss": 0.318,
+      "step": 6555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.484482909513613e-06,
+      "loss": 0.2923,
+      "step": 6556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.481738795846424e-06,
+      "loss": 0.2532,
+      "step": 6557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478994799199308e-06,
+      "loss": 0.2848,
+      "step": 6558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476250919783759e-06,
+      "loss": 0.2385,
+      "step": 6559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.473507157811254e-06,
+      "loss": 0.31,
+      "step": 6560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470763513493281e-06,
+      "loss": 0.3047,
+      "step": 6561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.468019987041298e-06,
+      "loss": 0.2968,
+      "step": 6562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465276578666766e-06,
+      "loss": 0.2969,
+      "step": 6563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462533288581128e-06,
+      "loss": 0.2549,
+      "step": 6564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45979011699583e-06,
+      "loss": 0.2863,
+      "step": 6565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4570470641223e-06,
+      "loss": 0.2489,
+      "step": 6566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454304130171956e-06,
+      "loss": 0.2518,
+      "step": 6567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451561315356216e-06,
+      "loss": 0.5439,
+      "step": 6568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448818619886477e-06,
+      "loss": 0.2722,
+      "step": 6569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44607604397413e-06,
+      "loss": 0.2662,
+      "step": 6570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443333587830568e-06,
+      "loss": 0.2607,
+      "step": 6571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.440591251667164e-06,
+      "loss": 0.2697,
+      "step": 6572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437849035695278e-06,
+      "loss": 0.2972,
+      "step": 6573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.435106940126266e-06,
+      "loss": 0.303,
+      "step": 6574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43236496517148e-06,
+      "loss": 0.2728,
+      "step": 6575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42962311104226e-06,
+      "loss": 0.3177,
+      "step": 6576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426881377949927e-06,
+      "loss": 0.2699,
+      "step": 6577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424139766105808e-06,
+      "loss": 0.2378,
+      "step": 6578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.421398275721208e-06,
+      "loss": 0.278,
+      "step": 6579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41865690700743e-06,
+      "loss": 0.2327,
+      "step": 6580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415915660175763e-06,
+      "loss": 0.2849,
+      "step": 6581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413174535437486e-06,
+      "loss": 0.2592,
+      "step": 6582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410433533003881e-06,
+      "loss": 0.285,
+      "step": 6583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4076926530862e-06,
+      "loss": 0.2789,
+      "step": 6584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.404951895895706e-06,
+      "loss": 0.3026,
+      "step": 6585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.402211261643638e-06,
+      "loss": 0.2661,
+      "step": 6586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.399470750541228e-06,
+      "loss": 0.2972,
+      "step": 6587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.396730362799704e-06,
+      "loss": 0.302,
+      "step": 6588
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.393990098630284e-06,
+      "loss": 0.2769,
+      "step": 6589
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391249958244173e-06,
+      "loss": 0.2604,
+      "step": 6590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.388509941852562e-06,
+      "loss": 0.2522,
+      "step": 6591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385770049666646e-06,
+      "loss": 0.2576,
+      "step": 6592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383030281897598e-06,
+      "loss": 0.2767,
+      "step": 6593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380290638756584e-06,
+      "loss": 0.3027,
+      "step": 6594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377551120454762e-06,
+      "loss": 0.2668,
+      "step": 6595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374811727203288e-06,
+      "loss": 0.2876,
+      "step": 6596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.372072459213296e-06,
+      "loss": 0.252,
+      "step": 6597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369333316695909e-06,
+      "loss": 0.278,
+      "step": 6598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366594299862258e-06,
+      "loss": 0.2847,
+      "step": 6599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363855408923446e-06,
+      "loss": 0.2445,
+      "step": 6600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361116644090576e-06,
+      "loss": 0.2927,
+      "step": 6601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358378005574731e-06,
+      "loss": 0.2589,
+      "step": 6602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355639493587005e-06,
+      "loss": 0.2673,
+      "step": 6603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.352901108338462e-06,
+      "loss": 0.251,
+      "step": 6604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350162850040158e-06,
+      "loss": 0.2782,
+      "step": 6605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347424718903152e-06,
+      "loss": 0.2795,
+      "step": 6606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344686715138482e-06,
+      "loss": 0.2858,
+      "step": 6607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.341948838957185e-06,
+      "loss": 0.2376,
+      "step": 6608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339211090570272e-06,
+      "loss": 0.2654,
+      "step": 6609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336473470188767e-06,
+      "loss": 0.2508,
+      "step": 6610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333735978023667e-06,
+      "loss": 0.254,
+      "step": 6611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33099861428596e-06,
+      "loss": 0.276,
+      "step": 6612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.328261379186636e-06,
+      "loss": 0.3005,
+      "step": 6613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325524272936668e-06,
+      "loss": 0.3035,
+      "step": 6614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.322787295747007e-06,
+      "loss": 0.2884,
+      "step": 6615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.320050447828622e-06,
+      "loss": 0.2643,
+      "step": 6616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.317313729392446e-06,
+      "loss": 0.2758,
+      "step": 6617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31457714064941e-06,
+      "loss": 0.2894,
+      "step": 6618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311840681810441e-06,
+      "loss": 0.3061,
+      "step": 6619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.309104353086452e-06,
+      "loss": 0.2903,
+      "step": 6620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.306368154688347e-06,
+      "loss": 0.2937,
+      "step": 6621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30363208682701e-06,
+      "loss": 0.3031,
+      "step": 6622
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.300896149713334e-06,
+      "loss": 0.3235,
+      "step": 6623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298160343558188e-06,
+      "loss": 0.329,
+      "step": 6624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.295424668572432e-06,
+      "loss": 0.2549,
+      "step": 6625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.292689124966917e-06,
+      "loss": 0.3129,
+      "step": 6626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.289953712952494e-06,
+      "loss": 0.2688,
+      "step": 6627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287218432739987e-06,
+      "loss": 0.2543,
+      "step": 6628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284483284540217e-06,
+      "loss": 0.2755,
+      "step": 6629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.281748268564002e-06,
+      "loss": 0.3182,
+      "step": 6630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279013385022142e-06,
+      "loss": 0.3085,
+      "step": 6631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.276278634125424e-06,
+      "loss": 0.2692,
+      "step": 6632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.273544016084629e-06,
+      "loss": 0.2648,
+      "step": 6633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270809531110536e-06,
+      "loss": 0.2913,
+      "step": 6634
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.268075179413899e-06,
+      "loss": 0.2413,
+      "step": 6635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.265340961205467e-06,
+      "loss": 0.2721,
+      "step": 6636
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262606876695984e-06,
+      "loss": 0.2873,
+      "step": 6637
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.259872926096177e-06,
+      "loss": 0.2712,
+      "step": 6638
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.257139109616769e-06,
+      "loss": 0.2597,
+      "step": 6639
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254405427468464e-06,
+      "loss": 0.2756,
+      "step": 6640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.251671879861966e-06,
+      "loss": 0.2834,
+      "step": 6641
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24893846700796e-06,
+      "loss": 0.2936,
+      "step": 6642
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246205189117122e-06,
+      "loss": 0.3012,
+      "step": 6643
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.243472046400126e-06,
+      "loss": 0.274,
+      "step": 6644
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240739039067623e-06,
+      "loss": 0.274,
+      "step": 6645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238006167330266e-06,
+      "loss": 0.2891,
+      "step": 6646
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235273431398681e-06,
+      "loss": 0.2826,
+      "step": 6647
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232540831483505e-06,
+      "loss": 0.2602,
+      "step": 6648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229808367795349e-06,
+      "loss": 0.3093,
+      "step": 6649
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227076040544813e-06,
+      "loss": 0.2997,
+      "step": 6650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224343849942496e-06,
+      "loss": 0.2674,
+      "step": 6651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221611796198984e-06,
+      "loss": 0.3036,
+      "step": 6652
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218879879524844e-06,
+      "loss": 0.2232,
+      "step": 6653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216148100130647e-06,
+      "loss": 0.5638,
+      "step": 6654
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213416458226939e-06,
+      "loss": 0.3172,
+      "step": 6655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210684954024261e-06,
+      "loss": 0.244,
+      "step": 6656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207953587733145e-06,
+      "loss": 0.2872,
+      "step": 6657
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205222359564113e-06,
+      "loss": 0.2779,
+      "step": 6658
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202491269727674e-06,
+      "loss": 0.3121,
+      "step": 6659
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.199760318434323e-06,
+      "loss": 0.2393,
+      "step": 6660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197029505894553e-06,
+      "loss": 0.2648,
+      "step": 6661
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194298832318843e-06,
+      "loss": 0.3467,
+      "step": 6662
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191568297917654e-06,
+      "loss": 0.2256,
+      "step": 6663
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188837902901441e-06,
+      "loss": 0.3017,
+      "step": 6664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186107647480659e-06,
+      "loss": 0.3411,
+      "step": 6665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183377531865737e-06,
+      "loss": 0.256,
+      "step": 6666
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180647556267093e-06,
+      "loss": 0.2773,
+      "step": 6667
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.177917720895152e-06,
+      "loss": 0.2856,
+      "step": 6668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17518802596031e-06,
+      "loss": 0.3012,
+      "step": 6669
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172458471672953e-06,
+      "loss": 0.2455,
+      "step": 6670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169729058243468e-06,
+      "loss": 0.2645,
+      "step": 6671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.166999785882226e-06,
+      "loss": 0.2902,
+      "step": 6672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164270654799584e-06,
+      "loss": 0.2939,
+      "step": 6673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161541665205885e-06,
+      "loss": 0.2874,
+      "step": 6674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158812817311474e-06,
+      "loss": 0.3354,
+      "step": 6675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.156084111326673e-06,
+      "loss": 0.3188,
+      "step": 6676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1533555474618e-06,
+      "loss": 0.3161,
+      "step": 6677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15062712592715e-06,
+      "loss": 0.2739,
+      "step": 6678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14789884693303e-06,
+      "loss": 0.2619,
+      "step": 6679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.145170710689712e-06,
+      "loss": 0.2914,
+      "step": 6680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142442717407469e-06,
+      "loss": 0.2628,
+      "step": 6681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139714867296567e-06,
+      "loss": 0.3356,
+      "step": 6682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13698716056725e-06,
+      "loss": 0.2923,
+      "step": 6683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134259597429757e-06,
+      "loss": 0.3265,
+      "step": 6684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13153217809431e-06,
+      "loss": 0.2641,
+      "step": 6685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.128804902771137e-06,
+      "loss": 0.275,
+      "step": 6686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126077771670438e-06,
+      "loss": 0.3011,
+      "step": 6687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123350785002398e-06,
+      "loss": 0.2404,
+      "step": 6688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.120623942977209e-06,
+      "loss": 0.2519,
+      "step": 6689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.117897245805044e-06,
+      "loss": 0.2847,
+      "step": 6690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115170693696058e-06,
+      "loss": 0.3049,
+      "step": 6691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112444286860397e-06,
+      "loss": 0.2795,
+      "step": 6692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109718025508208e-06,
+      "loss": 0.3232,
+      "step": 6693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106991909849613e-06,
+      "loss": 0.2578,
+      "step": 6694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.104265940094726e-06,
+      "loss": 0.2699,
+      "step": 6695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.101540116453655e-06,
+      "loss": 0.6116,
+      "step": 6696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098814439136492e-06,
+      "loss": 0.2864,
+      "step": 6697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096088908353316e-06,
+      "loss": 0.3403,
+      "step": 6698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.093363524314202e-06,
+      "loss": 0.2563,
+      "step": 6699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090638287229207e-06,
+      "loss": 0.3217,
+      "step": 6700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.087913197308376e-06,
+      "loss": 0.2843,
+      "step": 6701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085188254761744e-06,
+      "loss": 0.2822,
+      "step": 6702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082463459799346e-06,
+      "loss": 0.2934,
+      "step": 6703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07973881263119e-06,
+      "loss": 0.2802,
+      "step": 6704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077014313467274e-06,
+      "loss": 0.27,
+      "step": 6705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074289962517597e-06,
+      "loss": 0.2915,
+      "step": 6706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071565759992133e-06,
+      "loss": 0.2678,
+      "step": 6707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068841706100851e-06,
+      "loss": 0.3052,
+      "step": 6708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.066117801053706e-06,
+      "loss": 0.2889,
+      "step": 6709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063394045060648e-06,
+      "loss": 0.2965,
+      "step": 6710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06067043833161e-06,
+      "loss": 0.2789,
+      "step": 6711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057946981076506e-06,
+      "loss": 0.2478,
+      "step": 6712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055223673505258e-06,
+      "loss": 0.2797,
+      "step": 6713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.052500515827759e-06,
+      "loss": 0.2854,
+      "step": 6714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049777508253898e-06,
+      "loss": 0.2609,
+      "step": 6715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.047054650993545e-06,
+      "loss": 0.3015,
+      "step": 6716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.044331944256576e-06,
+      "loss": 0.261,
+      "step": 6717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041609388252836e-06,
+      "loss": 0.2498,
+      "step": 6718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.038886983192164e-06,
+      "loss": 0.2546,
+      "step": 6719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036164729284398e-06,
+      "loss": 0.2996,
+      "step": 6720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033442626739347e-06,
+      "loss": 0.3373,
+      "step": 6721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030720675766825e-06,
+      "loss": 0.2789,
+      "step": 6722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02799887657662e-06,
+      "loss": 0.261,
+      "step": 6723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.025277229378519e-06,
+      "loss": 0.2935,
+      "step": 6724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022555734382294e-06,
+      "loss": 0.2772,
+      "step": 6725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.019834391797696e-06,
+      "loss": 0.2841,
+      "step": 6726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.017113201834482e-06,
+      "loss": 0.2551,
+      "step": 6727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014392164702387e-06,
+      "loss": 0.305,
+      "step": 6728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011671280611132e-06,
+      "loss": 0.2997,
+      "step": 6729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008950549770426e-06,
+      "loss": 0.2642,
+      "step": 6730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.006229972389979e-06,
+      "loss": 0.2518,
+      "step": 6731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.003509548679471e-06,
+      "loss": 0.3026,
+      "step": 6732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000789278848582e-06,
+      "loss": 0.298,
+      "step": 6733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.998069163106977e-06,
+      "loss": 0.2606,
+      "step": 6734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.995349201664311e-06,
+      "loss": 0.2728,
+      "step": 6735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992629394730225e-06,
+      "loss": 0.3022,
+      "step": 6736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98990974251434e-06,
+      "loss": 0.2748,
+      "step": 6737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987190245226285e-06,
+      "loss": 0.2649,
+      "step": 6738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984470903075658e-06,
+      "loss": 0.2667,
+      "step": 6739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981751716272054e-06,
+      "loss": 0.2701,
+      "step": 6740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.979032685025057e-06,
+      "loss": 0.288,
+      "step": 6741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976313809544237e-06,
+      "loss": 0.3501,
+      "step": 6742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973595090039144e-06,
+      "loss": 0.2673,
+      "step": 6743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.970876526719333e-06,
+      "loss": 0.274,
+      "step": 6744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968158119794334e-06,
+      "loss": 0.2609,
+      "step": 6745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.965439869473664e-06,
+      "loss": 0.3143,
+      "step": 6746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962721775966836e-06,
+      "loss": 0.2941,
+      "step": 6747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960003839483348e-06,
+      "loss": 0.2725,
+      "step": 6748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957286060232687e-06,
+      "loss": 0.6223,
+      "step": 6749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954568438424315e-06,
+      "loss": 0.2475,
+      "step": 6750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95185097426771e-06,
+      "loss": 0.2858,
+      "step": 6751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.949133667972307e-06,
+      "loss": 0.5742,
+      "step": 6752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946416519747549e-06,
+      "loss": 0.6016,
+      "step": 6753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943699529802854e-06,
+      "loss": 0.2668,
+      "step": 6754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940982698347646e-06,
+      "loss": 0.2445,
+      "step": 6755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938266025591315e-06,
+      "loss": 0.2624,
+      "step": 6756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935549511743249e-06,
+      "loss": 0.2715,
+      "step": 6757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932833157012829e-06,
+      "loss": 0.3148,
+      "step": 6758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.930116961609413e-06,
+      "loss": 0.2615,
+      "step": 6759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927400925742357e-06,
+      "loss": 0.2903,
+      "step": 6760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92468504962099e-06,
+      "loss": 0.2978,
+      "step": 6761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.921969333454652e-06,
+      "loss": 0.2657,
+      "step": 6762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919253777452649e-06,
+      "loss": 0.2697,
+      "step": 6763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.916538381824279e-06,
+      "loss": 0.2648,
+      "step": 6764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913823146778838e-06,
+      "loss": 0.2726,
+      "step": 6765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.911108072525603e-06,
+      "loss": 0.2791,
+      "step": 6766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908393159273835e-06,
+      "loss": 0.3007,
+      "step": 6767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905678407232785e-06,
+      "loss": 0.3189,
+      "step": 6768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.902963816611699e-06,
+      "loss": 0.2844,
+      "step": 6769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900249387619797e-06,
+      "loss": 0.2731,
+      "step": 6770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897535120466297e-06,
+      "loss": 0.312,
+      "step": 6771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894821015360404e-06,
+      "loss": 0.275,
+      "step": 6772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892107072511303e-06,
+      "loss": 0.2603,
+      "step": 6773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889393292128177e-06,
+      "loss": 0.2852,
+      "step": 6774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886679674420182e-06,
+      "loss": 0.2716,
+      "step": 6775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88396621959648e-06,
+      "loss": 0.2916,
+      "step": 6776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881252927866203e-06,
+      "loss": 0.2948,
+      "step": 6777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.878539799438479e-06,
+      "loss": 0.2668,
+      "step": 6778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87582683452243e-06,
+      "loss": 0.2807,
+      "step": 6779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873114033327153e-06,
+      "loss": 0.289,
+      "step": 6780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.870401396061731e-06,
+      "loss": 0.3131,
+      "step": 6781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.867688922935253e-06,
+      "loss": 0.2866,
+      "step": 6782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864976614156776e-06,
+      "loss": 0.3062,
+      "step": 6783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86226446993535e-06,
+      "loss": 0.27,
+      "step": 6784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859552490480014e-06,
+      "loss": 0.2682,
+      "step": 6785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856840675999799e-06,
+      "loss": 0.3036,
+      "step": 6786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.854129026703716e-06,
+      "loss": 0.3032,
+      "step": 6787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85141754280076e-06,
+      "loss": 0.2999,
+      "step": 6788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848706224499928e-06,
+      "loss": 0.2505,
+      "step": 6789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.845995072010188e-06,
+      "loss": 0.2905,
+      "step": 6790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.843284085540505e-06,
+      "loss": 0.308,
+      "step": 6791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840573265299827e-06,
+      "loss": 0.283,
+      "step": 6792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.837862611497094e-06,
+      "loss": 0.2396,
+      "step": 6793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835152124341228e-06,
+      "loss": 0.2682,
+      "step": 6794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832441804041135e-06,
+      "loss": 0.3159,
+      "step": 6795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829731650805722e-06,
+      "loss": 0.2722,
+      "step": 6796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.827021664843867e-06,
+      "loss": 0.2944,
+      "step": 6797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.824311846364448e-06,
+      "loss": 0.2573,
+      "step": 6798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.821602195576316e-06,
+      "loss": 0.2988,
+      "step": 6799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818892712688328e-06,
+      "loss": 0.2433,
+      "step": 6800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816183397909312e-06,
+      "loss": 0.3422,
+      "step": 6801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.813474251448086e-06,
+      "loss": 0.3299,
+      "step": 6802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.810765273513463e-06,
+      "loss": 0.2822,
+      "step": 6803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.808056464314236e-06,
+      "loss": 0.2115,
+      "step": 6804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805347824059188e-06,
+      "loss": 0.3323,
+      "step": 6805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80263935295708e-06,
+      "loss": 0.2448,
+      "step": 6806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799931051216677e-06,
+      "loss": 0.3411,
+      "step": 6807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.797222919046717e-06,
+      "loss": 0.3226,
+      "step": 6808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.794514956655929e-06,
+      "loss": 0.265,
+      "step": 6809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79180716425303e-06,
+      "loss": 0.2964,
+      "step": 6810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789099542046727e-06,
+      "loss": 0.2639,
+      "step": 6811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786392090245708e-06,
+      "loss": 0.2372,
+      "step": 6812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783684809058642e-06,
+      "loss": 0.2568,
+      "step": 6813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780977698694206e-06,
+      "loss": 0.2457,
+      "step": 6814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778270759361044e-06,
+      "loss": 0.2863,
+      "step": 6815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.775563991267789e-06,
+      "loss": 0.3049,
+      "step": 6816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772857394623074e-06,
+      "loss": 0.2921,
+      "step": 6817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.770150969635509e-06,
+      "loss": 0.2805,
+      "step": 6818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767444716513686e-06,
+      "loss": 0.3014,
+      "step": 6819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.764738635466192e-06,
+      "loss": 0.2441,
+      "step": 6820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.762032726701602e-06,
+      "loss": 0.2899,
+      "step": 6821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.759326990428468e-06,
+      "loss": 0.2731,
+      "step": 6822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756621426855337e-06,
+      "loss": 0.3071,
+      "step": 6823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753916036190747e-06,
+      "loss": 0.2357,
+      "step": 6824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.751210818643209e-06,
+      "loss": 0.2946,
+      "step": 6825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748505774421227e-06,
+      "loss": 0.2566,
+      "step": 6826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745800903733298e-06,
+      "loss": 0.2599,
+      "step": 6827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.743096206787894e-06,
+      "loss": 0.2734,
+      "step": 6828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740391683793486e-06,
+      "loss": 0.2442,
+      "step": 6829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737687334958518e-06,
+      "loss": 0.2748,
+      "step": 6830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734983160491435e-06,
+      "loss": 0.242,
+      "step": 6831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73227916060066e-06,
+      "loss": 0.2683,
+      "step": 6832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729575335494595e-06,
+      "loss": 0.2817,
+      "step": 6833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726871685381652e-06,
+      "loss": 0.29,
+      "step": 6834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.724168210470203e-06,
+      "loss": 0.2422,
+      "step": 6835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.721464910968628e-06,
+      "loss": 0.2679,
+      "step": 6836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.718761787085271e-06,
+      "loss": 0.2892,
+      "step": 6837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71605883902849e-06,
+      "loss": 0.2853,
+      "step": 6838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713356067006609e-06,
+      "loss": 0.2867,
+      "step": 6839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710653471227939e-06,
+      "loss": 0.3144,
+      "step": 6840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70795105190079e-06,
+      "loss": 0.2573,
+      "step": 6841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70524880923345e-06,
+      "loss": 0.2829,
+      "step": 6842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702546743434193e-06,
+      "loss": 0.3174,
+      "step": 6843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.699844854711276e-06,
+      "loss": 0.2766,
+      "step": 6844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697143143272959e-06,
+      "loss": 0.261,
+      "step": 6845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694441609327465e-06,
+      "loss": 0.2397,
+      "step": 6846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.691740253083022e-06,
+      "loss": 0.264,
+      "step": 6847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689039074747832e-06,
+      "loss": 0.2981,
+      "step": 6848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686338074530095e-06,
+      "loss": 0.2701,
+      "step": 6849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.683637252637988e-06,
+      "loss": 0.3522,
+      "step": 6850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68093660927967e-06,
+      "loss": 0.2596,
+      "step": 6851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678236144663304e-06,
+      "loss": 0.2336,
+      "step": 6852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675535858997024e-06,
+      "loss": 0.2643,
+      "step": 6853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67283575248895e-06,
+      "loss": 0.2522,
+      "step": 6854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670135825347202e-06,
+      "loss": 0.2463,
+      "step": 6855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667436077779872e-06,
+      "loss": 0.2444,
+      "step": 6856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664736509995042e-06,
+      "loss": 0.2952,
+      "step": 6857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662037122200783e-06,
+      "loss": 0.2639,
+      "step": 6858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659337914605152e-06,
+      "loss": 0.282,
+      "step": 6859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.656638887416186e-06,
+      "loss": 0.2481,
+      "step": 6860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653940040841917e-06,
+      "loss": 0.2653,
+      "step": 6861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651241375090358e-06,
+      "loss": 0.2657,
+      "step": 6862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64854289036951e-06,
+      "loss": 0.2699,
+      "step": 6863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.645844586887353e-06,
+      "loss": 0.2776,
+      "step": 6864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643146464851867e-06,
+      "loss": 0.2554,
+      "step": 6865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640448524471002e-06,
+      "loss": 0.3297,
+      "step": 6866
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63775076595271e-06,
+      "loss": 0.2556,
+      "step": 6867
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635053189504913e-06,
+      "loss": 0.2626,
+      "step": 6868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632355795335533e-06,
+      "loss": 0.2658,
+      "step": 6869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629658583652471e-06,
+      "loss": 0.2875,
+      "step": 6870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.626961554663609e-06,
+      "loss": 0.241,
+      "step": 6871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.624264708576827e-06,
+      "loss": 0.6245,
+      "step": 6872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.621568045599983e-06,
+      "loss": 0.2819,
+      "step": 6873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6188715659409216e-06,
+      "loss": 0.2404,
+      "step": 6874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.616175269807472e-06,
+      "loss": 0.3386,
+      "step": 6875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613479157407457e-06,
+      "loss": 0.2772,
+      "step": 6876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6107832289486775e-06,
+      "loss": 0.2893,
+      "step": 6877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608087484638915e-06,
+      "loss": 0.2455,
+      "step": 6878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.605391924685954e-06,
+      "loss": 0.2654,
+      "step": 6879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6026965492975535e-06,
+      "loss": 0.2864,
+      "step": 6880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.600001358681457e-06,
+      "loss": 0.2517,
+      "step": 6881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597306353045393e-06,
+      "loss": 0.2388,
+      "step": 6882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.594611532597087e-06,
+      "loss": 0.3116,
+      "step": 6883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591916897544238e-06,
+      "loss": 0.2532,
+      "step": 6884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.589222448094535e-06,
+      "loss": 0.2809,
+      "step": 6885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.586528184455653e-06,
+      "loss": 0.2706,
+      "step": 6886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.583834106835256e-06,
+      "loss": 0.6118,
+      "step": 6887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.581140215440987e-06,
+      "loss": 0.248,
+      "step": 6888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.578446510480475e-06,
+      "loss": 0.2797,
+      "step": 6889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.575752992161345e-06,
+      "loss": 0.2577,
+      "step": 6890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573059660691192e-06,
+      "loss": 0.2634,
+      "step": 6891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.570366516277607e-06,
+      "loss": 0.288,
+      "step": 6892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567673559128171e-06,
+      "loss": 0.2911,
+      "step": 6893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.564980789450438e-06,
+      "loss": 0.2475,
+      "step": 6894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5622882074519544e-06,
+      "loss": 0.3536,
+      "step": 6895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559595813340246e-06,
+      "loss": 0.2318,
+      "step": 6896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556903607322839e-06,
+      "loss": 0.285,
+      "step": 6897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.554211589607227e-06,
+      "loss": 0.2857,
+      "step": 6898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551519760400898e-06,
+      "loss": 0.2746,
+      "step": 6899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.548828119911333e-06,
+      "loss": 0.2625,
+      "step": 6900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.546136668345985e-06,
+      "loss": 0.2668,
+      "step": 6901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.543445405912298e-06,
+      "loss": 0.2488,
+      "step": 6902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.540754332817695e-06,
+      "loss": 0.2884,
+      "step": 6903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.538063449269599e-06,
+      "loss": 0.3168,
+      "step": 6904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.535372755475411e-06,
+      "loss": 0.2642,
+      "step": 6905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.532682251642508e-06,
+      "loss": 0.2701,
+      "step": 6906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5299919379782695e-06,
+      "loss": 0.3134,
+      "step": 6907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.527301814690048e-06,
+      "loss": 0.2495,
+      "step": 6908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524611881985181e-06,
+      "loss": 0.3099,
+      "step": 6909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.521922140071003e-06,
+      "loss": 0.2658,
+      "step": 6910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519232589154819e-06,
+      "loss": 0.2795,
+      "step": 6911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516543229443931e-06,
+      "loss": 0.2455,
+      "step": 6912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513854061145617e-06,
+      "loss": 0.265,
+      "step": 6913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5111650844671515e-06,
+      "loss": 0.3412,
+      "step": 6914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5084762996157835e-06,
+      "loss": 0.2528,
+      "step": 6915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5057877067987464e-06,
+      "loss": 0.2936,
+      "step": 6916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.503099306223271e-06,
+      "loss": 0.3046,
+      "step": 6917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5004110980965664e-06,
+      "loss": 0.2598,
+      "step": 6918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4977230826258226e-06,
+      "loss": 0.3021,
+      "step": 6919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.495035260018215e-06,
+      "loss": 0.3339,
+      "step": 6920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492347630480917e-06,
+      "loss": 0.3074,
+      "step": 6921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489660194221071e-06,
+      "loss": 0.2566,
+      "step": 6922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486972951445812e-06,
+      "loss": 0.2756,
+      "step": 6923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484285902362263e-06,
+      "loss": 0.2369,
+      "step": 6924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481599047177527e-06,
+      "loss": 0.2578,
+      "step": 6925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478912386098692e-06,
+      "loss": 0.3085,
+      "step": 6926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47622591933283e-06,
+      "loss": 0.2516,
+      "step": 6927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473539647087007e-06,
+      "loss": 0.2854,
+      "step": 6928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470853569568264e-06,
+      "loss": 0.2673,
+      "step": 6929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468167686983627e-06,
+      "loss": 0.2991,
+      "step": 6930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46548199954012e-06,
+      "loss": 0.2335,
+      "step": 6931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462796507444736e-06,
+      "loss": 0.2685,
+      "step": 6932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4601112109044615e-06,
+      "loss": 0.3016,
+      "step": 6933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4574261101262604e-06,
+      "loss": 0.3145,
+      "step": 6934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4547412053170955e-06,
+      "loss": 0.2549,
+      "step": 6935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4520564966839e-06,
+      "loss": 0.2838,
+      "step": 6936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449371984433598e-06,
+      "loss": 0.2552,
+      "step": 6937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446687668773105e-06,
+      "loss": 0.2852,
+      "step": 6938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44400354990931e-06,
+      "loss": 0.2657,
+      "step": 6939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4413196280490905e-06,
+      "loss": 0.2704,
+      "step": 6940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.43863590339931e-06,
+      "loss": 0.2847,
+      "step": 6941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435952376166818e-06,
+      "loss": 0.3054,
+      "step": 6942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.433269046558449e-06,
+      "loss": 0.3448,
+      "step": 6943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.430585914781017e-06,
+      "loss": 0.2815,
+      "step": 6944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.427902981041329e-06,
+      "loss": 0.2787,
+      "step": 6945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.425220245546172e-06,
+      "loss": 0.2469,
+      "step": 6946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4225377085023105e-06,
+      "loss": 0.2617,
+      "step": 6947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419855370116511e-06,
+      "loss": 0.3333,
+      "step": 6948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4171732305955095e-06,
+      "loss": 0.3027,
+      "step": 6949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4144912901460355e-06,
+      "loss": 0.2921,
+      "step": 6950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 0.289,
+      "step": 6951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4091280072884854e-06,
+      "loss": 0.2651,
+      "step": 6952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.406446665293789e-06,
+      "loss": 0.262,
+      "step": 6953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.403765523197365e-06,
+      "loss": 0.2734,
+      "step": 6954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.401084581205869e-06,
+      "loss": 0.3026,
+      "step": 6955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3984038395259315e-06,
+      "loss": 0.3073,
+      "step": 6956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.395723298364174e-06,
+      "loss": 0.3115,
+      "step": 6957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39304295792719e-06,
+      "loss": 0.2886,
+      "step": 6958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.390362818421579e-06,
+      "loss": 0.2398,
+      "step": 6959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.387682880053906e-06,
+      "loss": 0.2817,
+      "step": 6960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.385003143030727e-06,
+      "loss": 0.2376,
+      "step": 6961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.382323607558585e-06,
+      "loss": 0.2755,
+      "step": 6962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379644273844008e-06,
+      "loss": 0.2515,
+      "step": 6963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.376965142093502e-06,
+      "loss": 0.2986,
+      "step": 6964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.374286212513558e-06,
+      "loss": 0.3193,
+      "step": 6965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3716074853106635e-06,
+      "loss": 0.2745,
+      "step": 6966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.368928960691275e-06,
+      "loss": 0.2859,
+      "step": 6967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366250638861838e-06,
+      "loss": 0.2816,
+      "step": 6968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3635725200287936e-06,
+      "loss": 0.3256,
+      "step": 6969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3608946043985515e-06,
+      "loss": 0.2828,
+      "step": 6970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358216892177514e-06,
+      "loss": 0.2491,
+      "step": 6971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355539383572059e-06,
+      "loss": 0.239,
+      "step": 6972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3528620787885676e-06,
+      "loss": 0.2664,
+      "step": 6973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350184978033386e-06,
+      "loss": 0.2633,
+      "step": 6974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.347508081512848e-06,
+      "loss": 0.3063,
+      "step": 6975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344831389433287e-06,
+      "loss": 0.2521,
+      "step": 6976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342154902001003e-06,
+      "loss": 0.2835,
+      "step": 6977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.339478619422287e-06,
+      "loss": 0.2777,
+      "step": 6978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336802541903408e-06,
+      "loss": 0.3508,
+      "step": 6979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3341266696506304e-06,
+      "loss": 0.2777,
+      "step": 6980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3314510028702e-06,
+      "loss": 0.2792,
+      "step": 6981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328775541768336e-06,
+      "loss": 0.2884,
+      "step": 6982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32610028655126e-06,
+      "loss": 0.2721,
+      "step": 6983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3234252374251614e-06,
+      "loss": 0.2717,
+      "step": 6984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320750394596217e-06,
+      "loss": 0.2684,
+      "step": 6985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318075758270593e-06,
+      "loss": 0.2686,
+      "step": 6986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315401328654439e-06,
+      "loss": 0.2759,
+      "step": 6987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312727105953888e-06,
+      "loss": 0.3135,
+      "step": 6988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310053090375049e-06,
+      "loss": 0.3065,
+      "step": 6989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.30737928212403e-06,
+      "loss": 0.2645,
+      "step": 6990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3047056814069115e-06,
+      "loss": 0.2267,
+      "step": 6991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3020322884297565e-06,
+      "loss": 0.2621,
+      "step": 6992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.299359103398626e-06,
+      "loss": 0.2657,
+      "step": 6993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296686126519552e-06,
+      "loss": 0.241,
+      "step": 6994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.294013357998554e-06,
+      "loss": 0.3229,
+      "step": 6995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291340798041631e-06,
+      "loss": 0.2821,
+      "step": 6996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288668446854781e-06,
+      "loss": 0.3004,
+      "step": 6997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2859963046439665e-06,
+      "loss": 0.3038,
+      "step": 6998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.283324371615147e-06,
+      "loss": 0.2745,
+      "step": 6999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280652647974263e-06,
+      "loss": 0.2761,
+      "step": 7000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.277981133927236e-06,
+      "loss": 0.2981,
+      "step": 7001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.275309829679973e-06,
+      "loss": 0.2807,
+      "step": 7002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2726387354383625e-06,
+      "loss": 0.2733,
+      "step": 7003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269967851408286e-06,
+      "loss": 0.2708,
+      "step": 7004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.267297177795596e-06,
+      "loss": 0.285,
+      "step": 7005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264626714806135e-06,
+      "loss": 0.2902,
+      "step": 7006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.261956462645734e-06,
+      "loss": 0.2422,
+      "step": 7007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.259286421520201e-06,
+      "loss": 0.278,
+      "step": 7008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256616591635328e-06,
+      "loss": 0.2925,
+      "step": 7009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.253946973196888e-06,
+      "loss": 0.3397,
+      "step": 7010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251277566410651e-06,
+      "loss": 0.3224,
+      "step": 7011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.248608371482355e-06,
+      "loss": 0.2954,
+      "step": 7012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24593938861773e-06,
+      "loss": 0.2712,
+      "step": 7013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243270618022492e-06,
+      "loss": 0.2603,
+      "step": 7014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240602059902333e-06,
+      "loss": 0.3207,
+      "step": 7015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.237933714462932e-06,
+      "loss": 0.3207,
+      "step": 7016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23526558190995e-06,
+      "loss": 0.28,
+      "step": 7017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232597662449038e-06,
+      "loss": 0.2385,
+      "step": 7018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.229929956285826e-06,
+      "loss": 0.2546,
+      "step": 7019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.22726246362592e-06,
+      "loss": 0.2901,
+      "step": 7020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.224595184674928e-06,
+      "loss": 0.2338,
+      "step": 7021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221928119638426e-06,
+      "loss": 0.3267,
+      "step": 7022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2192612687219755e-06,
+      "loss": 0.2772,
+      "step": 7023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2165946321311254e-06,
+      "loss": 0.2852,
+      "step": 7024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213928210071408e-06,
+      "loss": 0.2764,
+      "step": 7025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211262002748341e-06,
+      "loss": 0.3029,
+      "step": 7026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2085960103674146e-06,
+      "loss": 0.2753,
+      "step": 7027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.205930233134117e-06,
+      "loss": 0.2527,
+      "step": 7028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.203264671253915e-06,
+      "loss": 0.2659,
+      "step": 7029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200599324932246e-06,
+      "loss": 0.312,
+      "step": 7030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1979341943745515e-06,
+      "loss": 0.3139,
+      "step": 7031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195269279786247e-06,
+      "loss": 0.3,
+      "step": 7032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192604581372727e-06,
+      "loss": 0.2886,
+      "step": 7033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.18994009933937e-06,
+      "loss": 0.2709,
+      "step": 7034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.187275833891549e-06,
+      "loss": 0.2518,
+      "step": 7035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1846117852346075e-06,
+      "loss": 0.2731,
+      "step": 7036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181947953573878e-06,
+      "loss": 0.2617,
+      "step": 7037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.179284339114676e-06,
+      "loss": 0.2572,
+      "step": 7038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1766209420623e-06,
+      "loss": 0.2774,
+      "step": 7039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.173957762622032e-06,
+      "loss": 0.2656,
+      "step": 7040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171294800999134e-06,
+      "loss": 0.2863,
+      "step": 7041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168632057398857e-06,
+      "loss": 0.2738,
+      "step": 7042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.16596953202643e-06,
+      "loss": 0.3048,
+      "step": 7043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1633072250870665e-06,
+      "loss": 0.2772,
+      "step": 7044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160645136785968e-06,
+      "loss": 0.2604,
+      "step": 7045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.157983267328314e-06,
+      "loss": 0.2914,
+      "step": 7046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155321616919267e-06,
+      "loss": 0.2443,
+      "step": 7047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.152660185763969e-06,
+      "loss": 0.283,
+      "step": 7048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.149998974067558e-06,
+      "loss": 0.2876,
+      "step": 7049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.147337982035143e-06,
+      "loss": 0.2795,
+      "step": 7050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144677209871819e-06,
+      "loss": 0.2578,
+      "step": 7051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142016657782671e-06,
+      "loss": 0.285,
+      "step": 7052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139356325972757e-06,
+      "loss": 0.2953,
+      "step": 7053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.136696214647123e-06,
+      "loss": 0.2733,
+      "step": 7054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134036324010791e-06,
+      "loss": 0.2737,
+      "step": 7055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1313766542687824e-06,
+      "loss": 0.2746,
+      "step": 7056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1287172056260875e-06,
+      "loss": 0.2905,
+      "step": 7057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.12605797828768e-06,
+      "loss": 0.2798,
+      "step": 7058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123398972458526e-06,
+      "loss": 0.3279,
+      "step": 7059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120740188343567e-06,
+      "loss": 0.2869,
+      "step": 7060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118081626147724e-06,
+      "loss": 0.2475,
+      "step": 7061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.11542328607591e-06,
+      "loss": 0.2903,
+      "step": 7062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112765168333016e-06,
+      "loss": 0.2856,
+      "step": 7063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110107273123919e-06,
+      "loss": 0.2883,
+      "step": 7064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107449600653468e-06,
+      "loss": 0.2548,
+      "step": 7065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.104792151126515e-06,
+      "loss": 0.2606,
+      "step": 7066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102134924747877e-06,
+      "loss": 0.2535,
+      "step": 7067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099477921722358e-06,
+      "loss": 0.6453,
+      "step": 7068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096821142254747e-06,
+      "loss": 0.2961,
+      "step": 7069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094164586549821e-06,
+      "loss": 0.2875,
+      "step": 7070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.091508254812331e-06,
+      "loss": 0.2695,
+      "step": 7071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088852147247009e-06,
+      "loss": 0.2967,
+      "step": 7072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.086196264058584e-06,
+      "loss": 0.2778,
+      "step": 7073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0835406054517505e-06,
+      "loss": 0.2772,
+      "step": 7074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.080885171631198e-06,
+      "loss": 0.264,
+      "step": 7075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078229962801592e-06,
+      "loss": 0.2643,
+      "step": 7076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.075574979167585e-06,
+      "loss": 0.2778,
+      "step": 7077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.072920220933808e-06,
+      "loss": 0.298,
+      "step": 7078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070265688304873e-06,
+      "loss": 0.6022,
+      "step": 7079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.067611381485388e-06,
+      "loss": 0.3085,
+      "step": 7080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064957300679927e-06,
+      "loss": 0.2512,
+      "step": 7081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.062303446093051e-06,
+      "loss": 0.2683,
+      "step": 7082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.059649817929315e-06,
+      "loss": 0.2716,
+      "step": 7083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056996416393241e-06,
+      "loss": 0.3134,
+      "step": 7084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054343241689343e-06,
+      "loss": 0.2478,
+      "step": 7085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.051690294022108e-06,
+      "loss": 0.2875,
+      "step": 7086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.049037573596021e-06,
+      "loss": 0.2996,
+      "step": 7087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0463850806155355e-06,
+      "loss": 0.2771,
+      "step": 7088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.043732815285091e-06,
+      "loss": 0.2462,
+      "step": 7089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041080777809118e-06,
+      "loss": 0.3026,
+      "step": 7090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038428968392018e-06,
+      "loss": 0.247,
+      "step": 7091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0357773872381804e-06,
+      "loss": 0.2667,
+      "step": 7092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0331260345519705e-06,
+      "loss": 0.2906,
+      "step": 7093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.030474910537748e-06,
+      "loss": 0.3019,
+      "step": 7094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027824015399849e-06,
+      "loss": 0.3752,
+      "step": 7095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.025173349342584e-06,
+      "loss": 0.2483,
+      "step": 7096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022522912570262e-06,
+      "loss": 0.2938,
+      "step": 7097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019872705287163e-06,
+      "loss": 0.5854,
+      "step": 7098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017222727697548e-06,
+      "loss": 0.5742,
+      "step": 7099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014572980005667e-06,
+      "loss": 0.2359,
+      "step": 7100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.01192346241575e-06,
+      "loss": 0.2411,
+      "step": 7101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.009274175132009e-06,
+      "loss": 0.277,
+      "step": 7102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006625118358633e-06,
+      "loss": 0.274,
+      "step": 7103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.003976292299807e-06,
+      "loss": 0.2981,
+      "step": 7104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001327697159684e-06,
+      "loss": 0.2745,
+      "step": 7105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.998679333142403e-06,
+      "loss": 0.2916,
+      "step": 7106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996031200452087e-06,
+      "loss": 0.2646,
+      "step": 7107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9933832992928476e-06,
+      "loss": 0.269,
+      "step": 7108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.990735629868768e-06,
+      "loss": 0.27,
+      "step": 7109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9880881923839105e-06,
+      "loss": 0.2866,
+      "step": 7110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985440987042339e-06,
+      "loss": 0.272,
+      "step": 7111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9827940140480776e-06,
+      "loss": 0.2966,
+      "step": 7112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980147273605146e-06,
+      "loss": 0.2431,
+      "step": 7113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97750076591754e-06,
+      "loss": 0.259,
+      "step": 7114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.974854491189243e-06,
+      "loss": 0.2808,
+      "step": 7115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9722084496242146e-06,
+      "loss": 0.2999,
+      "step": 7116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969562641426394e-06,
+      "loss": 0.2514,
+      "step": 7117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.966917066799714e-06,
+      "loss": 0.2651,
+      "step": 7118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.96427172594808e-06,
+      "loss": 0.2206,
+      "step": 7119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.961626619075377e-06,
+      "loss": 0.253,
+      "step": 7120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.958981746385486e-06,
+      "loss": 0.3116,
+      "step": 7121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956337108082256e-06,
+      "loss": 0.332,
+      "step": 7122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.953692704369522e-06,
+      "loss": 0.261,
+      "step": 7123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951048535451099e-06,
+      "loss": 0.2593,
+      "step": 7124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.948404601530793e-06,
+      "loss": 0.2974,
+      "step": 7125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9457609028123795e-06,
+      "loss": 0.3338,
+      "step": 7126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.943117439499622e-06,
+      "loss": 0.3351,
+      "step": 7127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940474211796273e-06,
+      "loss": 0.3165,
+      "step": 7128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.937831219906055e-06,
+      "loss": 0.2773,
+      "step": 7129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.935188464032674e-06,
+      "loss": 0.3149,
+      "step": 7130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932545944379818e-06,
+      "loss": 0.2656,
+      "step": 7131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929903661151167e-06,
+      "loss": 0.2434,
+      "step": 7132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.927261614550375e-06,
+      "loss": 0.3065,
+      "step": 7133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924619804781069e-06,
+      "loss": 0.2697,
+      "step": 7134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921978232046878e-06,
+      "loss": 0.3273,
+      "step": 7135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.919336896551396e-06,
+      "loss": 0.3043,
+      "step": 7136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.916695798498201e-06,
+      "loss": 0.2528,
+      "step": 7137
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91405493809086e-06,
+      "loss": 0.298,
+      "step": 7138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.911414315532914e-06,
+      "loss": 0.2842,
+      "step": 7139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9087739310278956e-06,
+      "loss": 0.2744,
+      "step": 7140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906133784779303e-06,
+      "loss": 0.2769,
+      "step": 7141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.903493876990637e-06,
+      "loss": 0.2692,
+      "step": 7142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900854207865361e-06,
+      "loss": 0.5897,
+      "step": 7143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.898214777606927e-06,
+      "loss": 0.2576,
+      "step": 7144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.89557558641877e-06,
+      "loss": 0.2808,
+      "step": 7145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.892936634504313e-06,
+      "loss": 0.3228,
+      "step": 7146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890297922066947e-06,
+      "loss": 0.2793,
+      "step": 7147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.887659449310045e-06,
+      "loss": 0.3448,
+      "step": 7148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885021216436981e-06,
+      "loss": 0.2498,
+      "step": 7149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882383223651088e-06,
+      "loss": 0.268,
+      "step": 7150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879745471155692e-06,
+      "loss": 0.2971,
+      "step": 7151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.877107959154094e-06,
+      "loss": 0.275,
+      "step": 7152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8744706878495885e-06,
+      "loss": 0.2622,
+      "step": 7153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871833657445438e-06,
+      "loss": 0.235,
+      "step": 7154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8691968681448895e-06,
+      "loss": 0.2673,
+      "step": 7155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.866560320151179e-06,
+      "loss": 0.2803,
+      "step": 7156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8639240136675156e-06,
+      "loss": 0.3454,
+      "step": 7157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.861287948897091e-06,
+      "loss": 0.2888,
+      "step": 7158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858652126043086e-06,
+      "loss": 0.3086,
+      "step": 7159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.856016545308655e-06,
+      "loss": 0.2526,
+      "step": 7160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853381206896932e-06,
+      "loss": 0.2421,
+      "step": 7161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850746111011034e-06,
+      "loss": 0.2998,
+      "step": 7162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.848111257854069e-06,
+      "loss": 0.2838,
+      "step": 7163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.845476647629112e-06,
+      "loss": 0.3314,
+      "step": 7164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842842280539226e-06,
+      "loss": 0.2663,
+      "step": 7165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8402081567874625e-06,
+      "loss": 0.2894,
+      "step": 7166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8375742765768394e-06,
+      "loss": 0.3058,
+      "step": 7167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8349406401103655e-06,
+      "loss": 0.2712,
+      "step": 7168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832307247591026e-06,
+      "loss": 0.2624,
+      "step": 7169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8296740992217915e-06,
+      "loss": 0.2549,
+      "step": 7170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.827041195205615e-06,
+      "loss": 0.2476,
+      "step": 7171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824408535745422e-06,
+      "loss": 0.2554,
+      "step": 7172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82177612104413e-06,
+      "loss": 0.2761,
+      "step": 7173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819143951304632e-06,
+      "loss": 0.2582,
+      "step": 7174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8165120267297994e-06,
+      "loss": 0.3132,
+      "step": 7175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.813880347522489e-06,
+      "loss": 0.2543,
+      "step": 7176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811248913885539e-06,
+      "loss": 0.2748,
+      "step": 7177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8086177260217675e-06,
+      "loss": 0.2635,
+      "step": 7178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80598678413397e-06,
+      "loss": 0.2427,
+      "step": 7179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.803356088424933e-06,
+      "loss": 0.276,
+      "step": 7180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.800725639097412e-06,
+      "loss": 0.2837,
+      "step": 7181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7980954363541506e-06,
+      "loss": 0.2981,
+      "step": 7182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.795465480397868e-06,
+      "loss": 0.2987,
+      "step": 7183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792835771431278e-06,
+      "loss": 0.2744,
+      "step": 7184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790206309657058e-06,
+      "loss": 0.2968,
+      "step": 7185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.787577095277873e-06,
+      "loss": 0.2891,
+      "step": 7186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.784948128496376e-06,
+      "loss": 0.2675,
+      "step": 7187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782319409515188e-06,
+      "loss": 0.2592,
+      "step": 7188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7796909385369245e-06,
+      "loss": 0.2728,
+      "step": 7189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.777062715764166e-06,
+      "loss": 0.2361,
+      "step": 7190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774434741399493e-06,
+      "loss": 0.2347,
+      "step": 7191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771807015645453e-06,
+      "loss": 0.2879,
+      "step": 7192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7691795387045735e-06,
+      "loss": 0.2632,
+      "step": 7193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766552310779374e-06,
+      "loss": 0.2704,
+      "step": 7194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.763925332072343e-06,
+      "loss": 0.2549,
+      "step": 7195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761298602785957e-06,
+      "loss": 0.2593,
+      "step": 7196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758672123122675e-06,
+      "loss": 0.2761,
+      "step": 7197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7560458932849306e-06,
+      "loss": 0.2642,
+      "step": 7198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753419913475139e-06,
+      "loss": 0.2402,
+      "step": 7199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7507941838956946e-06,
+      "loss": 0.3021,
+      "step": 7200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748168704748984e-06,
+      "loss": 0.2784,
+      "step": 7201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74554347623736e-06,
+      "loss": 0.2567,
+      "step": 7202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.742918498563163e-06,
+      "loss": 0.3071,
+      "step": 7203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740293771928717e-06,
+      "loss": 0.2748,
+      "step": 7204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7376692965363196e-06,
+      "loss": 0.3159,
+      "step": 7205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.735045072588256e-06,
+      "loss": 0.2794,
+      "step": 7206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.732421100286779e-06,
+      "loss": 0.259,
+      "step": 7207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7297973798341405e-06,
+      "loss": 0.2919,
+      "step": 7208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.727173911432565e-06,
+      "loss": 0.3078,
+      "step": 7209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.724550695284247e-06,
+      "loss": 0.2858,
+      "step": 7210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.721927731591382e-06,
+      "loss": 0.2748,
+      "step": 7211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.71930502055613e-06,
+      "loss": 0.2647,
+      "step": 7212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716682562380634e-06,
+      "loss": 0.2778,
+      "step": 7213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714060357267023e-06,
+      "loss": 0.2635,
+      "step": 7214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711438405417403e-06,
+      "loss": 0.3018,
+      "step": 7215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708816707033865e-06,
+      "loss": 0.5948,
+      "step": 7216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.706195262318467e-06,
+      "loss": 0.2814,
+      "step": 7217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.703574071473269e-06,
+      "loss": 0.2925,
+      "step": 7218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7009531347002924e-06,
+      "loss": 0.298,
+      "step": 7219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.698332452201545e-06,
+      "loss": 0.2256,
+      "step": 7220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695712024179015e-06,
+      "loss": 0.3104,
+      "step": 7221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693091850834681e-06,
+      "loss": 0.264,
+      "step": 7222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690471932370487e-06,
+      "loss": 0.2758,
+      "step": 7223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6878522689883596e-06,
+      "loss": 0.2579,
+      "step": 7224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6852328608902165e-06,
+      "loss": 0.2759,
+      "step": 7225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682613708277945e-06,
+      "loss": 0.2706,
+      "step": 7226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679994811353419e-06,
+      "loss": 0.2902,
+      "step": 7227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.677376170318484e-06,
+      "loss": 0.2663,
+      "step": 7228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.674757785374979e-06,
+      "loss": 0.2781,
+      "step": 7229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672139656724715e-06,
+      "loss": 0.3488,
+      "step": 7230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669521784569479e-06,
+      "loss": 0.3009,
+      "step": 7231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66690416911105e-06,
+      "loss": 0.3086,
+      "step": 7232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664286810551177e-06,
+      "loss": 0.2546,
+      "step": 7233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6616697090915975e-06,
+      "loss": 0.277,
+      "step": 7234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.659052864934017e-06,
+      "loss": 0.2794,
+      "step": 7235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656436278280136e-06,
+      "loss": 0.2183,
+      "step": 7236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.653819949331628e-06,
+      "loss": 0.3157,
+      "step": 7237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.651203878290139e-06,
+      "loss": 0.3036,
+      "step": 7238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648588065357313e-06,
+      "loss": 0.3045,
+      "step": 7239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.645972510734756e-06,
+      "loss": 0.3218,
+      "step": 7240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643357214624064e-06,
+      "loss": 0.2398,
+      "step": 7241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640742177226816e-06,
+      "loss": 0.2417,
+      "step": 7242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638127398744563e-06,
+      "loss": 0.2916,
+      "step": 7243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635512879378837e-06,
+      "loss": 0.2627,
+      "step": 7244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632898619331151e-06,
+      "loss": 0.3278,
+      "step": 7245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630284618803003e-06,
+      "loss": 0.3138,
+      "step": 7246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6276708779958696e-06,
+      "loss": 0.2462,
+      "step": 7247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6250573971111975e-06,
+      "loss": 0.2772,
+      "step": 7248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622444176350429e-06,
+      "loss": 0.2576,
+      "step": 7249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619831215914974e-06,
+      "loss": 0.2695,
+      "step": 7250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6172185160062255e-06,
+      "loss": 0.3051,
+      "step": 7251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6146060768255596e-06,
+      "loss": 0.3267,
+      "step": 7252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611993898574329e-06,
+      "loss": 0.3086,
+      "step": 7253
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.609381981453869e-06,
+      "loss": 0.2958,
+      "step": 7254
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60677032566549e-06,
+      "loss": 0.2664,
+      "step": 7255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604158931410491e-06,
+      "loss": 0.3224,
+      "step": 7256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601547798890142e-06,
+      "loss": 0.2673,
+      "step": 7257
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.598936928305695e-06,
+      "loss": 0.3295,
+      "step": 7258
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596326319858382e-06,
+      "loss": 0.2814,
+      "step": 7259
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593715973749422e-06,
+      "loss": 0.3036,
+      "step": 7260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591105890180005e-06,
+      "loss": 0.3146,
+      "step": 7261
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5884960693512965e-06,
+      "loss": 0.2615,
+      "step": 7262
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5858865114644584e-06,
+      "loss": 0.269,
+      "step": 7263
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.583277216720618e-06,
+      "loss": 0.2655,
+      "step": 7264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.580668185320889e-06,
+      "loss": 0.295,
+      "step": 7265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578059417466356e-06,
+      "loss": 0.2618,
+      "step": 7266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575450913358099e-06,
+      "loss": 0.2559,
+      "step": 7267
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.572842673197164e-06,
+      "loss": 0.3317,
+      "step": 7268
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.570234697184578e-06,
+      "loss": 0.2997,
+      "step": 7269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5676269855213585e-06,
+      "loss": 0.3006,
+      "step": 7270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.565019538408488e-06,
+      "loss": 0.3048,
+      "step": 7271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562412356046943e-06,
+      "loss": 0.3724,
+      "step": 7272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559805438637663e-06,
+      "loss": 0.2789,
+      "step": 7273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557198786381584e-06,
+      "loss": 0.2672,
+      "step": 7274
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554592399479614e-06,
+      "loss": 0.2433,
+      "step": 7275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5519862781326315e-06,
+      "loss": 0.269,
+      "step": 7276
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549380422541514e-06,
+      "loss": 0.2672,
+      "step": 7277
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.546774832907101e-06,
+      "loss": 0.274,
+      "step": 7278
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544169509430219e-06,
+      "loss": 0.2919,
+      "step": 7279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541564452311681e-06,
+      "loss": 0.3206,
+      "step": 7280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.538959661752264e-06,
+      "loss": 0.3083,
+      "step": 7281
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.536355137952737e-06,
+      "loss": 0.2836,
+      "step": 7282
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.533750881113836e-06,
+      "loss": 0.2808,
+      "step": 7283
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.531146891436293e-06,
+      "loss": 0.3727,
+      "step": 7284
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.528543169120809e-06,
+      "loss": 0.2753,
+      "step": 7285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.52593971436806e-06,
+      "loss": 0.2754,
+      "step": 7286
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523336527378716e-06,
+      "loss": 0.2632,
+      "step": 7287
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.520733608353415e-06,
+      "loss": 0.269,
+      "step": 7288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518130957492774e-06,
+      "loss": 0.2343,
+      "step": 7289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515528574997394e-06,
+      "loss": 0.3079,
+      "step": 7290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.512926461067853e-06,
+      "loss": 0.2656,
+      "step": 7291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.510324615904713e-06,
+      "loss": 0.2411,
+      "step": 7292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507723039708505e-06,
+      "loss": 0.2785,
+      "step": 7293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5051217326797535e-06,
+      "loss": 0.3478,
+      "step": 7294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5025206950189475e-06,
+      "loss": 0.258,
+      "step": 7295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.499919926926566e-06,
+      "loss": 0.2908,
+      "step": 7296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.49731942860306e-06,
+      "loss": 0.2934,
+      "step": 7297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494719200248867e-06,
+      "loss": 0.275,
+      "step": 7298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.492119242064398e-06,
+      "loss": 0.2389,
+      "step": 7299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.489519554250043e-06,
+      "loss": 0.2653,
+      "step": 7300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4869201370061785e-06,
+      "loss": 0.2875,
+      "step": 7301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484320990533148e-06,
+      "loss": 0.3289,
+      "step": 7302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.481722115031287e-06,
+      "loss": 0.3198,
+      "step": 7303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479123510700896e-06,
+      "loss": 0.2668,
+      "step": 7304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.476525177742272e-06,
+      "loss": 0.2445,
+      "step": 7305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473927116355678e-06,
+      "loss": 0.3007,
+      "step": 7306
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4713293267413555e-06,
+      "loss": 0.29,
+      "step": 7307
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.468731809099536e-06,
+      "loss": 0.2662,
+      "step": 7308
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.466134563630418e-06,
+      "loss": 0.2924,
+      "step": 7309
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463537590534188e-06,
+      "loss": 0.2774,
+      "step": 7310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.460940890011004e-06,
+      "loss": 0.2723,
+      "step": 7311
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4583444622610126e-06,
+      "loss": 0.3069,
+      "step": 7312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.455748307484328e-06,
+      "loss": 0.2598,
+      "step": 7313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453152425881051e-06,
+      "loss": 0.3437,
+      "step": 7314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.450556817651261e-06,
+      "loss": 0.288,
+      "step": 7315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447961482995011e-06,
+      "loss": 0.2327,
+      "step": 7316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4453664221123425e-06,
+      "loss": 0.2458,
+      "step": 7317
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44277163520326e-06,
+      "loss": 0.2919,
+      "step": 7318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.440177122467769e-06,
+      "loss": 0.288,
+      "step": 7319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.437582884105835e-06,
+      "loss": 0.2764,
+      "step": 7320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434988920317407e-06,
+      "loss": 0.2935,
+      "step": 7321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432395231302418e-06,
+      "loss": 0.2794,
+      "step": 7322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429801817260779e-06,
+      "loss": 0.2689,
+      "step": 7323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4272086783923715e-06,
+      "loss": 0.2647,
+      "step": 7324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.424615814897068e-06,
+      "loss": 0.267,
+      "step": 7325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.422023226974713e-06,
+      "loss": 0.3537,
+      "step": 7326
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.419430914825125e-06,
+      "loss": 0.5598,
+      "step": 7327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4168388786481106e-06,
+      "loss": 0.5558,
+      "step": 7328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414247118643451e-06,
+      "loss": 0.2962,
+      "step": 7329
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411655635010907e-06,
+      "loss": 0.2992,
+      "step": 7330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409064427950213e-06,
+      "loss": 0.2709,
+      "step": 7331
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406473497661092e-06,
+      "loss": 0.2609,
+      "step": 7332
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.403882844343239e-06,
+      "loss": 0.3163,
+      "step": 7333
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4012924681963255e-06,
+      "loss": 0.2516,
+      "step": 7334
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3987023694200045e-06,
+      "loss": 0.2801,
+      "step": 7335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.396112548213913e-06,
+      "loss": 0.2522,
+      "step": 7336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393523004777661e-06,
+      "loss": 0.2338,
+      "step": 7337
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39093373931083e-06,
+      "loss": 0.2737,
+      "step": 7338
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.388344752012999e-06,
+      "loss": 0.2892,
+      "step": 7339
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385756043083706e-06,
+      "loss": 0.3145,
+      "step": 7340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383167612722481e-06,
+      "loss": 0.2906,
+      "step": 7341
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.38057946112882e-06,
+      "loss": 0.3256,
+      "step": 7342
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3779915885022145e-06,
+      "loss": 0.2568,
+      "step": 7343
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375403995042122e-06,
+      "loss": 0.3112,
+      "step": 7344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3728166809479744e-06,
+      "loss": 0.254,
+      "step": 7345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.370229646419199e-06,
+      "loss": 0.2955,
+      "step": 7346
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3676428916551856e-06,
+      "loss": 0.2528,
+      "step": 7347
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365056416855311e-06,
+      "loss": 0.2457,
+      "step": 7348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3624702222189235e-06,
+      "loss": 0.2742,
+      "step": 7349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.359884307945363e-06,
+      "loss": 0.2371,
+      "step": 7350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357298674233931e-06,
+      "loss": 0.2926,
+      "step": 7351
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.354713321283916e-06,
+      "loss": 0.2774,
+      "step": 7352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352128249294591e-06,
+      "loss": 0.2715,
+      "step": 7353
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349543458465193e-06,
+      "loss": 0.2508,
+      "step": 7354
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3469589489949504e-06,
+      "loss": 0.254,
+      "step": 7355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3443747210830565e-06,
+      "loss": 0.2607,
+      "step": 7356
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3417907749287e-06,
+      "loss": 0.285,
+      "step": 7357
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.339207110731036e-06,
+      "loss": 0.3003,
+      "step": 7358
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336623728689195e-06,
+      "loss": 0.2698,
+      "step": 7359
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3340406290022986e-06,
+      "loss": 0.2584,
+      "step": 7360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.331457811869437e-06,
+      "loss": 0.3047,
+      "step": 7361
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.328875277489677e-06,
+      "loss": 0.2912,
+      "step": 7362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326293026062075e-06,
+      "loss": 0.3234,
+      "step": 7363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3237110577856534e-06,
+      "loss": 0.2498,
+      "step": 7364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.321129372859418e-06,
+      "loss": 0.2927,
+      "step": 7365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318547971482352e-06,
+      "loss": 0.2851,
+      "step": 7366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.315966853853417e-06,
+      "loss": 0.2859,
+      "step": 7367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.313386020171557e-06,
+      "loss": 0.2474,
+      "step": 7368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310805470635682e-06,
+      "loss": 0.2416,
+      "step": 7369
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3082252054446955e-06,
+      "loss": 0.3397,
+      "step": 7370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30564522479747e-06,
+      "loss": 0.2685,
+      "step": 7371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303065528892853e-06,
+      "loss": 0.2521,
+      "step": 7372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300486117929676e-06,
+      "loss": 0.2812,
+      "step": 7373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297906992106755e-06,
+      "loss": 0.644,
+      "step": 7374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295328151622868e-06,
+      "loss": 0.3009,
+      "step": 7375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.292749596676779e-06,
+      "loss": 0.2661,
+      "step": 7376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290171327467238e-06,
+      "loss": 0.2781,
+      "step": 7377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287593344192957e-06,
+      "loss": 0.2819,
+      "step": 7378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285015647052639e-06,
+      "loss": 0.2754,
+      "step": 7379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282438236244956e-06,
+      "loss": 0.2919,
+      "step": 7380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2798611119685685e-06,
+      "loss": 0.2772,
+      "step": 7381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277284274422104e-06,
+      "loss": 0.3116,
+      "step": 7382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.27470772380417e-06,
+      "loss": 0.302,
+      "step": 7383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.27213146031336e-06,
+      "loss": 0.2711,
+      "step": 7384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.269555484148237e-06,
+      "loss": 0.2593,
+      "step": 7385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266979795507346e-06,
+      "loss": 0.3012,
+      "step": 7386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.264404394589202e-06,
+      "loss": 0.2141,
+      "step": 7387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261829281592313e-06,
+      "loss": 0.2646,
+      "step": 7388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259254456715154e-06,
+      "loss": 0.2444,
+      "step": 7389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.256679920156172e-06,
+      "loss": 0.2916,
+      "step": 7390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25410567211381e-06,
+      "loss": 0.2636,
+      "step": 7391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.251531712786473e-06,
+      "loss": 0.2279,
+      "step": 7392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.24895804237255e-06,
+      "loss": 0.2864,
+      "step": 7393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246384661070404e-06,
+      "loss": 0.2712,
+      "step": 7394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243811569078384e-06,
+      "loss": 0.2708,
+      "step": 7395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.24123876659481e-06,
+      "loss": 0.2595,
+      "step": 7396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238666253817974e-06,
+      "loss": 0.2948,
+      "step": 7397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.23609403094616e-06,
+      "loss": 0.2607,
+      "step": 7398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233522098177622e-06,
+      "loss": 0.2733,
+      "step": 7399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230950455710592e-06,
+      "loss": 0.3316,
+      "step": 7400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228379103743272e-06,
+      "loss": 0.2439,
+      "step": 7401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.225808042473857e-06,
+      "loss": 0.2556,
+      "step": 7402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.223237272100509e-06,
+      "loss": 0.2602,
+      "step": 7403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220666792821371e-06,
+      "loss": 0.2871,
+      "step": 7404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21809660483456e-06,
+      "loss": 0.2881,
+      "step": 7405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2155267083381795e-06,
+      "loss": 0.2601,
+      "step": 7406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212957103530297e-06,
+      "loss": 0.292,
+      "step": 7407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.210387790608972e-06,
+      "loss": 0.2257,
+      "step": 7408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.207818769772231e-06,
+      "loss": 0.2697,
+      "step": 7409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2052500412180805e-06,
+      "loss": 0.3264,
+      "step": 7410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202681605144503e-06,
+      "loss": 0.2346,
+      "step": 7411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200113461749469e-06,
+      "loss": 0.2527,
+      "step": 7412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197545611230913e-06,
+      "loss": 0.267,
+      "step": 7413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.194978053786749e-06,
+      "loss": 0.2754,
+      "step": 7414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19241078961488e-06,
+      "loss": 0.2548,
+      "step": 7415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.189843818913172e-06,
+      "loss": 0.2764,
+      "step": 7416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187277141879476e-06,
+      "loss": 0.271,
+      "step": 7417
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.184710758711616e-06,
+      "loss": 0.2069,
+      "step": 7418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.182144669607403e-06,
+      "loss": 0.2579,
+      "step": 7419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179578874764614e-06,
+      "loss": 0.2655,
+      "step": 7420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177013374381005e-06,
+      "loss": 0.2969,
+      "step": 7421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.174448168654317e-06,
+      "loss": 0.3246,
+      "step": 7422
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171883257782261e-06,
+      "loss": 0.3112,
+      "step": 7423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16931864196253e-06,
+      "loss": 0.2684,
+      "step": 7424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.166754321392785e-06,
+      "loss": 0.248,
+      "step": 7425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164190296270683e-06,
+      "loss": 0.2527,
+      "step": 7426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161626566793837e-06,
+      "loss": 0.2909,
+      "step": 7427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159063133159846e-06,
+      "loss": 0.2526,
+      "step": 7428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.156499995566294e-06,
+      "loss": 0.2944,
+      "step": 7429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1539371542107295e-06,
+      "loss": 0.2761,
+      "step": 7430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151374609290688e-06,
+      "loss": 0.2386,
+      "step": 7431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.148812361003669e-06,
+      "loss": 0.309,
+      "step": 7432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14625040954717e-06,
+      "loss": 0.2401,
+      "step": 7433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1436887551186466e-06,
+      "loss": 0.2916,
+      "step": 7434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141127397915534e-06,
+      "loss": 0.3116,
+      "step": 7435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138566338135259e-06,
+      "loss": 0.2842,
+      "step": 7436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13600557597521e-06,
+      "loss": 0.2538,
+      "step": 7437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133445111632761e-06,
+      "loss": 0.283,
+      "step": 7438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.130884945305252e-06,
+      "loss": 0.2909,
+      "step": 7439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.128325077190018e-06,
+      "loss": 0.2332,
+      "step": 7440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.125765507484356e-06,
+      "loss": 0.2864,
+      "step": 7441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123206236385543e-06,
+      "loss": 0.2509,
+      "step": 7442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.120647264090839e-06,
+      "loss": 0.2197,
+      "step": 7443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1180885907974775e-06,
+      "loss": 0.3011,
+      "step": 7444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115530216702661e-06,
+      "loss": 0.2401,
+      "step": 7445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.112972142003587e-06,
+      "loss": 0.3022,
+      "step": 7446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.110414366897413e-06,
+      "loss": 0.2604,
+      "step": 7447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.107856891581281e-06,
+      "loss": 0.285,
+      "step": 7448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105299716252303e-06,
+      "loss": 0.2635,
+      "step": 7449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102742841107585e-06,
+      "loss": 0.2951,
+      "step": 7450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1001862663441906e-06,
+      "loss": 0.2982,
+      "step": 7451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0976299921591645e-06,
+      "loss": 0.2589,
+      "step": 7452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.095074018749542e-06,
+      "loss": 0.2439,
+      "step": 7453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092518346312317e-06,
+      "loss": 0.2491,
+      "step": 7454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089962975044472e-06,
+      "loss": 0.2677,
+      "step": 7455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.087407905142957e-06,
+      "loss": 0.3097,
+      "step": 7456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084853136804711e-06,
+      "loss": 0.2666,
+      "step": 7457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082298670226642e-06,
+      "loss": 0.2827,
+      "step": 7458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.079744505605628e-06,
+      "loss": 0.2264,
+      "step": 7459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077190643138542e-06,
+      "loss": 0.3216,
+      "step": 7460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0746370830222145e-06,
+      "loss": 0.2244,
+      "step": 7461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0720838254534675e-06,
+      "loss": 0.2762,
+      "step": 7462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069530870629088e-06,
+      "loss": 0.3168,
+      "step": 7463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0669782187458515e-06,
+      "loss": 0.306,
+      "step": 7464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064425870000499e-06,
+      "loss": 0.2759,
+      "step": 7465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.061873824589751e-06,
+      "loss": 0.2559,
+      "step": 7466
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059322082710315e-06,
+      "loss": 0.2316,
+      "step": 7467
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056770644558858e-06,
+      "loss": 0.2603,
+      "step": 7468
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054219510332038e-06,
+      "loss": 0.2874,
+      "step": 7469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.051668680226477e-06,
+      "loss": 0.2875,
+      "step": 7470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049118154438789e-06,
+      "loss": 0.281,
+      "step": 7471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046567933165552e-06,
+      "loss": 0.2738,
+      "step": 7472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.044018016603321e-06,
+      "loss": 0.2448,
+      "step": 7473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0414684049486335e-06,
+      "loss": 0.2523,
+      "step": 7474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.038919098398006e-06,
+      "loss": 0.2759,
+      "step": 7475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.036370097147922e-06,
+      "loss": 0.2946,
+      "step": 7476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.033821401394842e-06,
+      "loss": 0.2922,
+      "step": 7477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.031273011335215e-06,
+      "loss": 0.2779,
+      "step": 7478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.028724927165452e-06,
+      "loss": 0.2715,
+      "step": 7479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.026177149081949e-06,
+      "loss": 0.3035,
+      "step": 7480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.023629677281075e-06,
+      "loss": 0.2566,
+      "step": 7481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0210825119591806e-06,
+      "loss": 0.2521,
+      "step": 7482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018535653312586e-06,
+      "loss": 0.3024,
+      "step": 7483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.015989101537586e-06,
+      "loss": 0.2773,
+      "step": 7484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0134428568304645e-06,
+      "loss": 0.2871,
+      "step": 7485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0108969193874675e-06,
+      "loss": 0.2777,
+      "step": 7486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008351289404824e-06,
+      "loss": 0.2424,
+      "step": 7487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.005805967078741e-06,
+      "loss": 0.3023,
+      "step": 7488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.003260952605401e-06,
+      "loss": 0.2704,
+      "step": 7489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000716246180953e-06,
+      "loss": 0.2845,
+      "step": 7490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9981718480015416e-06,
+      "loss": 0.3026,
+      "step": 7491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995627758263267e-06,
+      "loss": 0.2785,
+      "step": 7492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9930839771622196e-06,
+      "loss": 0.2854,
+      "step": 7493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9905405048944575e-06,
+      "loss": 0.2971,
+      "step": 7494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.987997341656027e-06,
+      "loss": 0.2958,
+      "step": 7495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9854544876429364e-06,
+      "loss": 0.2897,
+      "step": 7496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.982911943051173e-06,
+      "loss": 0.2921,
+      "step": 7497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.980369708076713e-06,
+      "loss": 0.288,
+      "step": 7498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977827782915493e-06,
+      "loss": 0.3124,
+      "step": 7499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.975286167763433e-06,
+      "loss": 0.2576,
+      "step": 7500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972744862816426e-06,
+      "loss": 0.2802,
+      "step": 7501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.97020386827035e-06,
+      "loss": 0.2531,
+      "step": 7502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.967663184321047e-06,
+      "loss": 0.2457,
+      "step": 7503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9651228111643385e-06,
+      "loss": 0.2734,
+      "step": 7504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962582748996031e-06,
+      "loss": 0.2679,
+      "step": 7505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960042998011892e-06,
+      "loss": 0.3422,
+      "step": 7506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95750355840768e-06,
+      "loss": 0.2347,
+      "step": 7507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.954964430379115e-06,
+      "loss": 0.2913,
+      "step": 7508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952425614121908e-06,
+      "loss": 0.3139,
+      "step": 7509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.949887109831736e-06,
+      "loss": 0.3038,
+      "step": 7510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.947348917704248e-06,
+      "loss": 0.3193,
+      "step": 7511
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944811037935083e-06,
+      "loss": 0.2969,
+      "step": 7512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94227347071985e-06,
+      "loss": 0.2983,
+      "step": 7513
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.939736216254126e-06,
+      "loss": 0.2701,
+      "step": 7514
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.937199274733468e-06,
+      "loss": 0.2819,
+      "step": 7515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93466264635342e-06,
+      "loss": 0.3,
+      "step": 7516
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932126331309486e-06,
+      "loss": 0.2663,
+      "step": 7517
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929590329797154e-06,
+      "loss": 0.2562,
+      "step": 7518
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9270546420118855e-06,
+      "loss": 0.2665,
+      "step": 7519
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924519268149123e-06,
+      "loss": 0.2967,
+      "step": 7520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.921984208404279e-06,
+      "loss": 0.3071,
+      "step": 7521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919449462972737e-06,
+      "loss": 0.2625,
+      "step": 7522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916915032049873e-06,
+      "loss": 0.2537,
+      "step": 7523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.91438091583102e-06,
+      "loss": 0.2715,
+      "step": 7524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.911847114511497e-06,
+      "loss": 0.3113,
+      "step": 7525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9093136282866014e-06,
+      "loss": 0.3052,
+      "step": 7526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9067804573516e-06,
+      "loss": 0.2943,
+      "step": 7527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9042476019017304e-06,
+      "loss": 0.2795,
+      "step": 7528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.901715062132223e-06,
+      "loss": 0.2703,
+      "step": 7529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899182838238265e-06,
+      "loss": 0.2491,
+      "step": 7530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8966509304150354e-06,
+      "loss": 0.597,
+      "step": 7531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894119338857671e-06,
+      "loss": 0.2837,
+      "step": 7532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891588063761304e-06,
+      "loss": 0.2944,
+      "step": 7533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8890571053210295e-06,
+      "loss": 0.259,
+      "step": 7534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886526463731916e-06,
+      "loss": 0.3053,
+      "step": 7535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8839961391890234e-06,
+      "loss": 0.252,
+      "step": 7536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8814661318873665e-06,
+      "loss": 0.6045,
+      "step": 7537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878936442021952e-06,
+      "loss": 0.3126,
+      "step": 7538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87640706978775e-06,
+      "loss": 0.2875,
+      "step": 7539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873878015379722e-06,
+      "loss": 0.2729,
+      "step": 7540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.871349278992786e-06,
+      "loss": 0.2519,
+      "step": 7541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868820860821844e-06,
+      "loss": 0.2601,
+      "step": 7542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866292761061781e-06,
+      "loss": 0.2795,
+      "step": 7543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863764979907446e-06,
+      "loss": 0.2809,
+      "step": 7544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.861237517553669e-06,
+      "loss": 0.2676,
+      "step": 7545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858710374195251e-06,
+      "loss": 0.311,
+      "step": 7546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8561835500269795e-06,
+      "loss": 0.2461,
+      "step": 7547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853657045243604e-06,
+      "loss": 0.3074,
+      "step": 7548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851130860039851e-06,
+      "loss": 0.2747,
+      "step": 7549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848604994610434e-06,
+      "loss": 0.351,
+      "step": 7550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8460794491500325e-06,
+      "loss": 0.2626,
+      "step": 7551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.843554223853303e-06,
+      "loss": 0.3003,
+      "step": 7552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8410293189148704e-06,
+      "loss": 0.299,
+      "step": 7553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838504734529353e-06,
+      "loss": 0.2704,
+      "step": 7554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.835980470891327e-06,
+      "loss": 0.2344,
+      "step": 7555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.83345652819535e-06,
+      "loss": 0.2782,
+      "step": 7556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8309329066359575e-06,
+      "loss": 0.288,
+      "step": 7557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828409606407659e-06,
+      "loss": 0.2661,
+      "step": 7558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.825886627704933e-06,
+      "loss": 0.2749,
+      "step": 7559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823363970722237e-06,
+      "loss": 0.258,
+      "step": 7560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.820841635654015e-06,
+      "loss": 0.3162,
+      "step": 7561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818319622694668e-06,
+      "loss": 0.5818,
+      "step": 7562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815797932038581e-06,
+      "loss": 0.2512,
+      "step": 7563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813276563880114e-06,
+      "loss": 0.2894,
+      "step": 7564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.810755518413605e-06,
+      "loss": 0.2413,
+      "step": 7565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8082347958333625e-06,
+      "loss": 0.2911,
+      "step": 7566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.805714396333662e-06,
+      "loss": 0.2607,
+      "step": 7567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8031943201087805e-06,
+      "loss": 0.2923,
+      "step": 7568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.800674567352938e-06,
+      "loss": 0.3039,
+      "step": 7569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798155138260352e-06,
+      "loss": 0.2632,
+      "step": 7570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795636033025205e-06,
+      "loss": 0.2619,
+      "step": 7571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.793117251841659e-06,
+      "loss": 0.2997,
+      "step": 7572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790598794903847e-06,
+      "loss": 0.3199,
+      "step": 7573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788080662405881e-06,
+      "loss": 0.2531,
+      "step": 7574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7855628545418515e-06,
+      "loss": 0.2776,
+      "step": 7575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783045371505809e-06,
+      "loss": 0.295,
+      "step": 7576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780528213491792e-06,
+      "loss": 0.2551,
+      "step": 7577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7780113806938095e-06,
+      "loss": 0.2497,
+      "step": 7578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77549487330585e-06,
+      "loss": 0.2665,
+      "step": 7579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772978691521871e-06,
+      "loss": 0.2389,
+      "step": 7580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770462835535809e-06,
+      "loss": 0.3396,
+      "step": 7581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.767947305541577e-06,
+      "loss": 0.2678,
+      "step": 7582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7654321017330505e-06,
+      "loss": 0.3085,
+      "step": 7583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762917224304094e-06,
+      "loss": 0.2903,
+      "step": 7584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760402673448544e-06,
+      "loss": 0.2639,
+      "step": 7585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757888449360205e-06,
+      "loss": 0.3151,
+      "step": 7586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755374552232864e-06,
+      "loss": 0.2701,
+      "step": 7587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.75286098226028e-06,
+      "loss": 0.3142,
+      "step": 7588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750347739636188e-06,
+      "loss": 0.3275,
+      "step": 7589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.747834824554293e-06,
+      "loss": 0.2758,
+      "step": 7590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745322237208273e-06,
+      "loss": 0.3183,
+      "step": 7591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7428099777918e-06,
+      "loss": 0.3133,
+      "step": 7592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740298046498496e-06,
+      "loss": 0.2865,
+      "step": 7593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.737786443521968e-06,
+      "loss": 0.3123,
+      "step": 7594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7352751690558025e-06,
+      "loss": 0.2478,
+      "step": 7595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.732764223293559e-06,
+      "loss": 0.2595,
+      "step": 7596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730253606428759e-06,
+      "loss": 0.2635,
+      "step": 7597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727743318654911e-06,
+      "loss": 0.2667,
+      "step": 7598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.725233360165505e-06,
+      "loss": 0.2758,
+      "step": 7599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.722723731153986e-06,
+      "loss": 0.2652,
+      "step": 7600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.720214431813786e-06,
+      "loss": 0.3037,
+      "step": 7601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.717705462338311e-06,
+      "loss": 0.2795,
+      "step": 7602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71519682292094e-06,
+      "loss": 0.2213,
+      "step": 7603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71268851375503e-06,
+      "loss": 0.2542,
+      "step": 7604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.710180535033897e-06,
+      "loss": 0.3026,
+      "step": 7605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.707672886950859e-06,
+      "loss": 0.3079,
+      "step": 7606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7051655696991825e-06,
+      "loss": 0.2803,
+      "step": 7607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7026585834721225e-06,
+      "loss": 0.2959,
+      "step": 7608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7001519284629045e-06,
+      "loss": 0.2888,
+      "step": 7609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697645604864732e-06,
+      "loss": 0.2676,
+      "step": 7610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6951396128707745e-06,
+      "loss": 0.6045,
+      "step": 7611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.692633952674187e-06,
+      "loss": 0.2894,
+      "step": 7612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6901286244680946e-06,
+      "loss": 0.2874,
+      "step": 7613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687623628445588e-06,
+      "loss": 0.2729,
+      "step": 7614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685118964799743e-06,
+      "loss": 0.2735,
+      "step": 7615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.682614633723609e-06,
+      "loss": 0.2899,
+      "step": 7616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680110635410205e-06,
+      "loss": 0.2418,
+      "step": 7617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67760697005253e-06,
+      "loss": 0.2729,
+      "step": 7618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675103637843551e-06,
+      "loss": 0.2515,
+      "step": 7619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672600638976218e-06,
+      "loss": 0.2402,
+      "step": 7620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67009797364344e-06,
+      "loss": 0.2917,
+      "step": 7621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667595642038117e-06,
+      "loss": 0.5909,
+      "step": 7622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665093644353115e-06,
+      "loss": 0.2448,
+      "step": 7623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.662591980781276e-06,
+      "loss": 0.2631,
+      "step": 7624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.660090651515413e-06,
+      "loss": 0.2591,
+      "step": 7625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.657589656748321e-06,
+      "loss": 0.2198,
+      "step": 7626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.655088996672764e-06,
+      "loss": 0.2787,
+      "step": 7627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.652588671481475e-06,
+      "loss": 0.2795,
+      "step": 7628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650088681367166e-06,
+      "loss": 0.2906,
+      "step": 7629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647589026522535e-06,
+      "loss": 0.3565,
+      "step": 7630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645089707140234e-06,
+      "loss": 0.2897,
+      "step": 7631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642590723412898e-06,
+      "loss": 0.3113,
+      "step": 7632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64009207553314e-06,
+      "loss": 0.6165,
+      "step": 7633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637593763693545e-06,
+      "loss": 0.298,
+      "step": 7634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635095788086664e-06,
+      "loss": 0.2778,
+      "step": 7635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.632598148905027e-06,
+      "loss": 0.3043,
+      "step": 7636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630100846341153e-06,
+      "loss": 0.2673,
+      "step": 7637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.627603880587511e-06,
+      "loss": 0.2756,
+      "step": 7638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.625107251836556e-06,
+      "loss": 0.2678,
+      "step": 7639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622610960280717e-06,
+      "loss": 0.3,
+      "step": 7640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.620115006112396e-06,
+      "loss": 0.2802,
+      "step": 7641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617619389523973e-06,
+      "loss": 0.2824,
+      "step": 7642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.615124110707786e-06,
+      "loss": 0.3045,
+      "step": 7643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612629169856172e-06,
+      "loss": 0.2823,
+      "step": 7644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61013456716142e-06,
+      "loss": 0.243,
+      "step": 7645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607640302815806e-06,
+      "loss": 0.2424,
+      "step": 7646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605146377011572e-06,
+      "loss": 0.2728,
+      "step": 7647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602652789940941e-06,
+      "loss": 0.3058,
+      "step": 7648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60015954179611e-06,
+      "loss": 0.3168,
+      "step": 7649
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597666632769232e-06,
+      "loss": 0.2645,
+      "step": 7650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595174063052465e-06,
+      "loss": 0.2794,
+      "step": 7651
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.592681832837913e-06,
+      "loss": 0.303,
+      "step": 7652
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5901899423176674e-06,
+      "loss": 0.2482,
+      "step": 7653
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587698391683792e-06,
+      "loss": 0.2643,
+      "step": 7654
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.585207181128323e-06,
+      "loss": 0.2599,
+      "step": 7655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5827163108432704e-06,
+      "loss": 0.247,
+      "step": 7656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.580225781020618e-06,
+      "loss": 0.3073,
+      "step": 7657
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577735591852327e-06,
+      "loss": 0.2609,
+      "step": 7658
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575245743530322e-06,
+      "loss": 0.2872,
+      "step": 7659
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572756236246512e-06,
+      "loss": 0.3033,
+      "step": 7660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.570267070192776e-06,
+      "loss": 0.2974,
+      "step": 7661
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567778245560966e-06,
+      "loss": 0.2669,
+      "step": 7662
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565289762542908e-06,
+      "loss": 0.2764,
+      "step": 7663
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562801621330402e-06,
+      "loss": 0.2768,
+      "step": 7664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.560313822115229e-06,
+      "loss": 0.3415,
+      "step": 7665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5578263650891225e-06,
+      "loss": 0.2858,
+      "step": 7666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.555339250443808e-06,
+      "loss": 0.2777,
+      "step": 7667
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.552852478370989e-06,
+      "loss": 0.2498,
+      "step": 7668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.550366049062323e-06,
+      "loss": 0.2642,
+      "step": 7669
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547879962709457e-06,
+      "loss": 0.2096,
+      "step": 7670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.545394219504005e-06,
+      "loss": 0.281,
+      "step": 7671
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.542908819637558e-06,
+      "loss": 0.308,
+      "step": 7672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.540423763301674e-06,
+      "loss": 0.2758,
+      "step": 7673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.537939050687886e-06,
+      "loss": 0.3063,
+      "step": 7674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535454681987715e-06,
+      "loss": 0.2714,
+      "step": 7675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532970657392635e-06,
+      "loss": 0.2983,
+      "step": 7676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.530486977094104e-06,
+      "loss": 0.2545,
+      "step": 7677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528003641283552e-06,
+      "loss": 0.2699,
+      "step": 7678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525520650152383e-06,
+      "loss": 0.2586,
+      "step": 7679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523038003891976e-06,
+      "loss": 0.2681,
+      "step": 7680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5205557026936714e-06,
+      "loss": 0.2833,
+      "step": 7681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5180737467488085e-06,
+      "loss": 0.2584,
+      "step": 7682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51559213624867e-06,
+      "loss": 0.2679,
+      "step": 7683
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.513110871384532e-06,
+      "loss": 0.2529,
+      "step": 7684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510629952347637e-06,
+      "loss": 0.2674,
+      "step": 7685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.508149379329204e-06,
+      "loss": 0.243,
+      "step": 7686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.505669152520425e-06,
+      "loss": 0.2602,
+      "step": 7687
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.503189272112452e-06,
+      "loss": 0.2945,
+      "step": 7688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.50070973829644e-06,
+      "loss": 0.2533,
+      "step": 7689
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4982305512634845e-06,
+      "loss": 0.2582,
+      "step": 7690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.495751711204675e-06,
+      "loss": 0.2778,
+      "step": 7691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493273218311067e-06,
+      "loss": 0.2935,
+      "step": 7692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.490795072773692e-06,
+      "loss": 0.283,
+      "step": 7693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48831727478355e-06,
+      "loss": 0.262,
+      "step": 7694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.485839824531621e-06,
+      "loss": 0.3019,
+      "step": 7695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.483362722208858e-06,
+      "loss": 0.2448,
+      "step": 7696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4808859680061734e-06,
+      "loss": 0.2444,
+      "step": 7697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.478409562114469e-06,
+      "loss": 0.2437,
+      "step": 7698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4759335047246154e-06,
+      "loss": 0.2203,
+      "step": 7699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4734577960274515e-06,
+      "loss": 0.2722,
+      "step": 7700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4709824362137945e-06,
+      "loss": 0.2975,
+      "step": 7701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4685074254744346e-06,
+      "loss": 0.3494,
+      "step": 7702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4660327640001335e-06,
+      "loss": 0.2834,
+      "step": 7703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4635584519816195e-06,
+      "loss": 0.2932,
+      "step": 7704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461084489609603e-06,
+      "loss": 0.2642,
+      "step": 7705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.458610877074773e-06,
+      "loss": 0.2798,
+      "step": 7706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456137614567773e-06,
+      "loss": 0.2859,
+      "step": 7707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453664702279235e-06,
+      "loss": 0.2778,
+      "step": 7708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.451192140399757e-06,
+      "loss": 0.2631,
+      "step": 7709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448719929119916e-06,
+      "loss": 0.3259,
+      "step": 7710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446248068630251e-06,
+      "loss": 0.2751,
+      "step": 7711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.443776559121279e-06,
+      "loss": 0.2667,
+      "step": 7712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4413054007835055e-06,
+      "loss": 0.2521,
+      "step": 7713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4388345938073824e-06,
+      "loss": 0.2806,
+      "step": 7714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.436364138383352e-06,
+      "loss": 0.2941,
+      "step": 7715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433894034701824e-06,
+      "loss": 0.319,
+      "step": 7716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431424282953181e-06,
+      "loss": 0.277,
+      "step": 7717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4289548833277865e-06,
+      "loss": 0.6084,
+      "step": 7718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426485836015953e-06,
+      "loss": 0.6002,
+      "step": 7719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424017141208002e-06,
+      "loss": 0.2696,
+      "step": 7720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421548799094196e-06,
+      "loss": 0.2503,
+      "step": 7721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419080809864785e-06,
+      "loss": 0.3042,
+      "step": 7722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41661317370999e-06,
+      "loss": 0.3444,
+      "step": 7723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414145890820004e-06,
+      "loss": 0.2747,
+      "step": 7724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.411678961384998e-06,
+      "loss": 0.2639,
+      "step": 7725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.409212385595098e-06,
+      "loss": 0.2527,
+      "step": 7726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406746163640432e-06,
+      "loss": 0.2514,
+      "step": 7727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404280295711071e-06,
+      "loss": 0.2802,
+      "step": 7728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401814781997077e-06,
+      "loss": 0.2713,
+      "step": 7729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399349622688479e-06,
+      "loss": 0.2701,
+      "step": 7730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396884817975281e-06,
+      "loss": 0.2708,
+      "step": 7731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.394420368047459e-06,
+      "loss": 0.2332,
+      "step": 7732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391956273094952e-06,
+      "loss": 0.308,
+      "step": 7733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389492533307692e-06,
+      "loss": 0.28,
+      "step": 7734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387029148875563e-06,
+      "loss": 0.2489,
+      "step": 7735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384566119988435e-06,
+      "loss": 0.2212,
+      "step": 7736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382103446836144e-06,
+      "loss": 0.2369,
+      "step": 7737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379641129608501e-06,
+      "loss": 0.2687,
+      "step": 7738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.377179168495292e-06,
+      "loss": 0.2842,
+      "step": 7739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.374717563686269e-06,
+      "loss": 0.226,
+      "step": 7740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.372256315371167e-06,
+      "loss": 0.2617,
+      "step": 7741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3697954237396764e-06,
+      "loss": 0.2941,
+      "step": 7742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367334888981474e-06,
+      "loss": 0.269,
+      "step": 7743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3648747112862145e-06,
+      "loss": 0.2697,
+      "step": 7744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362414890843504e-06,
+      "loss": 0.3158,
+      "step": 7745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3599554278429415e-06,
+      "loss": 0.3206,
+      "step": 7746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357496322474086e-06,
+      "loss": 0.2864,
+      "step": 7747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.355037574926478e-06,
+      "loss": 0.2402,
+      "step": 7748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35257918538962e-06,
+      "loss": 0.2917,
+      "step": 7749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35012115405299e-06,
+      "loss": 0.2921,
+      "step": 7750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347663481106053e-06,
+      "loss": 0.3192,
+      "step": 7751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345206166738225e-06,
+      "loss": 0.3267,
+      "step": 7752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3427492111389045e-06,
+      "loss": 0.2687,
+      "step": 7753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3402926144974625e-06,
+      "loss": 0.2976,
+      "step": 7754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337836377003242e-06,
+      "loss": 0.2578,
+      "step": 7755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.335380498845559e-06,
+      "loss": 0.2601,
+      "step": 7756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.332924980213694e-06,
+      "loss": 0.2804,
+      "step": 7757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.330469821296916e-06,
+      "loss": 0.2832,
+      "step": 7758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32801502228445e-06,
+      "loss": 0.292,
+      "step": 7759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325560583365499e-06,
+      "loss": 0.2629,
+      "step": 7760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.323106504729241e-06,
+      "loss": 0.2498,
+      "step": 7761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.320652786564826e-06,
+      "loss": 0.2895,
+      "step": 7762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318199429061375e-06,
+      "loss": 0.2509,
+      "step": 7763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315746432407972e-06,
+      "loss": 0.2601,
+      "step": 7764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313293796793696e-06,
+      "loss": 0.2717,
+      "step": 7765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3108415224075725e-06,
+      "loss": 0.2889,
+      "step": 7766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.308389609438615e-06,
+      "loss": 0.2663,
+      "step": 7767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305938058075804e-06,
+      "loss": 0.3458,
+      "step": 7768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303486868508093e-06,
+      "loss": 0.2789,
+      "step": 7769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.301036040924412e-06,
+      "loss": 0.26,
+      "step": 7770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298585575513649e-06,
+      "loss": 0.3165,
+      "step": 7771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296135472464686e-06,
+      "loss": 0.2643,
+      "step": 7772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.293685731966353e-06,
+      "loss": 0.2613,
+      "step": 7773
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2912363542074695e-06,
+      "loss": 0.2805,
+      "step": 7774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288787339376822e-06,
+      "loss": 0.3005,
+      "step": 7775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2863386876631674e-06,
+      "loss": 0.2213,
+      "step": 7776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2838903992552355e-06,
+      "loss": 0.3118,
+      "step": 7777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281442474341729e-06,
+      "loss": 0.2983,
+      "step": 7778
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.278994913111326e-06,
+      "loss": 0.2526,
+      "step": 7779
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276547715752663e-06,
+      "loss": 0.2634,
+      "step": 7780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274100882454364e-06,
+      "loss": 0.2818,
+      "step": 7781
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.271654413405016e-06,
+      "loss": 0.3231,
+      "step": 7782
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269208308793183e-06,
+      "loss": 0.2935,
+      "step": 7783
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2667625688074e-06,
+      "loss": 0.2848,
+      "step": 7784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.264317193636168e-06,
+      "loss": 0.2892,
+      "step": 7785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261872183467972e-06,
+      "loss": 0.2807,
+      "step": 7786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2594275384912526e-06,
+      "loss": 0.2913,
+      "step": 7787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25698325889443e-06,
+      "loss": 0.2629,
+      "step": 7788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25453934486591e-06,
+      "loss": 0.2546,
+      "step": 7789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252095796594046e-06,
+      "loss": 0.2908,
+      "step": 7790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249652614267178e-06,
+      "loss": 0.2775,
+      "step": 7791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.247209798073614e-06,
+      "loss": 0.2923,
+      "step": 7792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2447673482016335e-06,
+      "loss": 0.2929,
+      "step": 7793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242325264839494e-06,
+      "loss": 0.2772,
+      "step": 7794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239883548175407e-06,
+      "loss": 0.6228,
+      "step": 7795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2374421983975846e-06,
+      "loss": 0.3563,
+      "step": 7796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23500121569418e-06,
+      "loss": 0.2564,
+      "step": 7797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.232560600253336e-06,
+      "loss": 0.2518,
+      "step": 7798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230120352263166e-06,
+      "loss": 0.2386,
+      "step": 7799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2276804719117504e-06,
+      "loss": 0.3195,
+      "step": 7800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225240959387147e-06,
+      "loss": 0.2998,
+      "step": 7801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22280181487737e-06,
+      "loss": 0.2983,
+      "step": 7802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.220363038570432e-06,
+      "loss": 0.2548,
+      "step": 7803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21792463065429e-06,
+      "loss": 0.2581,
+      "step": 7804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.215486591316888e-06,
+      "loss": 0.3034,
+      "step": 7805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21304892074614e-06,
+      "loss": 0.2719,
+      "step": 7806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210611619129927e-06,
+      "loss": 0.2829,
+      "step": 7807
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20817468665611e-06,
+      "loss": 0.2479,
+      "step": 7808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.205738123512503e-06,
+      "loss": 0.2667,
+      "step": 7809
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20330192988692e-06,
+      "loss": 0.2583,
+      "step": 7810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200866105967119e-06,
+      "loss": 0.2859,
+      "step": 7811
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198430651940846e-06,
+      "loss": 0.2289,
+      "step": 7812
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195995567995813e-06,
+      "loss": 0.2433,
+      "step": 7813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1935608543197035e-06,
+      "loss": 0.3417,
+      "step": 7814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.191126511100179e-06,
+      "loss": 0.2767,
+      "step": 7815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188692538524854e-06,
+      "loss": 0.2842,
+      "step": 7816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.186258936781341e-06,
+      "loss": 0.3012,
+      "step": 7817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183825706057199e-06,
+      "loss": 0.2468,
+      "step": 7818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1813928465399765e-06,
+      "loss": 0.3082,
+      "step": 7819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178960358417184e-06,
+      "loss": 0.2372,
+      "step": 7820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1765282418763045e-06,
+      "loss": 0.3051,
+      "step": 7821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1740964971047945e-06,
+      "loss": 0.3117,
+      "step": 7822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171665124290082e-06,
+      "loss": 0.2623,
+      "step": 7823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.169234123619569e-06,
+      "loss": 0.2814,
+      "step": 7824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.166803495280614e-06,
+      "loss": 0.2733,
+      "step": 7825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164373239460561e-06,
+      "loss": 0.2468,
+      "step": 7826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.161943356346734e-06,
+      "loss": 0.2553,
+      "step": 7827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159513846126403e-06,
+      "loss": 0.2561,
+      "step": 7828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157084708986826e-06,
+      "loss": 0.2724,
+      "step": 7829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154655945115233e-06,
+      "loss": 0.2518,
+      "step": 7830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152227554698814e-06,
+      "loss": 0.2718,
+      "step": 7831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149799537924749e-06,
+      "loss": 0.2728,
+      "step": 7832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14737189498016e-06,
+      "loss": 0.3082,
+      "step": 7833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.144944626052178e-06,
+      "loss": 0.2567,
+      "step": 7834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142517731327868e-06,
+      "loss": 0.2644,
+      "step": 7835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1400912109942915e-06,
+      "loss": 0.3019,
+      "step": 7836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137665065238471e-06,
+      "loss": 0.2584,
+      "step": 7837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1352392942474005e-06,
+      "loss": 0.2751,
+      "step": 7838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.132813898208053e-06,
+      "loss": 0.2803,
+      "step": 7839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130388877307353e-06,
+      "loss": 0.2812,
+      "step": 7840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127964231732223e-06,
+      "loss": 0.3066,
+      "step": 7841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1255399616695345e-06,
+      "loss": 0.2782,
+      "step": 7842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12311606730614e-06,
+      "loss": 0.2733,
+      "step": 7843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12069254882886e-06,
+      "loss": 0.2774,
+      "step": 7844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.118269406424492e-06,
+      "loss": 0.275,
+      "step": 7845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.115846640279798e-06,
+      "loss": 0.2661,
+      "step": 7846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113424250581505e-06,
+      "loss": 0.3121,
+      "step": 7847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111002237516334e-06,
+      "loss": 0.3069,
+      "step": 7848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108580601270947e-06,
+      "loss": 0.2819,
+      "step": 7849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106159342032e-06,
+      "loss": 0.272,
+      "step": 7850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10373845998611e-06,
+      "loss": 0.2482,
+      "step": 7851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101317955319866e-06,
+      "loss": 0.2791,
+      "step": 7852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098897828219831e-06,
+      "loss": 0.2651,
+      "step": 7853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096478078872528e-06,
+      "loss": 0.2767,
+      "step": 7854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.094058707464474e-06,
+      "loss": 0.2966,
+      "step": 7855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091639714182129e-06,
+      "loss": 0.2748,
+      "step": 7856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089221099211943e-06,
+      "loss": 0.3063,
+      "step": 7857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08680286274033e-06,
+      "loss": 0.253,
+      "step": 7858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084385004953674e-06,
+      "loss": 0.2944,
+      "step": 7859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081967526038334e-06,
+      "loss": 0.2981,
+      "step": 7860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.079550426180635e-06,
+      "loss": 0.2892,
+      "step": 7861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0771337055668826e-06,
+      "loss": 0.287,
+      "step": 7862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.074717364383335e-06,
+      "loss": 0.2927,
+      "step": 7863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072301402816231e-06,
+      "loss": 0.2622,
+      "step": 7864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069885821051796e-06,
+      "loss": 0.2474,
+      "step": 7865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.067470619276196e-06,
+      "loss": 0.2961,
+      "step": 7866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.06505579767559e-06,
+      "loss": 0.2751,
+      "step": 7867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.062641356436098e-06,
+      "loss": 0.3049,
+      "step": 7868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060227295743813e-06,
+      "loss": 0.2701,
+      "step": 7869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057813615784806e-06,
+      "loss": 0.5499,
+      "step": 7870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.055400316745096e-06,
+      "loss": 0.2789,
+      "step": 7871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052987398810706e-06,
+      "loss": 0.2699,
+      "step": 7872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0505748621676e-06,
+      "loss": 0.2661,
+      "step": 7873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048162707001727e-06,
+      "loss": 0.3229,
+      "step": 7874
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045750933499005e-06,
+      "loss": 0.2847,
+      "step": 7875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043339541845321e-06,
+      "loss": 0.295,
+      "step": 7876
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.040928532226539e-06,
+      "loss": 0.2955,
+      "step": 7877
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038517904828473e-06,
+      "loss": 0.2385,
+      "step": 7878
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03610765983694e-06,
+      "loss": 0.2499,
+      "step": 7879
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033697797437695e-06,
+      "loss": 0.2856,
+      "step": 7880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0312883178164875e-06,
+      "loss": 0.3056,
+      "step": 7881
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.028879221159025e-06,
+      "loss": 0.2779,
+      "step": 7882
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.026470507650988e-06,
+      "loss": 0.2896,
+      "step": 7883
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.024062177478033e-06,
+      "loss": 0.2896,
+      "step": 7884
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021654230825772e-06,
+      "loss": 0.2329,
+      "step": 7885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0192466678798116e-06,
+      "loss": 0.2853,
+      "step": 7886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016839488825703e-06,
+      "loss": 0.2591,
+      "step": 7887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014432693848985e-06,
+      "loss": 0.2347,
+      "step": 7888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0120262831351595e-06,
+      "loss": 0.2668,
+      "step": 7889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.009620256869703e-06,
+      "loss": 0.2458,
+      "step": 7890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007214615238061e-06,
+      "loss": 0.2581,
+      "step": 7891
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.004809358425639e-06,
+      "loss": 0.2809,
+      "step": 7892
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.002404486617839e-06,
+      "loss": 0.2753,
+      "step": 7893
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.2739,
+      "step": 7894
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9975958987574604e-06,
+      "loss": 0.3065,
+      "step": 7895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.995192183075509e-06,
+      "loss": 0.3094,
+      "step": 7896
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.992788853139414e-06,
+      "loss": 0.3174,
+      "step": 7897
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9903859091344175e-06,
+      "loss": 0.257,
+      "step": 7898
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987983351245713e-06,
+      "loss": 0.2749,
+      "step": 7899
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985581179658495e-06,
+      "loss": 0.2996,
+      "step": 7900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9831793945578995e-06,
+      "loss": 0.3027,
+      "step": 7901
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980777996129043e-06,
+      "loss": 0.283,
+      "step": 7902
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978376984557026e-06,
+      "loss": 0.2543,
+      "step": 7903
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975976360026894e-06,
+      "loss": 0.5981,
+      "step": 7904
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.97357612272368e-06,
+      "loss": 0.3047,
+      "step": 7905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971176272832382e-06,
+      "loss": 0.2738,
+      "step": 7906
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9687768105379685e-06,
+      "loss": 0.2667,
+      "step": 7907
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966377736025383e-06,
+      "loss": 0.284,
+      "step": 7908
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963979049479522e-06,
+      "loss": 0.2688,
+      "step": 7909
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9615807510852795e-06,
+      "loss": 0.287,
+      "step": 7910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959182841027494e-06,
+      "loss": 0.3025,
+      "step": 7911
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956785319490986e-06,
+      "loss": 0.2818,
+      "step": 7912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954388186660548e-06,
+      "loss": 0.2675,
+      "step": 7913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951991442720937e-06,
+      "loss": 0.2991,
+      "step": 7914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.949595087856887e-06,
+      "loss": 0.2449,
+      "step": 7915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947199122253083e-06,
+      "loss": 0.3584,
+      "step": 7916
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944803546094214e-06,
+      "loss": 0.2966,
+      "step": 7917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.942408359564906e-06,
+      "loss": 0.2963,
+      "step": 7918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940013562849769e-06,
+      "loss": 0.2566,
+      "step": 7919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.937619156133385e-06,
+      "loss": 0.2581,
+      "step": 7920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935225139600304e-06,
+      "loss": 0.2724,
+      "step": 7921
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932831513435045e-06,
+      "loss": 0.5986,
+      "step": 7922
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.93043827782209e-06,
+      "loss": 0.2997,
+      "step": 7923
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928045432945909e-06,
+      "loss": 0.2913,
+      "step": 7924
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925652978990921e-06,
+      "loss": 0.2437,
+      "step": 7925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92326091614153e-06,
+      "loss": 0.576,
+      "step": 7926
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.920869244582102e-06,
+      "loss": 0.2755,
+      "step": 7927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.918477964496975e-06,
+      "loss": 0.2682,
+      "step": 7928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.916087076070462e-06,
+      "loss": 0.2874,
+      "step": 7929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913696579486829e-06,
+      "loss": 0.25,
+      "step": 7930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91130647493034e-06,
+      "loss": 0.2695,
+      "step": 7931
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9089167625852e-06,
+      "loss": 0.2504,
+      "step": 7932
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906527442635599e-06,
+      "loss": 0.2741,
+      "step": 7933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904138515265696e-06,
+      "loss": 0.2186,
+      "step": 7934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.901749980659617e-06,
+      "loss": 0.2565,
+      "step": 7935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.899361839001462e-06,
+      "loss": 0.2446,
+      "step": 7936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.896974090475286e-06,
+      "loss": 0.2684,
+      "step": 7937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.89458673526514e-06,
+      "loss": 0.2877,
+      "step": 7938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892199773555018e-06,
+      "loss": 0.6144,
+      "step": 7939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889813205528895e-06,
+      "loss": 0.2814,
+      "step": 7940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887427031370727e-06,
+      "loss": 0.2676,
+      "step": 7941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885041251264419e-06,
+      "loss": 0.2526,
+      "step": 7942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.882655865393856e-06,
+      "loss": 0.2839,
+      "step": 7943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.880270873942895e-06,
+      "loss": 0.2771,
+      "step": 7944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.877886277095356e-06,
+      "loss": 0.3147,
+      "step": 7945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875502075035039e-06,
+      "loss": 0.3027,
+      "step": 7946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.873118267945692e-06,
+      "loss": 0.2527,
+      "step": 7947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.870734856011066e-06,
+      "loss": 0.3033,
+      "step": 7948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868351839414848e-06,
+      "loss": 0.2677,
+      "step": 7949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8659692183407135e-06,
+      "loss": 0.3055,
+      "step": 7950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.863586992972304e-06,
+      "loss": 0.2398,
+      "step": 7951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861205163493229e-06,
+      "loss": 0.2754,
+      "step": 7952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.858823730087072e-06,
+      "loss": 0.3032,
+      "step": 7953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856442692937372e-06,
+      "loss": 0.302,
+      "step": 7954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854062052227662e-06,
+      "loss": 0.2488,
+      "step": 7955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851681808141418e-06,
+      "loss": 0.2108,
+      "step": 7956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.849301960862101e-06,
+      "loss": 0.2521,
+      "step": 7957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.846922510573139e-06,
+      "loss": 0.2726,
+      "step": 7958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8445434574579275e-06,
+      "loss": 0.3016,
+      "step": 7959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.842164801699836e-06,
+      "loss": 0.254,
+      "step": 7960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8397865434821886e-06,
+      "loss": 0.294,
+      "step": 7961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837408682988305e-06,
+      "loss": 0.2191,
+      "step": 7962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8350312204014475e-06,
+      "loss": 0.3165,
+      "step": 7963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.832654155904863e-06,
+      "loss": 0.2407,
+      "step": 7964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830277489681762e-06,
+      "loss": 0.2639,
+      "step": 7965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8279012219153284e-06,
+      "loss": 0.3093,
+      "step": 7966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.825525352788716e-06,
+      "loss": 0.2513,
+      "step": 7967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8231498824850356e-06,
+      "loss": 0.2851,
+      "step": 7968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820774811187389e-06,
+      "loss": 0.2749,
+      "step": 7969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818400139078824e-06,
+      "loss": 0.271,
+      "step": 7970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.816025866342374e-06,
+      "loss": 0.2677,
+      "step": 7971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.813651993161036e-06,
+      "loss": 0.2302,
+      "step": 7972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811278519717775e-06,
+      "loss": 0.2811,
+      "step": 7973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808905446195532e-06,
+      "loss": 0.2969,
+      "step": 7974
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8065327727772005e-06,
+      "loss": 0.3164,
+      "step": 7975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.804160499645667e-06,
+      "loss": 0.2782,
+      "step": 7976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8017886269837665e-06,
+      "loss": 0.2904,
+      "step": 7977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7994171549743085e-06,
+      "loss": 0.2961,
+      "step": 7978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.797046083800087e-06,
+      "loss": 0.3427,
+      "step": 7979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794675413643842e-06,
+      "loss": 0.2944,
+      "step": 7980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7923051446883e-06,
+      "loss": 0.278,
+      "step": 7981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7899352771161355e-06,
+      "loss": 0.2711,
+      "step": 7982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.787565811110022e-06,
+      "loss": 0.306,
+      "step": 7983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785196746852584e-06,
+      "loss": 0.2742,
+      "step": 7984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7828280845264056e-06,
+      "loss": 0.2942,
+      "step": 7985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7804598243140664e-06,
+      "loss": 0.2606,
+      "step": 7986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778091966398091e-06,
+      "loss": 0.2525,
+      "step": 7987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.775724510960984e-06,
+      "loss": 0.2756,
+      "step": 7988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7733574581852185e-06,
+      "loss": 0.274,
+      "step": 7989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.770990808253234e-06,
+      "loss": 0.2756,
+      "step": 7990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7686245613474445e-06,
+      "loss": 0.2682,
+      "step": 7991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.766258717650218e-06,
+      "loss": 0.2888,
+      "step": 7992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.763893277343915e-06,
+      "loss": 0.2914,
+      "step": 7993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.761528240610842e-06,
+      "loss": 0.608,
+      "step": 7994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.759163607633289e-06,
+      "loss": 0.2693,
+      "step": 7995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756799378593509e-06,
+      "loss": 0.2768,
+      "step": 7996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754435553673726e-06,
+      "loss": 0.2953,
+      "step": 7997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.752072133056135e-06,
+      "loss": 0.2979,
+      "step": 7998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.749709116922886e-06,
+      "loss": 0.2753,
+      "step": 7999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747346505456123e-06,
+      "loss": 0.2813,
+      "step": 8000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.744984298837933e-06,
+      "loss": 0.2518,
+      "step": 8001
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742622497250389e-06,
+      "loss": 0.3054,
+      "step": 8002
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.740261100875524e-06,
+      "loss": 0.2849,
+      "step": 8003
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737900109895345e-06,
+      "loss": 0.2591,
+      "step": 8004
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.735539524491828e-06,
+      "loss": 0.2394,
+      "step": 8005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7331793448469045e-06,
+      "loss": 0.2416,
+      "step": 8006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7308195711425e-06,
+      "loss": 0.2706,
+      "step": 8007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7284602035604845e-06,
+      "loss": 0.2851,
+      "step": 8008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726101242282708e-06,
+      "loss": 0.2498,
+      "step": 8009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723742687490988e-06,
+      "loss": 0.3086,
+      "step": 8010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7213845393671136e-06,
+      "loss": 0.2362,
+      "step": 8011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719026798092838e-06,
+      "loss": 0.2952,
+      "step": 8012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7166694638498755e-06,
+      "loss": 0.2143,
+      "step": 8013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7143125368199335e-06,
+      "loss": 0.2373,
+      "step": 8014
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.71195601718466e-06,
+      "loss": 0.2871,
+      "step": 8015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.709599905125683e-06,
+      "loss": 0.2692,
+      "step": 8016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7072442008246135e-06,
+      "loss": 0.2756,
+      "step": 8017
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.704888904463003e-06,
+      "loss": 0.252,
+      "step": 8018
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702534016222398e-06,
+      "loss": 0.2737,
+      "step": 8019
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700179536284286e-06,
+      "loss": 0.2926,
+      "step": 8020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697825464830153e-06,
+      "loss": 0.2598,
+      "step": 8021
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695471802041437e-06,
+      "loss": 0.2375,
+      "step": 8022
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693118548099538e-06,
+      "loss": 0.2793,
+      "step": 8023
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.690765703185846e-06,
+      "loss": 0.3109,
+      "step": 8024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6884132674816964e-06,
+      "loss": 0.2486,
+      "step": 8025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686061241168406e-06,
+      "loss": 0.2942,
+      "step": 8026
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6837096244272586e-06,
+      "loss": 0.2729,
+      "step": 8027
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.681358417439505e-06,
+      "loss": 0.5831,
+      "step": 8028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.679007620386369e-06,
+      "loss": 0.3109,
+      "step": 8029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.676657233449025e-06,
+      "loss": 0.2721,
+      "step": 8030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674307256808646e-06,
+      "loss": 0.2817,
+      "step": 8031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6719576906463445e-06,
+      "loss": 0.2647,
+      "step": 8032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669608535143218e-06,
+      "loss": 0.3099,
+      "step": 8033
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667259790480327e-06,
+      "loss": 0.3391,
+      "step": 8034
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6649114568387024e-06,
+      "loss": 0.2907,
+      "step": 8035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.662563534399345e-06,
+      "loss": 0.2653,
+      "step": 8036
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6602160233432085e-06,
+      "loss": 0.2228,
+      "step": 8037
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657868923851244e-06,
+      "loss": 0.252,
+      "step": 8038
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655522236104344e-06,
+      "loss": 0.2551,
+      "step": 8039
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6531759602833815e-06,
+      "loss": 0.345,
+      "step": 8040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.650830096569196e-06,
+      "loss": 0.2882,
+      "step": 8041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648484645142597e-06,
+      "loss": 0.2986,
+      "step": 8042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.646139606184362e-06,
+      "loss": 0.2451,
+      "step": 8043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.643794979875225e-06,
+      "loss": 0.2682,
+      "step": 8044
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6414507663959115e-06,
+      "loss": 0.286,
+      "step": 8045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639106965927093e-06,
+      "loss": 0.2797,
+      "step": 8046
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.636763578649419e-06,
+      "loss": 0.3387,
+      "step": 8047
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634420604743509e-06,
+      "loss": 0.2574,
+      "step": 8048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.632078044389947e-06,
+      "loss": 0.2545,
+      "step": 8049
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629735897769289e-06,
+      "loss": 0.3049,
+      "step": 8050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.627394165062045e-06,
+      "loss": 0.2468,
+      "step": 8051
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.62505284644872e-06,
+      "loss": 0.3224,
+      "step": 8052
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622711942109759e-06,
+      "loss": 0.2697,
+      "step": 8053
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.620371452225587e-06,
+      "loss": 0.2632,
+      "step": 8054
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.61803137697661e-06,
+      "loss": 0.2638,
+      "step": 8055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.615691716543179e-06,
+      "loss": 0.2703,
+      "step": 8056
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613352471105627e-06,
+      "loss": 0.2521,
+      "step": 8057
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611013640844245e-06,
+      "loss": 0.2441,
+      "step": 8058
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.608675225939308e-06,
+      "loss": 0.2761,
+      "step": 8059
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606337226571047e-06,
+      "loss": 0.2935,
+      "step": 8060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6039996429196555e-06,
+      "loss": 0.285,
+      "step": 8061
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.601662475165316e-06,
+      "loss": 0.2796,
+      "step": 8062
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5993257234881565e-06,
+      "loss": 0.2687,
+      "step": 8063
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596989388068283e-06,
+      "loss": 0.2836,
+      "step": 8064
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5946534690857705e-06,
+      "loss": 0.2661,
+      "step": 8065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592317966720661e-06,
+      "loss": 0.2847,
+      "step": 8066
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.589982881152966e-06,
+      "loss": 0.3113,
+      "step": 8067
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587648212562651e-06,
+      "loss": 0.2533,
+      "step": 8068
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585313961129676e-06,
+      "loss": 0.2955,
+      "step": 8069
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582980127033943e-06,
+      "loss": 0.252,
+      "step": 8070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5806467104553345e-06,
+      "loss": 0.2835,
+      "step": 8071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5783137115737e-06,
+      "loss": 0.2668,
+      "step": 8072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575981130568856e-06,
+      "loss": 0.3032,
+      "step": 8073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573648967620589e-06,
+      "loss": 0.2485,
+      "step": 8074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57131722290864e-06,
+      "loss": 0.312,
+      "step": 8075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568985896612742e-06,
+      "loss": 0.2449,
+      "step": 8076
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5666549889125726e-06,
+      "loss": 0.2367,
+      "step": 8077
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.56432449998779e-06,
+      "loss": 0.2734,
+      "step": 8078
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561994430018016e-06,
+      "loss": 0.2636,
+      "step": 8079
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559664779182842e-06,
+      "loss": 0.2562,
+      "step": 8080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.557335547661828e-06,
+      "loss": 0.2773,
+      "step": 8081
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.55500673563449e-06,
+      "loss": 0.2729,
+      "step": 8082
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.552678343280337e-06,
+      "loss": 0.2978,
+      "step": 8083
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.550350370778815e-06,
+      "loss": 0.2931,
+      "step": 8084
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548022818309361e-06,
+      "loss": 0.2603,
+      "step": 8085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545695686051369e-06,
+      "loss": 0.2605,
+      "step": 8086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5433689741842024e-06,
+      "loss": 0.2934,
+      "step": 8087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5410426828871965e-06,
+      "loss": 0.2527,
+      "step": 8088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5387168123396406e-06,
+      "loss": 0.6045,
+      "step": 8089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536391362720816e-06,
+      "loss": 0.2437,
+      "step": 8090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5340663342099435e-06,
+      "loss": 0.2952,
+      "step": 8091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.531741726986226e-06,
+      "loss": 0.2439,
+      "step": 8092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529417541228846e-06,
+      "loss": 0.2712,
+      "step": 8093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527093777116925e-06,
+      "loss": 0.2842,
+      "step": 8094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5247704348295785e-06,
+      "loss": 0.3024,
+      "step": 8095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522447514545865e-06,
+      "loss": 0.2585,
+      "step": 8096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520125016444835e-06,
+      "loss": 0.2802,
+      "step": 8097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5178029407054965e-06,
+      "loss": 0.2325,
+      "step": 8098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.515481287506811e-06,
+      "loss": 0.3073,
+      "step": 8099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.513160057027736e-06,
+      "loss": 0.2603,
+      "step": 8100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.510839249447169e-06,
+      "loss": 0.3051,
+      "step": 8101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508518864943989e-06,
+      "loss": 0.3133,
+      "step": 8102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506198903697041e-06,
+      "loss": 0.2342,
+      "step": 8103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5038793658851365e-06,
+      "loss": 0.3011,
+      "step": 8104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501560251687056e-06,
+      "loss": 0.2777,
+      "step": 8105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4992415612815355e-06,
+      "loss": 0.2684,
+      "step": 8106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496923294847303e-06,
+      "loss": 0.2896,
+      "step": 8107
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.494605452563028e-06,
+      "loss": 0.3073,
+      "step": 8108
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492288034607361e-06,
+      "loss": 0.2422,
+      "step": 8109
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.489971041158919e-06,
+      "loss": 0.2971,
+      "step": 8110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487654472396284e-06,
+      "loss": 0.251,
+      "step": 8111
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4853383284980064e-06,
+      "loss": 0.3059,
+      "step": 8112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.483022609642596e-06,
+      "loss": 0.2935,
+      "step": 8113
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.480707316008549e-06,
+      "loss": 0.2644,
+      "step": 8114
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478392447774307e-06,
+      "loss": 0.2598,
+      "step": 8115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.476078005118293e-06,
+      "loss": 0.2622,
+      "step": 8116
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.473763988218891e-06,
+      "loss": 0.299,
+      "step": 8117
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4714503972544545e-06,
+      "loss": 0.289,
+      "step": 8118
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.469137232403308e-06,
+      "loss": 0.2335,
+      "step": 8119
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466824493843728e-06,
+      "loss": 0.2941,
+      "step": 8120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.464512181753982e-06,
+      "loss": 0.257,
+      "step": 8121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.462200296312284e-06,
+      "loss": 0.291,
+      "step": 8122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.459888837696822e-06,
+      "loss": 0.2661,
+      "step": 8123
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457577806085754e-06,
+      "loss": 0.2953,
+      "step": 8124
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.455267201657203e-06,
+      "loss": 0.2645,
+      "step": 8125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4529570245892625e-06,
+      "loss": 0.2548,
+      "step": 8126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.450647275059979e-06,
+      "loss": 0.2951,
+      "step": 8127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4483379532473906e-06,
+      "loss": 0.2943,
+      "step": 8128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.446029059329477e-06,
+      "loss": 0.3195,
+      "step": 8129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443720593484198e-06,
+      "loss": 0.2298,
+      "step": 8130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441412555889487e-06,
+      "loss": 0.2728,
+      "step": 8131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439104946723228e-06,
+      "loss": 0.2642,
+      "step": 8132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436797766163285e-06,
+      "loss": 0.2825,
+      "step": 8133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4344910143874755e-06,
+      "loss": 0.3176,
+      "step": 8134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.432184691573602e-06,
+      "loss": 0.2866,
+      "step": 8135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429878797899424e-06,
+      "loss": 0.2526,
+      "step": 8136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.42757333354266e-06,
+      "loss": 0.2643,
+      "step": 8137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425268298681015e-06,
+      "loss": 0.2716,
+      "step": 8138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422963693492141e-06,
+      "loss": 0.2578,
+      "step": 8139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420659518153667e-06,
+      "loss": 0.3008,
+      "step": 8140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.41835577284319e-06,
+      "loss": 0.2524,
+      "step": 8141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416052457738271e-06,
+      "loss": 0.2606,
+      "step": 8142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.41374957301644e-06,
+      "loss": 0.2066,
+      "step": 8143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.411447118855183e-06,
+      "loss": 0.2757,
+      "step": 8144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.409145095431976e-06,
+      "loss": 0.2831,
+      "step": 8145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406843502924235e-06,
+      "loss": 0.3225,
+      "step": 8146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40454234150936e-06,
+      "loss": 0.5675,
+      "step": 8147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402241611364715e-06,
+      "loss": 0.2445,
+      "step": 8148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399941312667626e-06,
+      "loss": 0.2208,
+      "step": 8149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397641445595393e-06,
+      "loss": 0.28,
+      "step": 8150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39534201032527e-06,
+      "loss": 0.2852,
+      "step": 8151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.393043007034496e-06,
+      "loss": 0.3055,
+      "step": 8152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390744435900262e-06,
+      "loss": 0.251,
+      "step": 8153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388446297099728e-06,
+      "loss": 0.3126,
+      "step": 8154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386148590810027e-06,
+      "loss": 0.3075,
+      "step": 8155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383851317208253e-06,
+      "loss": 0.2547,
+      "step": 8156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381554476471473e-06,
+      "loss": 0.2348,
+      "step": 8157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379258068776706e-06,
+      "loss": 0.2521,
+      "step": 8158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3769620943009615e-06,
+      "loss": 0.3021,
+      "step": 8159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374666553221191e-06,
+      "loss": 0.2736,
+      "step": 8160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372371445714325e-06,
+      "loss": 0.262,
+      "step": 8161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370076771957264e-06,
+      "loss": 0.3145,
+      "step": 8162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367782532126864e-06,
+      "loss": 0.3156,
+      "step": 8163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365488726399962e-06,
+      "loss": 0.2725,
+      "step": 8164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363195354953341e-06,
+      "loss": 0.2673,
+      "step": 8165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.360902417963777e-06,
+      "loss": 0.2756,
+      "step": 8166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358609915607987e-06,
+      "loss": 0.2993,
+      "step": 8167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3563178480626665e-06,
+      "loss": 0.2783,
+      "step": 8168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354026215504488e-06,
+      "loss": 0.2764,
+      "step": 8169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351735018110066e-06,
+      "loss": 0.2222,
+      "step": 8170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349444256056005e-06,
+      "loss": 0.2797,
+      "step": 8171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347153929518852e-06,
+      "loss": 0.244,
+      "step": 8172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344864038675147e-06,
+      "loss": 0.2549,
+      "step": 8173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342574583701382e-06,
+      "loss": 0.2751,
+      "step": 8174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340285564774007e-06,
+      "loss": 0.2625,
+      "step": 8175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3379969820694636e-06,
+      "loss": 0.2565,
+      "step": 8176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335708835764131e-06,
+      "loss": 0.2946,
+      "step": 8177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333421126034374e-06,
+      "loss": 0.2819,
+      "step": 8178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331133853056516e-06,
+      "loss": 0.2598,
+      "step": 8179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3288470170068505e-06,
+      "loss": 0.2549,
+      "step": 8180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326560618061639e-06,
+      "loss": 0.2265,
+      "step": 8181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324274656397095e-06,
+      "loss": 0.2296,
+      "step": 8182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.321989132189422e-06,
+      "loss": 0.2867,
+      "step": 8183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319704045614768e-06,
+      "loss": 0.2393,
+      "step": 8184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317419396849258e-06,
+      "loss": 0.304,
+      "step": 8185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315135186068984e-06,
+      "loss": 0.2891,
+      "step": 8186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312851413449999e-06,
+      "loss": 0.2687,
+      "step": 8187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31056807916833e-06,
+      "loss": 0.2867,
+      "step": 8188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308285183399954e-06,
+      "loss": 0.2703,
+      "step": 8189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306002726320839e-06,
+      "loss": 0.3225,
+      "step": 8190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3037207081068965e-06,
+      "loss": 0.2919,
+      "step": 8191
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.301439128934015e-06,
+      "loss": 0.2985,
+      "step": 8192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299157988978048e-06,
+      "loss": 0.297,
+      "step": 8193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296877288414815e-06,
+      "loss": 0.6129,
+      "step": 8194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2945970274201045e-06,
+      "loss": 0.2766,
+      "step": 8195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292317206169655e-06,
+      "loss": 0.2724,
+      "step": 8196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290037824839202e-06,
+      "loss": 0.2853,
+      "step": 8197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287758883604415e-06,
+      "loss": 0.2875,
+      "step": 8198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.285480382640947e-06,
+      "loss": 0.2808,
+      "step": 8199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283202322124417e-06,
+      "loss": 0.2679,
+      "step": 8200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280924702230403e-06,
+      "loss": 0.2629,
+      "step": 8201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.278647523134459e-06,
+      "loss": 0.3494,
+      "step": 8202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276370785012086e-06,
+      "loss": 0.2757,
+      "step": 8203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2740944880387795e-06,
+      "loss": 0.262,
+      "step": 8204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.271818632389973e-06,
+      "loss": 0.28,
+      "step": 8205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.269543218241079e-06,
+      "loss": 0.2706,
+      "step": 8206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267268245767486e-06,
+      "loss": 0.2501,
+      "step": 8207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2649937151445275e-06,
+      "loss": 0.2593,
+      "step": 8208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262719626547519e-06,
+      "loss": 0.2443,
+      "step": 8209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.260445980151725e-06,
+      "loss": 0.2626,
+      "step": 8210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2581727761324e-06,
+      "loss": 0.2586,
+      "step": 8211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2559000146647485e-06,
+      "loss": 0.2744,
+      "step": 8212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253627695923934e-06,
+      "loss": 0.2627,
+      "step": 8213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2513558200851115e-06,
+      "loss": 0.2453,
+      "step": 8214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.249084387323373e-06,
+      "loss": 0.3441,
+      "step": 8215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2468133978137945e-06,
+      "loss": 0.5818,
+      "step": 8216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2445428517314116e-06,
+      "loss": 0.2794,
+      "step": 8217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242272749251228e-06,
+      "loss": 0.3177,
+      "step": 8218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240003090548213e-06,
+      "loss": 0.304,
+      "step": 8219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237733875797293e-06,
+      "loss": 0.2581,
+      "step": 8220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2354651051733795e-06,
+      "loss": 0.3061,
+      "step": 8221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2331967788513295e-06,
+      "loss": 0.3313,
+      "step": 8222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230928897005978e-06,
+      "loss": 0.3079,
+      "step": 8223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.22866145981212e-06,
+      "loss": 0.2622,
+      "step": 8224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.22639446744452e-06,
+      "loss": 0.2564,
+      "step": 8225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2241279200779105e-06,
+      "loss": 0.2591,
+      "step": 8226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221861817886973e-06,
+      "loss": 0.2531,
+      "step": 8227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2195961610463845e-06,
+      "loss": 0.2885,
+      "step": 8228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.217330949730758e-06,
+      "loss": 0.2784,
+      "step": 8229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.215066184114689e-06,
+      "loss": 0.3102,
+      "step": 8230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212801864372734e-06,
+      "loss": 0.5885,
+      "step": 8231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210537990679417e-06,
+      "loss": 0.2969,
+      "step": 8232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208274563209227e-06,
+      "loss": 0.2432,
+      "step": 8233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2060115821366085e-06,
+      "loss": 0.2612,
+      "step": 8234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203749047635998e-06,
+      "loss": 0.5912,
+      "step": 8235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201486959881766e-06,
+      "loss": 0.3296,
+      "step": 8236
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199225319048267e-06,
+      "loss": 0.2829,
+      "step": 8237
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196964125309818e-06,
+      "loss": 0.2664,
+      "step": 8238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194703378840701e-06,
+      "loss": 0.257,
+      "step": 8239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192443079815166e-06,
+      "loss": 0.2762,
+      "step": 8240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190183228407416e-06,
+      "loss": 0.2682,
+      "step": 8241
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187923824791642e-06,
+      "loss": 0.2657,
+      "step": 8242
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.185664869141979e-06,
+      "loss": 0.2876,
+      "step": 8243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183406361632534e-06,
+      "loss": 0.2501,
+      "step": 8244
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181148302437392e-06,
+      "loss": 0.2637,
+      "step": 8245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.178890691730585e-06,
+      "loss": 0.2749,
+      "step": 8246
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176633529686124e-06,
+      "loss": 0.2677,
+      "step": 8247
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174376816477969e-06,
+      "loss": 0.2653,
+      "step": 8248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172120552280067e-06,
+      "loss": 0.2803,
+      "step": 8249
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169864737266321e-06,
+      "loss": 0.2349,
+      "step": 8250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1676093716105845e-06,
+      "loss": 0.2369,
+      "step": 8251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165354455486707e-06,
+      "loss": 0.2574,
+      "step": 8252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163099989068476e-06,
+      "loss": 0.234,
+      "step": 8253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160845972529656e-06,
+      "loss": 0.2217,
+      "step": 8254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1585924060439755e-06,
+      "loss": 0.2908,
+      "step": 8255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156339289785129e-06,
+      "loss": 0.2633,
+      "step": 8256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154086623926781e-06,
+      "loss": 0.2374,
+      "step": 8257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151834408642542e-06,
+      "loss": 0.261,
+      "step": 8258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149582644106018e-06,
+      "loss": 0.2505,
+      "step": 8259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147331330490752e-06,
+      "loss": 0.2325,
+      "step": 8260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1450804679702685e-06,
+      "loss": 0.2436,
+      "step": 8261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142830056718052e-06,
+      "loss": 0.2469,
+      "step": 8262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140580096907554e-06,
+      "loss": 0.2831,
+      "step": 8263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138330588712194e-06,
+      "loss": 0.29,
+      "step": 8264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1360815323053406e-06,
+      "loss": 0.306,
+      "step": 8265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.133832927860356e-06,
+      "loss": 0.2546,
+      "step": 8266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.13158477555054e-06,
+      "loss": 0.25,
+      "step": 8267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1293370755491725e-06,
+      "loss": 0.2916,
+      "step": 8268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.127089828029496e-06,
+      "loss": 0.3303,
+      "step": 8269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124843033164716e-06,
+      "loss": 0.3043,
+      "step": 8270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122596691128009e-06,
+      "loss": 0.2563,
+      "step": 8271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.120350802092501e-06,
+      "loss": 0.3027,
+      "step": 8272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1181053662313075e-06,
+      "loss": 0.2294,
+      "step": 8273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.115860383717486e-06,
+      "loss": 0.2878,
+      "step": 8274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.113615854724071e-06,
+      "loss": 0.2259,
+      "step": 8275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1113717794240615e-06,
+      "loss": 0.2839,
+      "step": 8276
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109128157990418e-06,
+      "loss": 0.2674,
+      "step": 8277
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.106884990596073e-06,
+      "loss": 0.2569,
+      "step": 8278
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1046422774139065e-06,
+      "loss": 0.2555,
+      "step": 8279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.10240001861679e-06,
+      "loss": 0.2634,
+      "step": 8280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100158214377536e-06,
+      "loss": 0.2503,
+      "step": 8281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097916864868932e-06,
+      "loss": 0.2673,
+      "step": 8282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.095675970263738e-06,
+      "loss": 0.265,
+      "step": 8283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093435530734664e-06,
+      "loss": 0.2879,
+      "step": 8284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091195546454398e-06,
+      "loss": 0.3255,
+      "step": 8285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088956017595575e-06,
+      "loss": 0.3093,
+      "step": 8286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0867169443308196e-06,
+      "loss": 0.2482,
+      "step": 8287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.084478326832706e-06,
+      "loss": 0.2867,
+      "step": 8288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082240165273767e-06,
+      "loss": 0.2518,
+      "step": 8289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080002459826523e-06,
+      "loss": 0.2887,
+      "step": 8290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0777652106634334e-06,
+      "loss": 0.3162,
+      "step": 8291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.07552841795694e-06,
+      "loss": 0.2935,
+      "step": 8292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.073292081879442e-06,
+      "loss": 0.3007,
+      "step": 8293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071056202603305e-06,
+      "loss": 0.2941,
+      "step": 8294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068820780300864e-06,
+      "loss": 0.283,
+      "step": 8295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.066585815144404e-06,
+      "loss": 0.2927,
+      "step": 8296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0643513073061966e-06,
+      "loss": 0.2428,
+      "step": 8297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.06211725695846e-06,
+      "loss": 0.2556,
+      "step": 8298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.059883664273385e-06,
+      "loss": 0.2646,
+      "step": 8299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.057650529423126e-06,
+      "loss": 0.2399,
+      "step": 8300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055417852579802e-06,
+      "loss": 0.288,
+      "step": 8301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053185633915501e-06,
+      "loss": 0.305,
+      "step": 8302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050953873602259e-06,
+      "loss": 0.3251,
+      "step": 8303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048722571812105e-06,
+      "loss": 0.2581,
+      "step": 8304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0464917287170055e-06,
+      "loss": 0.2963,
+      "step": 8305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0442613444889065e-06,
+      "loss": 0.2631,
+      "step": 8306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042031419299714e-06,
+      "loss": 0.2841,
+      "step": 8307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039801953321302e-06,
+      "loss": 0.3099,
+      "step": 8308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0375729467255074e-06,
+      "loss": 0.2529,
+      "step": 8309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.035344399684124e-06,
+      "loss": 0.2891,
+      "step": 8310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033116312368926e-06,
+      "loss": 0.2452,
+      "step": 8311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030888684951638e-06,
+      "loss": 0.2714,
+      "step": 8312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028661517603956e-06,
+      "loss": 0.2919,
+      "step": 8313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026434810497538e-06,
+      "loss": 0.2911,
+      "step": 8314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024208563804008e-06,
+      "loss": 0.2867,
+      "step": 8315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02198277769496e-06,
+      "loss": 0.253,
+      "step": 8316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019757452341934e-06,
+      "loss": 0.2496,
+      "step": 8317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017532587916461e-06,
+      "loss": 0.2726,
+      "step": 8318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015308184590011e-06,
+      "loss": 0.2758,
+      "step": 8319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013084242534032e-06,
+      "loss": 0.2532,
+      "step": 8320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010860761919946e-06,
+      "loss": 0.2753,
+      "step": 8321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.008637742919114e-06,
+      "loss": 0.2874,
+      "step": 8322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.006415185702885e-06,
+      "loss": 0.2992,
+      "step": 8323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004193090442551e-06,
+      "loss": 0.2872,
+      "step": 8324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001971457309391e-06,
+      "loss": 0.2856,
+      "step": 8325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999750286474637e-06,
+      "loss": 0.2642,
+      "step": 8326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997529578109476e-06,
+      "loss": 0.2352,
+      "step": 8327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995309332385083e-06,
+      "loss": 0.2813,
+      "step": 8328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.993089549472574e-06,
+      "loss": 0.3192,
+      "step": 8329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99087022954304e-06,
+      "loss": 0.3025,
+      "step": 8330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.988651372767538e-06,
+      "loss": 0.2973,
+      "step": 8331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986432979317085e-06,
+      "loss": 0.3057,
+      "step": 8332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984215049362667e-06,
+      "loss": 0.2794,
+      "step": 8333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.981997583075222e-06,
+      "loss": 0.3143,
+      "step": 8334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979780580625674e-06,
+      "loss": 0.2734,
+      "step": 8335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977564042184888e-06,
+      "loss": 0.2735,
+      "step": 8336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975347967923708e-06,
+      "loss": 0.2598,
+      "step": 8337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973132358012939e-06,
+      "loss": 0.3055,
+      "step": 8338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970917212623347e-06,
+      "loss": 0.2514,
+      "step": 8339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.96870253192567e-06,
+      "loss": 0.2948,
+      "step": 8340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.966488316090593e-06,
+      "loss": 0.2532,
+      "step": 8341
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964274565288792e-06,
+      "loss": 0.283,
+      "step": 8342
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9620612796908794e-06,
+      "loss": 0.2628,
+      "step": 8343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95984845946745e-06,
+      "loss": 0.3083,
+      "step": 8344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957636104789056e-06,
+      "loss": 0.2852,
+      "step": 8345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9554242158262134e-06,
+      "loss": 0.2607,
+      "step": 8346
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95321279274941e-06,
+      "loss": 0.312,
+      "step": 8347
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.951001835729079e-06,
+      "loss": 0.2365,
+      "step": 8348
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9487913449356454e-06,
+      "loss": 0.2803,
+      "step": 8349
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94658132053947e-06,
+      "loss": 0.2827,
+      "step": 8350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.944371762710897e-06,
+      "loss": 0.3231,
+      "step": 8351
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942162671620225e-06,
+      "loss": 0.3166,
+      "step": 8352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939954047437723e-06,
+      "loss": 0.2458,
+      "step": 8353
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937745890333623e-06,
+      "loss": 0.2772,
+      "step": 8354
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935538200478108e-06,
+      "loss": 0.2287,
+      "step": 8355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.933330978041351e-06,
+      "loss": 0.2947,
+      "step": 8356
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931124223193461e-06,
+      "loss": 0.2973,
+      "step": 8357
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928917936104529e-06,
+      "loss": 0.265,
+      "step": 8358
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9267121169446056e-06,
+      "loss": 0.2961,
+      "step": 8359
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924506765883701e-06,
+      "loss": 0.2796,
+      "step": 8360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9223018830918004e-06,
+      "loss": 0.2588,
+      "step": 8361
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920097468738833e-06,
+      "loss": 0.3008,
+      "step": 8362
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917893522994713e-06,
+      "loss": 0.2415,
+      "step": 8363
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9156900460293126e-06,
+      "loss": 0.5614,
+      "step": 8364
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913487038012451e-06,
+      "loss": 0.2741,
+      "step": 8365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911284499113943e-06,
+      "loss": 0.237,
+      "step": 8366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909082429503537e-06,
+      "loss": 0.2568,
+      "step": 8367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.906880829350961e-06,
+      "loss": 0.2489,
+      "step": 8368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.904679698825903e-06,
+      "loss": 0.2836,
+      "step": 8369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902479038098017e-06,
+      "loss": 0.2688,
+      "step": 8370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90027884733692e-06,
+      "loss": 0.3223,
+      "step": 8371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898079126712184e-06,
+      "loss": 0.2689,
+      "step": 8372
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895879876393366e-06,
+      "loss": 0.2696,
+      "step": 8373
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893681096549961e-06,
+      "loss": 0.2723,
+      "step": 8374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.891482787351446e-06,
+      "loss": 0.2868,
+      "step": 8375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889284948967253e-06,
+      "loss": 0.2629,
+      "step": 8376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.887087581566784e-06,
+      "loss": 0.2762,
+      "step": 8377
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884890685319402e-06,
+      "loss": 0.2468,
+      "step": 8378
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.882694260394424e-06,
+      "loss": 0.282,
+      "step": 8379
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880498306961153e-06,
+      "loss": 0.2477,
+      "step": 8380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.87830282518883e-06,
+      "loss": 0.2911,
+      "step": 8381
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.876107815246678e-06,
+      "loss": 0.3041,
+      "step": 8382
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8739132773038765e-06,
+      "loss": 0.2927,
+      "step": 8383
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.871719211529571e-06,
+      "loss": 0.2674,
+      "step": 8384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.86952561809287e-06,
+      "loss": 0.2781,
+      "step": 8385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.867332497162836e-06,
+      "loss": 0.2375,
+      "step": 8386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8651398489085176e-06,
+      "loss": 0.278,
+      "step": 8387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862947673498904e-06,
+      "loss": 0.3192,
+      "step": 8388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8607559711029586e-06,
+      "loss": 0.2703,
+      "step": 8389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858564741889608e-06,
+      "loss": 0.2401,
+      "step": 8390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8563739860277415e-06,
+      "loss": 0.282,
+      "step": 8391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854183703686216e-06,
+      "loss": 0.2543,
+      "step": 8392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851993895033836e-06,
+      "loss": 0.2818,
+      "step": 8393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849804560239394e-06,
+      "loss": 0.2881,
+      "step": 8394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847615699471625e-06,
+      "loss": 0.2687,
+      "step": 8395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845427312899238e-06,
+      "loss": 0.2484,
+      "step": 8396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.843239400690903e-06,
+      "loss": 0.5842,
+      "step": 8397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.841051963015254e-06,
+      "loss": 0.2821,
+      "step": 8398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83886500004089e-06,
+      "loss": 0.2713,
+      "step": 8399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8366785119363624e-06,
+      "loss": 0.2335,
+      "step": 8400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.834492498870205e-06,
+      "loss": 0.2289,
+      "step": 8401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8323069610109046e-06,
+      "loss": 0.2355,
+      "step": 8402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830121898526901e-06,
+      "loss": 0.3112,
+      "step": 8403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827937311586622e-06,
+      "loss": 0.2593,
+      "step": 8404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825753200358434e-06,
+      "loss": 0.3228,
+      "step": 8405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823569565010682e-06,
+      "loss": 0.2648,
+      "step": 8406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821386405711669e-06,
+      "loss": 0.2551,
+      "step": 8407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819203722629663e-06,
+      "loss": 0.3141,
+      "step": 8408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817021515932897e-06,
+      "loss": 0.2149,
+      "step": 8409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814839785789555e-06,
+      "loss": 0.2427,
+      "step": 8410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8126585323678066e-06,
+      "loss": 0.263,
+      "step": 8411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8104777558357632e-06,
+      "loss": 0.2725,
+      "step": 8412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8082974563615104e-06,
+      "loss": 0.2427,
+      "step": 8413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8061176341130955e-06,
+      "loss": 0.2381,
+      "step": 8414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8039382892585297e-06,
+      "loss": 0.2234,
+      "step": 8415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8017594219657872e-06,
+      "loss": 0.2787,
+      "step": 8416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799581032402796e-06,
+      "loss": 0.3348,
+      "step": 8417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7974031207374685e-06,
+      "loss": 0.2542,
+      "step": 8418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795225687137657e-06,
+      "loss": 0.2748,
+      "step": 8419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7930487317711907e-06,
+      "loss": 0.2807,
+      "step": 8420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7908722548058586e-06,
+      "loss": 0.2467,
+      "step": 8421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.788696256409412e-06,
+      "loss": 0.2634,
+      "step": 8422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7865207367495716e-06,
+      "loss": 0.2833,
+      "step": 8423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.784345695994004e-06,
+      "loss": 0.3292,
+      "step": 8424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7821711343103652e-06,
+      "loss": 0.3304,
+      "step": 8425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7799970518662477e-06,
+      "loss": 0.3697,
+      "step": 8426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.777823448829224e-06,
+      "loss": 0.338,
+      "step": 8427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7756503253668244e-06,
+      "loss": 0.2753,
+      "step": 8428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7734776816465413e-06,
+      "loss": 0.2821,
+      "step": 8429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.771305517835837e-06,
+      "loss": 0.2437,
+      "step": 8430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.769133834102119e-06,
+      "loss": 0.2803,
+      "step": 8431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.766962630612785e-06,
+      "loss": 0.3024,
+      "step": 8432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.764791907535168e-06,
+      "loss": 0.3219,
+      "step": 8433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7626216650365833e-06,
+      "loss": 0.2435,
+      "step": 8434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7604519032843e-06,
+      "loss": 0.2917,
+      "step": 8435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7582826224455537e-06,
+      "loss": 0.2458,
+      "step": 8436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.756113822687546e-06,
+      "loss": 0.2998,
+      "step": 8437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7539455041774255e-06,
+      "loss": 0.2598,
+      "step": 8438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.751777667082326e-06,
+      "loss": 0.2447,
+      "step": 8439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749610311569334e-06,
+      "loss": 0.2561,
+      "step": 8440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7474434378054913e-06,
+      "loss": 0.2655,
+      "step": 8441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7452770459578134e-06,
+      "loss": 0.3026,
+      "step": 8442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743111136193277e-06,
+      "loss": 0.2927,
+      "step": 8443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.740945708678817e-06,
+      "loss": 0.2655,
+      "step": 8444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7387807635813343e-06,
+      "loss": 0.275,
+      "step": 8445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7366163010676937e-06,
+      "loss": 0.3058,
+      "step": 8446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7344523213047236e-06,
+      "loss": 0.3077,
+      "step": 8447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7322888244592028e-06,
+      "loss": 0.2726,
+      "step": 8448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7301258106978953e-06,
+      "loss": 0.2393,
+      "step": 8449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7279632801875076e-06,
+      "loss": 0.285,
+      "step": 8450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.725801233094719e-06,
+      "loss": 0.2836,
+      "step": 8451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7236396695861697e-06,
+      "loss": 0.2501,
+      "step": 8452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7214785898284635e-06,
+      "loss": 0.2289,
+      "step": 8453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7193179939881665e-06,
+      "loss": 0.5929,
+      "step": 8454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.717157882231798e-06,
+      "loss": 0.2479,
+      "step": 8455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.714998254725862e-06,
+      "loss": 0.3182,
+      "step": 8456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7128391116368035e-06,
+      "loss": 0.2921,
+      "step": 8457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.710680453131039e-06,
+      "loss": 0.2568,
+      "step": 8458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7085222793749486e-06,
+      "loss": 0.2676,
+      "step": 8459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7063645905348753e-06,
+      "loss": 0.2787,
+      "step": 8460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7042073867771243e-06,
+      "loss": 0.3097,
+      "step": 8461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7020506682679524e-06,
+      "loss": 0.2638,
+      "step": 8462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6998944351736034e-06,
+      "loss": 0.3032,
+      "step": 8463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.697738687660257e-06,
+      "loss": 0.3312,
+      "step": 8464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6955834258940726e-06,
+      "loss": 0.2757,
+      "step": 8465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6934286500411675e-06,
+      "loss": 0.259,
+      "step": 8466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6912743602676195e-06,
+      "loss": 0.2856,
+      "step": 8467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.689120556739475e-06,
+      "loss": 0.259,
+      "step": 8468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6869672396227283e-06,
+      "loss": 0.2773,
+      "step": 8469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6848144090833602e-06,
+      "loss": 0.2533,
+      "step": 8470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.68266206528729e-06,
+      "loss": 0.2699,
+      "step": 8471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680510208400413e-06,
+      "loss": 0.2847,
+      "step": 8472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6783588385885826e-06,
+      "loss": 0.2516,
+      "step": 8473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.676207956017618e-06,
+      "loss": 0.2451,
+      "step": 8474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6740575608533e-06,
+      "loss": 0.2958,
+      "step": 8475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6719076532613606e-06,
+      "loss": 0.2924,
+      "step": 8476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6697582334075156e-06,
+      "loss": 0.329,
+      "step": 8477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.667609301457431e-06,
+      "loss": 0.3004,
+      "step": 8478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.665460857576728e-06,
+      "loss": 0.3032,
+      "step": 8479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.663312901931002e-06,
+      "loss": 0.3025,
+      "step": 8480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6611654346858074e-06,
+      "loss": 0.3052,
+      "step": 8481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.65901845600666e-06,
+      "loss": 0.2354,
+      "step": 8482
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.656871966059038e-06,
+      "loss": 0.2615,
+      "step": 8483
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.654725965008383e-06,
+      "loss": 0.2446,
+      "step": 8484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6525804530201015e-06,
+      "loss": 0.3176,
+      "step": 8485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650435430259548e-06,
+      "loss": 0.2448,
+      "step": 8486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6482908968920636e-06,
+      "loss": 0.2935,
+      "step": 8487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6461468530829293e-06,
+      "loss": 0.2845,
+      "step": 8488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6440032989974004e-06,
+      "loss": 0.2479,
+      "step": 8489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6418602348006903e-06,
+      "loss": 0.2798,
+      "step": 8490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6397176606579775e-06,
+      "loss": 0.2497,
+      "step": 8491
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6375755767344047e-06,
+      "loss": 0.3129,
+      "step": 8492
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6354339831950603e-06,
+      "loss": 0.2673,
+      "step": 8493
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.633292880205024e-06,
+      "loss": 0.2958,
+      "step": 8494
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6311522679293097e-06,
+      "loss": 0.2579,
+      "step": 8495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6290121465329096e-06,
+      "loss": 0.2701,
+      "step": 8496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.626872516180774e-06,
+      "loss": 0.2783,
+      "step": 8497
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6247333770378133e-06,
+      "loss": 0.2784,
+      "step": 8498
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.622594729268907e-06,
+      "loss": 0.2903,
+      "step": 8499
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6204565730388795e-06,
+      "loss": 0.2931,
+      "step": 8500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618318908512545e-06,
+      "loss": 0.2687,
+      "step": 8501
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6161817358546513e-06,
+      "loss": 0.2609,
+      "step": 8502
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6140450552299268e-06,
+      "loss": 0.2789,
+      "step": 8503
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6119088668030557e-06,
+      "loss": 0.3287,
+      "step": 8504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.609773170738685e-06,
+      "loss": 0.2912,
+      "step": 8505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6076379672014263e-06,
+      "loss": 0.2817,
+      "step": 8506
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6055032563558402e-06,
+      "loss": 0.2711,
+      "step": 8507
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6033690383664745e-06,
+      "loss": 0.2748,
+      "step": 8508
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.601235313397813e-06,
+      "loss": 0.2513,
+      "step": 8509
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5991020816143164e-06,
+      "loss": 0.2733,
+      "step": 8510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.596969343180403e-06,
+      "loss": 0.2764,
+      "step": 8511
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.594837098260454e-06,
+      "loss": 0.264,
+      "step": 8512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5927053470188176e-06,
+      "loss": 0.2702,
+      "step": 8513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.590574089619786e-06,
+      "loss": 0.2794,
+      "step": 8514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5884433262276376e-06,
+      "loss": 0.2744,
+      "step": 8515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5863130570065998e-06,
+      "loss": 0.2683,
+      "step": 8516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5841832821208577e-06,
+      "loss": 0.27,
+      "step": 8517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5820540017345663e-06,
+      "loss": 0.2721,
+      "step": 8518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5799252160118405e-06,
+      "loss": 0.2531,
+      "step": 8519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5777969251167568e-06,
+      "loss": 0.2673,
+      "step": 8520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.575669129213353e-06,
+      "loss": 0.2453,
+      "step": 8521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5735418284656287e-06,
+      "loss": 0.2398,
+      "step": 8522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.57141502303755e-06,
+      "loss": 0.254,
+      "step": 8523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.569288713093034e-06,
+      "loss": 0.2614,
+      "step": 8524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5671628987959685e-06,
+      "loss": 0.3077,
+      "step": 8525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.565037580310201e-06,
+      "loss": 0.2505,
+      "step": 8526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5629127577995405e-06,
+      "loss": 0.2555,
+      "step": 8527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5607884314277583e-06,
+      "loss": 0.2615,
+      "step": 8528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558664601358587e-06,
+      "loss": 0.2531,
+      "step": 8529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5565412677557233e-06,
+      "loss": 0.2532,
+      "step": 8530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5544184307828142e-06,
+      "loss": 0.2345,
+      "step": 8531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.552296090603491e-06,
+      "loss": 0.2524,
+      "step": 8532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5501742473813226e-06,
+      "loss": 0.2336,
+      "step": 8533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.548052901279854e-06,
+      "loss": 0.246,
+      "step": 8534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.545932052462587e-06,
+      "loss": 0.2652,
+      "step": 8535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5438117010929875e-06,
+      "loss": 0.2462,
+      "step": 8536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.541691847334484e-06,
+      "loss": 0.2732,
+      "step": 8537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5395724913504546e-06,
+      "loss": 0.3038,
+      "step": 8538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.537453633304263e-06,
+      "loss": 0.3066,
+      "step": 8539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5353352733592095e-06,
+      "loss": 0.2835,
+      "step": 8540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.53321741167857e-06,
+      "loss": 0.2787,
+      "step": 8541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5311000484255796e-06,
+      "loss": 0.5907,
+      "step": 8542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5289831837634334e-06,
+      "loss": 0.2612,
+      "step": 8543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5268668178552922e-06,
+      "loss": 0.3006,
+      "step": 8544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5247509508642653e-06,
+      "loss": 0.2737,
+      "step": 8545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5226355829534475e-06,
+      "loss": 0.2662,
+      "step": 8546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.520520714285869e-06,
+      "loss": 0.2731,
+      "step": 8547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5184063450245386e-06,
+      "loss": 0.2498,
+      "step": 8548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5162924753324202e-06,
+      "loss": 0.2605,
+      "step": 8549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5141791053724405e-06,
+      "loss": 0.277,
+      "step": 8550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.512066235307492e-06,
+      "loss": 0.3588,
+      "step": 8551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.509953865300414e-06,
+      "loss": 0.2222,
+      "step": 8552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5078419955140263e-06,
+      "loss": 0.2911,
+      "step": 8553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5057306261111024e-06,
+      "loss": 0.2678,
+      "step": 8554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5036197572543697e-06,
+      "loss": 0.3234,
+      "step": 8555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5015093891065253e-06,
+      "loss": 0.2738,
+      "step": 8556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.499399521830229e-06,
+      "loss": 0.2598,
+      "step": 8557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4972901555880957e-06,
+      "loss": 0.2654,
+      "step": 8558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4951812905427073e-06,
+      "loss": 0.2504,
+      "step": 8559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4930729268566035e-06,
+      "loss": 0.3011,
+      "step": 8560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49096506469229e-06,
+      "loss": 0.2445,
+      "step": 8561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.488857704212224e-06,
+      "loss": 0.2291,
+      "step": 8562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4867508455788336e-06,
+      "loss": 0.2383,
+      "step": 8563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484644488954505e-06,
+      "loss": 0.2345,
+      "step": 8564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4825386345015865e-06,
+      "loss": 0.2863,
+      "step": 8565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4804332823823862e-06,
+      "loss": 0.2668,
+      "step": 8566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4783284327591736e-06,
+      "loss": 0.5621,
+      "step": 8567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4762240857941843e-06,
+      "loss": 0.5822,
+      "step": 8568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.474120241649601e-06,
+      "loss": 0.3455,
+      "step": 8569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4720169004875914e-06,
+      "loss": 0.2673,
+      "step": 8570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.46991406247026e-06,
+      "loss": 0.2792,
+      "step": 8571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4678117277596856e-06,
+      "loss": 0.2955,
+      "step": 8572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.465709896517908e-06,
+      "loss": 0.274,
+      "step": 8573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4636085689069244e-06,
+      "loss": 0.3306,
+      "step": 8574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.461507745088698e-06,
+      "loss": 0.2629,
+      "step": 8575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4594074252251397e-06,
+      "loss": 0.27,
+      "step": 8576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4573076094781466e-06,
+      "loss": 0.2626,
+      "step": 8577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4552082980095514e-06,
+      "loss": 0.3026,
+      "step": 8578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4531094909811614e-06,
+      "loss": 0.2465,
+      "step": 8579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4510111885547426e-06,
+      "loss": 0.2574,
+      "step": 8580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4489133908920225e-06,
+      "loss": 0.2577,
+      "step": 8581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446816098154692e-06,
+      "loss": 0.2302,
+      "step": 8582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4447193105043884e-06,
+      "loss": 0.2762,
+      "step": 8583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4426230281027374e-06,
+      "loss": 0.2634,
+      "step": 8584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4405272511112986e-06,
+      "loss": 0.267,
+      "step": 8585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4384319796916075e-06,
+      "loss": 0.2787,
+      "step": 8586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4363372140051586e-06,
+      "loss": 0.247,
+      "step": 8587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4342429542134047e-06,
+      "loss": 0.2725,
+      "step": 8588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432149200477766e-06,
+      "loss": 0.2563,
+      "step": 8589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.430055952959607e-06,
+      "loss": 0.2482,
+      "step": 8590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4279632118202744e-06,
+      "loss": 0.3,
+      "step": 8591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4258709772210686e-06,
+      "loss": 0.2694,
+      "step": 8592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4237792493232402e-06,
+      "loss": 0.2949,
+      "step": 8593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4216880282880128e-06,
+      "loss": 0.2317,
+      "step": 8594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4195973142765694e-06,
+      "loss": 0.3146,
+      "step": 8595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.417507107450049e-06,
+      "loss": 0.2536,
+      "step": 8596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4154174079695555e-06,
+      "loss": 0.297,
+      "step": 8597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4133282159961535e-06,
+      "loss": 0.2405,
+      "step": 8598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4112395316908697e-06,
+      "loss": 0.2518,
+      "step": 8599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4091513552146836e-06,
+      "loss": 0.2872,
+      "step": 8600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4070636867285455e-06,
+      "loss": 0.3068,
+      "step": 8601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.404976526393361e-06,
+      "loss": 0.3159,
+      "step": 8602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4028898743699988e-06,
+      "loss": 0.2369,
+      "step": 8603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4008037308192874e-06,
+      "loss": 0.2592,
+      "step": 8604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3987180959020184e-06,
+      "loss": 0.3264,
+      "step": 8605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3966329697789424e-06,
+      "loss": 0.2519,
+      "step": 8606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.394548352610767e-06,
+      "loss": 0.2443,
+      "step": 8607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3924642445581647e-06,
+      "loss": 0.2999,
+      "step": 8608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.39038064578177e-06,
+      "loss": 0.2908,
+      "step": 8609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3882975564421773e-06,
+      "loss": 0.2714,
+      "step": 8610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3862149766999385e-06,
+      "loss": 0.6063,
+      "step": 8611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3841329067155693e-06,
+      "loss": 0.2453,
+      "step": 8612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.38205134664955e-06,
+      "loss": 0.2421,
+      "step": 8613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.379970296662305e-06,
+      "loss": 0.2287,
+      "step": 8614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3778897569142454e-06,
+      "loss": 0.2653,
+      "step": 8615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3758097275657208e-06,
+      "loss": 0.2332,
+      "step": 8616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3737302087770497e-06,
+      "loss": 0.2803,
+      "step": 8617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3716512007085133e-06,
+      "loss": 0.2875,
+      "step": 8618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.369572703520352e-06,
+      "loss": 0.2805,
+      "step": 8619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3674947173727667e-06,
+      "loss": 0.2669,
+      "step": 8620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3654172424259103e-06,
+      "loss": 0.3019,
+      "step": 8621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.363340278839916e-06,
+      "loss": 0.2605,
+      "step": 8622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3612638267748575e-06,
+      "loss": 0.2546,
+      "step": 8623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.359187886390781e-06,
+      "loss": 0.2698,
+      "step": 8624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3571124578476888e-06,
+      "loss": 0.2451,
+      "step": 8625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355037541305545e-06,
+      "loss": 0.247,
+      "step": 8626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3529631369242764e-06,
+      "loss": 0.2975,
+      "step": 8627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.350889244863759e-06,
+      "loss": 0.2687,
+      "step": 8628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.348815865283848e-06,
+      "loss": 0.2657,
+      "step": 8629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3467429983443477e-06,
+      "loss": 0.2597,
+      "step": 8630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3446706442050203e-06,
+      "loss": 0.2668,
+      "step": 8631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.342598803025595e-06,
+      "loss": 0.2994,
+      "step": 8632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3405274749657577e-06,
+      "loss": 0.2716,
+      "step": 8633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3384566601851574e-06,
+      "loss": 0.2849,
+      "step": 8634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.336386358843403e-06,
+      "loss": 0.3148,
+      "step": 8635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3343165711000613e-06,
+      "loss": 0.2504,
+      "step": 8636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.332247297114666e-06,
+      "loss": 0.2939,
+      "step": 8637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.330178537046699e-06,
+      "loss": 0.2566,
+      "step": 8638
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.328110291055614e-06,
+      "loss": 0.2408,
+      "step": 8639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3260425593008207e-06,
+      "loss": 0.2856,
+      "step": 8640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.323975341941691e-06,
+      "loss": 0.2764,
+      "step": 8641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.321908639137553e-06,
+      "loss": 0.29,
+      "step": 8642
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3198424510477014e-06,
+      "loss": 0.2631,
+      "step": 8643
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3177767778313884e-06,
+      "loss": 0.2785,
+      "step": 8644
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.31571161964782e-06,
+      "loss": 0.307,
+      "step": 8645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.313646976656172e-06,
+      "loss": 0.2286,
+      "step": 8646
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3115828490155775e-06,
+      "loss": 0.2911,
+      "step": 8647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.309519236885128e-06,
+      "loss": 0.3014,
+      "step": 8648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.307456140423877e-06,
+      "loss": 0.2965,
+      "step": 8649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.305393559790838e-06,
+      "loss": 0.2422,
+      "step": 8650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.303331495144988e-06,
+      "loss": 0.2131,
+      "step": 8651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3012699466452503e-06,
+      "loss": 0.3192,
+      "step": 8652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.299208914450532e-06,
+      "loss": 0.2818,
+      "step": 8653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2971483987196783e-06,
+      "loss": 0.289,
+      "step": 8654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.295088399611507e-06,
+      "loss": 0.33,
+      "step": 8655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2930289172847905e-06,
+      "loss": 0.2634,
+      "step": 8656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290969951898265e-06,
+      "loss": 0.2961,
+      "step": 8657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288911503610629e-06,
+      "loss": 0.2738,
+      "step": 8658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.286853572580527e-06,
+      "loss": 0.3051,
+      "step": 8659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.284796158966589e-06,
+      "loss": 0.2639,
+      "step": 8660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.282739262927377e-06,
+      "loss": 0.2726,
+      "step": 8661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2806828846214324e-06,
+      "loss": 0.2635,
+      "step": 8662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2786270242072504e-06,
+      "loss": 0.2471,
+      "step": 8663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.276571681843286e-06,
+      "loss": 0.5651,
+      "step": 8664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.274516857687957e-06,
+      "loss": 0.2816,
+      "step": 8665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2724625518996322e-06,
+      "loss": 0.3109,
+      "step": 8666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2704087646366546e-06,
+      "loss": 0.2872,
+      "step": 8667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2683554960573207e-06,
+      "loss": 0.2745,
+      "step": 8668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2663027463198794e-06,
+      "loss": 0.3009,
+      "step": 8669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.264250515582551e-06,
+      "loss": 0.2886,
+      "step": 8670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2621988040035103e-06,
+      "loss": 0.2413,
+      "step": 8671
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2601476117408937e-06,
+      "loss": 0.2578,
+      "step": 8672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.258096938952796e-06,
+      "loss": 0.2976,
+      "step": 8673
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2560467857972744e-06,
+      "loss": 0.5608,
+      "step": 8674
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2539971524323455e-06,
+      "loss": 0.2701,
+      "step": 8675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2519480390159806e-06,
+      "loss": 0.2593,
+      "step": 8676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2498994457061184e-06,
+      "loss": 0.2593,
+      "step": 8677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.247851372660653e-06,
+      "loss": 0.2738,
+      "step": 8678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2458038200374408e-06,
+      "loss": 0.2813,
+      "step": 8679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2437567879942966e-06,
+      "loss": 0.3186,
+      "step": 8680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241710276688995e-06,
+      "loss": 0.2585,
+      "step": 8681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.239664286279276e-06,
+      "loss": 0.2809,
+      "step": 8682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.237618816922826e-06,
+      "loss": 0.2548,
+      "step": 8683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.235573868777304e-06,
+      "loss": 0.2615,
+      "step": 8684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.233529442000324e-06,
+      "loss": 0.5789,
+      "step": 8685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.23148553674946e-06,
+      "loss": 0.2833,
+      "step": 8686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2294421531822475e-06,
+      "loss": 0.2785,
+      "step": 8687
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.227399291456179e-06,
+      "loss": 0.2759,
+      "step": 8688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225356951728712e-06,
+      "loss": 0.2632,
+      "step": 8689
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223315134157253e-06,
+      "loss": 0.2714,
+      "step": 8690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2212738388991803e-06,
+      "loss": 0.2669,
+      "step": 8691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2192330661118243e-06,
+      "loss": 0.2535,
+      "step": 8692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217192815952479e-06,
+      "loss": 0.2918,
+      "step": 8693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2151530885783967e-06,
+      "loss": 0.282,
+      "step": 8694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.213113884146789e-06,
+      "loss": 0.2822,
+      "step": 8695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.211075202814832e-06,
+      "loss": 0.2908,
+      "step": 8696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2090370447396468e-06,
+      "loss": 0.2706,
+      "step": 8697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2069994100783376e-06,
+      "loss": 0.256,
+      "step": 8698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2049622989879446e-06,
+      "loss": 0.3157,
+      "step": 8699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.202925711625483e-06,
+      "loss": 0.2403,
+      "step": 8700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2008896481479223e-06,
+      "loss": 0.2863,
+      "step": 8701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1988541087121916e-06,
+      "loss": 0.2985,
+      "step": 8702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.196819093475184e-06,
+      "loss": 0.3075,
+      "step": 8703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.194784602593739e-06,
+      "loss": 0.2827,
+      "step": 8704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1927506362246753e-06,
+      "loss": 0.2224,
+      "step": 8705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1907171945247595e-06,
+      "loss": 0.269,
+      "step": 8706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1886842776507133e-06,
+      "loss": 0.3062,
+      "step": 8707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1866518857592267e-06,
+      "loss": 0.2865,
+      "step": 8708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1846200190069476e-06,
+      "loss": 0.3038,
+      "step": 8709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.182588677550482e-06,
+      "loss": 0.2447,
+      "step": 8710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.180557861546395e-06,
+      "loss": 0.2328,
+      "step": 8711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1785275711512122e-06,
+      "loss": 0.2377,
+      "step": 8712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1764978065214215e-06,
+      "loss": 0.2819,
+      "step": 8713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.174468567813461e-06,
+      "loss": 0.31,
+      "step": 8714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1724398551837364e-06,
+      "loss": 0.2526,
+      "step": 8715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1704116687886132e-06,
+      "loss": 0.2666,
+      "step": 8716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.168384008784412e-06,
+      "loss": 0.3362,
+      "step": 8717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1663568753274153e-06,
+      "loss": 0.2383,
+      "step": 8718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.164330268573864e-06,
+      "loss": 0.2584,
+      "step": 8719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1623041886799643e-06,
+      "loss": 0.2503,
+      "step": 8720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1602786358018668e-06,
+      "loss": 0.2581,
+      "step": 8721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1582536100956973e-06,
+      "loss": 0.274,
+      "step": 8722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1562291117175324e-06,
+      "loss": 0.2448,
+      "step": 8723
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.154205140823412e-06,
+      "loss": 0.3199,
+      "step": 8724
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.152181697569334e-06,
+      "loss": 0.2867,
+      "step": 8725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1501587821112532e-06,
+      "loss": 0.2816,
+      "step": 8726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1481363946050925e-06,
+      "loss": 0.2725,
+      "step": 8727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.146114535206718e-06,
+      "loss": 0.2487,
+      "step": 8728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1440932040719694e-06,
+      "loss": 0.2401,
+      "step": 8729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1420724013566408e-06,
+      "loss": 0.2852,
+      "step": 8730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1400521272164854e-06,
+      "loss": 0.3552,
+      "step": 8731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1380323818072155e-06,
+      "loss": 0.2738,
+      "step": 8732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.136013165284504e-06,
+      "loss": 0.2919,
+      "step": 8733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1339944778039844e-06,
+      "loss": 0.5553,
+      "step": 8734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1319763195212382e-06,
+      "loss": 0.2923,
+      "step": 8735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.129958690591829e-06,
+      "loss": 0.3061,
+      "step": 8736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.127941591171254e-06,
+      "loss": 0.283,
+      "step": 8737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.125925021414985e-06,
+      "loss": 0.2739,
+      "step": 8738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1239089814784505e-06,
+      "loss": 0.2598,
+      "step": 8739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1218934715170355e-06,
+      "loss": 0.2942,
+      "step": 8740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.119878491686089e-06,
+      "loss": 0.3181,
+      "step": 8741
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1178640421409057e-06,
+      "loss": 0.2711,
+      "step": 8742
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.115850123036761e-06,
+      "loss": 0.2656,
+      "step": 8743
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1138367345288757e-06,
+      "loss": 0.2692,
+      "step": 8744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.111823876772426e-06,
+      "loss": 0.389,
+      "step": 8745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1098115499225567e-06,
+      "loss": 0.2974,
+      "step": 8746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1077997541343672e-06,
+      "loss": 0.2835,
+      "step": 8747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1057884895629174e-06,
+      "loss": 0.2287,
+      "step": 8748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1037777563632264e-06,
+      "loss": 0.2888,
+      "step": 8749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1017675546902704e-06,
+      "loss": 0.2555,
+      "step": 8750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0997578846989886e-06,
+      "loss": 0.2608,
+      "step": 8751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.097748746544271e-06,
+      "loss": 0.3033,
+      "step": 8752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.095740140380975e-06,
+      "loss": 0.3217,
+      "step": 8753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0937320663639148e-06,
+      "loss": 0.3027,
+      "step": 8754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091724524647861e-06,
+      "loss": 0.2849,
+      "step": 8755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0897175153875467e-06,
+      "loss": 0.2658,
+      "step": 8756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087711038737662e-06,
+      "loss": 0.2956,
+      "step": 8757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0857050948528576e-06,
+      "loss": 0.2896,
+      "step": 8758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083699683887739e-06,
+      "loss": 0.3032,
+      "step": 8759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.081694805996872e-06,
+      "loss": 0.2695,
+      "step": 8760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0796904613347855e-06,
+      "loss": 0.2991,
+      "step": 8761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0776866500559654e-06,
+      "loss": 0.3035,
+      "step": 8762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0756833723148526e-06,
+      "loss": 0.3132,
+      "step": 8763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0736806282658514e-06,
+      "loss": 0.2821,
+      "step": 8764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0716784180633276e-06,
+      "loss": 0.2738,
+      "step": 8765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0696767418615945e-06,
+      "loss": 0.2892,
+      "step": 8766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0676755998149333e-06,
+      "loss": 0.2974,
+      "step": 8767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.065674992077584e-06,
+      "loss": 0.3019,
+      "step": 8768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.063674918803743e-06,
+      "loss": 0.2569,
+      "step": 8769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0616753801475653e-06,
+      "loss": 0.2845,
+      "step": 8770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0596763762631655e-06,
+      "loss": 0.2559,
+      "step": 8771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0576779073046214e-06,
+      "loss": 0.2432,
+      "step": 8772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055679973425958e-06,
+      "loss": 0.2878,
+      "step": 8773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0536825747811695e-06,
+      "loss": 0.289,
+      "step": 8774
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.051685711524205e-06,
+      "loss": 0.2903,
+      "step": 8775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0496893838089736e-06,
+      "loss": 0.3218,
+      "step": 8776
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0476935917893413e-06,
+      "loss": 0.2651,
+      "step": 8777
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.045698335619135e-06,
+      "loss": 0.2812,
+      "step": 8778
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0437036154521426e-06,
+      "loss": 0.2305,
+      "step": 8779
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0417094314420958e-06,
+      "loss": 0.3206,
+      "step": 8780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039715783742708e-06,
+      "loss": 0.2581,
+      "step": 8781
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0377226725076394e-06,
+      "loss": 0.2673,
+      "step": 8782
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0357300978905025e-06,
+      "loss": 0.2631,
+      "step": 8783
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0337380600448774e-06,
+      "loss": 0.2949,
+      "step": 8784
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.031746559124301e-06,
+      "loss": 0.2809,
+      "step": 8785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.02975559528227e-06,
+      "loss": 0.248,
+      "step": 8786
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0277651686722353e-06,
+      "loss": 0.2822,
+      "step": 8787
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0257752794476113e-06,
+      "loss": 0.2659,
+      "step": 8788
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.023785927761772e-06,
+      "loss": 0.2891,
+      "step": 8789
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.021797113768039e-06,
+      "loss": 0.2844,
+      "step": 8790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.019808837619704e-06,
+      "loss": 0.2602,
+      "step": 8791
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017821099470014e-06,
+      "loss": 0.2915,
+      "step": 8792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0158338994721737e-06,
+      "loss": 0.2745,
+      "step": 8793
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.013847237779346e-06,
+      "loss": 0.2757,
+      "step": 8794
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011861114544654e-06,
+      "loss": 0.3434,
+      "step": 8795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.009875529921181e-06,
+      "loss": 0.2637,
+      "step": 8796
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0078904840619607e-06,
+      "loss": 0.2512,
+      "step": 8797
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.005905977119992e-06,
+      "loss": 0.264,
+      "step": 8798
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0039220092482313e-06,
+      "loss": 0.2604,
+      "step": 8799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0019385805995936e-06,
+      "loss": 0.2977,
+      "step": 8800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.999955691326952e-06,
+      "loss": 0.2452,
+      "step": 8801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997973341583138e-06,
+      "loss": 0.2404,
+      "step": 8802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9959915315209444e-06,
+      "loss": 0.2798,
+      "step": 8803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.994010261293111e-06,
+      "loss": 0.2356,
+      "step": 8804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9920295310523496e-06,
+      "loss": 0.2933,
+      "step": 8805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9900493409513256e-06,
+      "loss": 0.291,
+      "step": 8806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.98806969114266e-06,
+      "loss": 0.2778,
+      "step": 8807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9860905817789354e-06,
+      "loss": 0.2945,
+      "step": 8808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.984112013012692e-06,
+      "loss": 0.2766,
+      "step": 8809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9821339849964324e-06,
+      "loss": 0.3086,
+      "step": 8810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.980156497882605e-06,
+      "loss": 0.2699,
+      "step": 8811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9781795518236288e-06,
+      "loss": 0.2803,
+      "step": 8812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9762031469718777e-06,
+      "loss": 0.289,
+      "step": 8813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9742272834796813e-06,
+      "loss": 0.2631,
+      "step": 8814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9722519614993306e-06,
+      "loss": 0.2943,
+      "step": 8815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.970277181183074e-06,
+      "loss": 0.2924,
+      "step": 8816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.968302942683121e-06,
+      "loss": 0.2261,
+      "step": 8817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.966329246151626e-06,
+      "loss": 0.2853,
+      "step": 8818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.964356091740721e-06,
+      "loss": 0.2446,
+      "step": 8819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9623834796024874e-06,
+      "loss": 0.6219,
+      "step": 8820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9604114098889592e-06,
+      "loss": 0.3151,
+      "step": 8821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9584398827521343e-06,
+      "loss": 0.264,
+      "step": 8822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9564688983439716e-06,
+      "loss": 0.2772,
+      "step": 8823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9544984568163815e-06,
+      "loss": 0.2816,
+      "step": 8824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9525285583212382e-06,
+      "loss": 0.2513,
+      "step": 8825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.950559203010371e-06,
+      "loss": 0.2559,
+      "step": 8826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9485903910355716e-06,
+      "loss": 0.257,
+      "step": 8827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.946622122548579e-06,
+      "loss": 0.2834,
+      "step": 8828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9446543977011023e-06,
+      "loss": 0.2933,
+      "step": 8829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942687216644803e-06,
+      "loss": 0.2908,
+      "step": 8830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.940720579531301e-06,
+      "loss": 0.2822,
+      "step": 8831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9387544865121763e-06,
+      "loss": 0.2466,
+      "step": 8832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.936788937738966e-06,
+      "loss": 0.2953,
+      "step": 8833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9348239333631655e-06,
+      "loss": 0.303,
+      "step": 8834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9328594735362237e-06,
+      "loss": 0.2895,
+      "step": 8835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9308955584095544e-06,
+      "loss": 0.2406,
+      "step": 8836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.262,
+      "step": 8837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.926969362862465e-06,
+      "loss": 0.2534,
+      "step": 8838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9250070827446563e-06,
+      "loss": 0.3336,
+      "step": 8839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.923045347932344e-06,
+      "loss": 0.2393,
+      "step": 8840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.92108415857673e-06,
+      "loss": 0.309,
+      "step": 8841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.919123514828969e-06,
+      "loss": 0.2438,
+      "step": 8842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9171634168401797e-06,
+      "loss": 0.2555,
+      "step": 8843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9152038647614357e-06,
+      "loss": 0.3085,
+      "step": 8844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9132448587437722e-06,
+      "loss": 0.244,
+      "step": 8845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.911286398938178e-06,
+      "loss": 0.2952,
+      "step": 8846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9093284854956017e-06,
+      "loss": 0.2805,
+      "step": 8847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.907371118566953e-06,
+      "loss": 0.2447,
+      "step": 8848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9054142983030884e-06,
+      "loss": 0.2794,
+      "step": 8849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9034580248548363e-06,
+      "loss": 0.2491,
+      "step": 8850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9015022983729733e-06,
+      "loss": 0.2717,
+      "step": 8851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.899547119008239e-06,
+      "loss": 0.2768,
+      "step": 8852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.897592486911328e-06,
+      "loss": 0.3257,
+      "step": 8853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8956384022328943e-06,
+      "loss": 0.3244,
+      "step": 8854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8936848651235516e-06,
+      "loss": 0.3015,
+      "step": 8855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.891731875733863e-06,
+      "loss": 0.2301,
+      "step": 8856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.889779434214356e-06,
+      "loss": 0.275,
+      "step": 8857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8878275407155244e-06,
+      "loss": 0.2593,
+      "step": 8858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.885876195387799e-06,
+      "loss": 0.2585,
+      "step": 8859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.883925398381585e-06,
+      "loss": 0.269,
+      "step": 8860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8819751498472405e-06,
+      "loss": 0.2657,
+      "step": 8861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8800254499350797e-06,
+      "loss": 0.2767,
+      "step": 8862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.878076298795376e-06,
+      "loss": 0.332,
+      "step": 8863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8761276965783613e-06,
+      "loss": 0.2704,
+      "step": 8864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874179643434227e-06,
+      "loss": 0.2747,
+      "step": 8865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8722321395131127e-06,
+      "loss": 0.2766,
+      "step": 8866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8702851849651258e-06,
+      "loss": 0.2656,
+      "step": 8867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.868338779940327e-06,
+      "loss": 0.2666,
+      "step": 8868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.866392924588738e-06,
+      "loss": 0.3022,
+      "step": 8869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.864447619060333e-06,
+      "loss": 0.2492,
+      "step": 8870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8625028635050477e-06,
+      "loss": 0.2811,
+      "step": 8871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8605586580727783e-06,
+      "loss": 0.2629,
+      "step": 8872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8586150029133663e-06,
+      "loss": 0.2952,
+      "step": 8873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8566718981766238e-06,
+      "loss": 0.2839,
+      "step": 8874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.854729344012316e-06,
+      "loss": 0.3015,
+      "step": 8875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8527873405701636e-06,
+      "loss": 0.2704,
+      "step": 8876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850845887999848e-06,
+      "loss": 0.2374,
+      "step": 8877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8489049864510053e-06,
+      "loss": 0.277,
+      "step": 8878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8469646360732362e-06,
+      "loss": 0.2593,
+      "step": 8879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.845024837016085e-06,
+      "loss": 0.3086,
+      "step": 8880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8430855894290655e-06,
+      "loss": 0.2775,
+      "step": 8881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.841146893461646e-06,
+      "loss": 0.2693,
+      "step": 8882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8392087492632504e-06,
+      "loss": 0.2789,
+      "step": 8883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.837271156983261e-06,
+      "loss": 0.2928,
+      "step": 8884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.83533411677102e-06,
+      "loss": 0.2761,
+      "step": 8885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.833397628775827e-06,
+      "loss": 0.264,
+      "step": 8886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8314616931469295e-06,
+      "loss": 0.3126,
+      "step": 8887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8295263100335447e-06,
+      "loss": 0.2698,
+      "step": 8888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8275914795848415e-06,
+      "loss": 0.2929,
+      "step": 8889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8256572019499474e-06,
+      "loss": 0.2853,
+      "step": 8890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8237234772779455e-06,
+      "loss": 0.3002,
+      "step": 8891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8217903057178796e-06,
+      "loss": 0.2512,
+      "step": 8892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8198576874187513e-06,
+      "loss": 0.2697,
+      "step": 8893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8179256225295114e-06,
+      "loss": 0.2708,
+      "step": 8894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.815994111199074e-06,
+      "loss": 0.2601,
+      "step": 8895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8140631535763195e-06,
+      "loss": 0.3077,
+      "step": 8896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.812132749810067e-06,
+      "loss": 0.2704,
+      "step": 8897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.810202900049106e-06,
+      "loss": 0.2992,
+      "step": 8898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.80827360444218e-06,
+      "loss": 0.2306,
+      "step": 8899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.806344863137989e-06,
+      "loss": 0.2949,
+      "step": 8900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8044166762851898e-06,
+      "loss": 0.2784,
+      "step": 8901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8024890440324e-06,
+      "loss": 0.2444,
+      "step": 8902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8005619665281935e-06,
+      "loss": 0.2661,
+      "step": 8903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.798635443921094e-06,
+      "loss": 0.5709,
+      "step": 8904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7967094763595917e-06,
+      "loss": 0.3088,
+      "step": 8905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7947840639921308e-06,
+      "loss": 0.2856,
+      "step": 8906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7928592069671113e-06,
+      "loss": 0.2781,
+      "step": 8907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7909349054328937e-06,
+      "loss": 0.2794,
+      "step": 8908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.789011159537792e-06,
+      "loss": 0.2968,
+      "step": 8909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7870879694300825e-06,
+      "loss": 0.2686,
+      "step": 8910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7851653352579886e-06,
+      "loss": 0.2484,
+      "step": 8911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783243257169702e-06,
+      "loss": 0.2371,
+      "step": 8912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.781321735313366e-06,
+      "loss": 0.2642,
+      "step": 8913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.779400769837083e-06,
+      "loss": 0.2697,
+      "step": 8914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.777480360888909e-06,
+      "loss": 0.2253,
+      "step": 8915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7755605086168624e-06,
+      "loss": 0.2537,
+      "step": 8916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7736412131689173e-06,
+      "loss": 0.2699,
+      "step": 8917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7717224746929984e-06,
+      "loss": 0.2928,
+      "step": 8918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.769804293336994e-06,
+      "loss": 0.2757,
+      "step": 8919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7678866692487503e-06,
+      "loss": 0.3195,
+      "step": 8920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7659696025760674e-06,
+      "loss": 0.2523,
+      "step": 8921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.764053093466702e-06,
+      "loss": 0.2746,
+      "step": 8922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7621371420683717e-06,
+      "loss": 0.3416,
+      "step": 8923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7602217485287497e-06,
+      "loss": 0.2775,
+      "step": 8924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7583069129954585e-06,
+      "loss": 0.2582,
+      "step": 8925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.75639263561609e-06,
+      "loss": 0.3271,
+      "step": 8926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7544789165381834e-06,
+      "loss": 0.2703,
+      "step": 8927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.752565755909242e-06,
+      "loss": 0.2889,
+      "step": 8928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.750653153876721e-06,
+      "loss": 0.2397,
+      "step": 8929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7487411105880356e-06,
+      "loss": 0.3184,
+      "step": 8930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.746829626190558e-06,
+      "loss": 0.2506,
+      "step": 8931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7449187008316113e-06,
+      "loss": 0.2531,
+      "step": 8932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7430083346584802e-06,
+      "loss": 0.267,
+      "step": 8933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7410985278184144e-06,
+      "loss": 0.2317,
+      "step": 8934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.739189280458604e-06,
+      "loss": 0.2836,
+      "step": 8935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7372805927262057e-06,
+      "loss": 0.2641,
+      "step": 8936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7353724647683344e-06,
+      "loss": 0.301,
+      "step": 8937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7334648967320587e-06,
+      "loss": 0.2787,
+      "step": 8938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7315578887644057e-06,
+      "loss": 0.2788,
+      "step": 8939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.72965144101235e-06,
+      "loss": 0.2899,
+      "step": 8940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7277455536228438e-06,
+      "loss": 0.2459,
+      "step": 8941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.725840226742774e-06,
+      "loss": 0.2461,
+      "step": 8942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.723935460518997e-06,
+      "loss": 0.3032,
+      "step": 8943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.722031255098323e-06,
+      "loss": 0.2745,
+      "step": 8944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7201276106275176e-06,
+      "loss": 0.2956,
+      "step": 8945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7182245272533046e-06,
+      "loss": 0.3193,
+      "step": 8946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.716322005122366e-06,
+      "loss": 0.2828,
+      "step": 8947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7144200443813394e-06,
+      "loss": 0.2609,
+      "step": 8948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.712518645176815e-06,
+      "loss": 0.2618,
+      "step": 8949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7106178076553446e-06,
+      "loss": 0.3062,
+      "step": 8950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.708717531963435e-06,
+      "loss": 0.2448,
+      "step": 8951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7068178182475514e-06,
+      "loss": 0.3056,
+      "step": 8952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7049186666541126e-06,
+      "loss": 0.2924,
+      "step": 8953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.703020077329498e-06,
+      "loss": 0.2819,
+      "step": 8954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7011220504200432e-06,
+      "loss": 0.3284,
+      "step": 8955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6992245860720325e-06,
+      "loss": 0.288,
+      "step": 8956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6973276844317166e-06,
+      "loss": 0.275,
+      "step": 8957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6954313456452995e-06,
+      "loss": 0.2579,
+      "step": 8958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6935355698589417e-06,
+      "loss": 0.2821,
+      "step": 8959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.691640357218759e-06,
+      "loss": 0.2684,
+      "step": 8960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6897457078708267e-06,
+      "loss": 0.3161,
+      "step": 8961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6878516219611773e-06,
+      "loss": 0.2882,
+      "step": 8962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6859580996357905e-06,
+      "loss": 0.2635,
+      "step": 8963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.684065141040614e-06,
+      "loss": 0.2587,
+      "step": 8964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6821727463215476e-06,
+      "loss": 0.2642,
+      "step": 8965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.680280915624448e-06,
+      "loss": 0.2742,
+      "step": 8966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6783896490951266e-06,
+      "loss": 0.2397,
+      "step": 8967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6764989468793544e-06,
+      "loss": 0.2552,
+      "step": 8968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6746088091228594e-06,
+      "loss": 0.2483,
+      "step": 8969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6727192359713196e-06,
+      "loss": 0.2735,
+      "step": 8970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6708302275703725e-06,
+      "loss": 0.2709,
+      "step": 8971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.668941784065623e-06,
+      "loss": 0.3055,
+      "step": 8972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6670539056026134e-06,
+      "loss": 0.2679,
+      "step": 8973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6651665923268555e-06,
+      "loss": 0.2426,
+      "step": 8974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.663279844383815e-06,
+      "loss": 0.2935,
+      "step": 8975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.661393661918912e-06,
+      "loss": 0.2654,
+      "step": 8976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6595080450775268e-06,
+      "loss": 0.2708,
+      "step": 8977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.657622994004986e-06,
+      "loss": 0.2574,
+      "step": 8978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6557385088465906e-06,
+      "loss": 0.2859,
+      "step": 8979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.65385458974758e-06,
+      "loss": 0.2717,
+      "step": 8980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.651971236853158e-06,
+      "loss": 0.271,
+      "step": 8981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6500884503084857e-06,
+      "loss": 0.2737,
+      "step": 8982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6482062302586774e-06,
+      "loss": 0.278,
+      "step": 8983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.646324576848809e-06,
+      "loss": 0.2837,
+      "step": 8984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6444434902239047e-06,
+      "loss": 0.326,
+      "step": 8985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6425629705289556e-06,
+      "loss": 0.3069,
+      "step": 8986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.640683017908895e-06,
+      "loss": 0.2611,
+      "step": 8987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6388036325086243e-06,
+      "loss": 0.2891,
+      "step": 8988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.636924814472995e-06,
+      "loss": 0.2443,
+      "step": 8989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6350465639468213e-06,
+      "loss": 0.2504,
+      "step": 8990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633168881074867e-06,
+      "loss": 0.3121,
+      "step": 8991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6312917660018534e-06,
+      "loss": 0.2984,
+      "step": 8992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6294152188724644e-06,
+      "loss": 0.2879,
+      "step": 8993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.627539239831328e-06,
+      "loss": 0.2751,
+      "step": 8994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6256638290230385e-06,
+      "loss": 0.2329,
+      "step": 8995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.623788986592144e-06,
+      "loss": 0.261,
+      "step": 8996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6219147126831467e-06,
+      "loss": 0.28,
+      "step": 8997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.620041007440508e-06,
+      "loss": 0.2652,
+      "step": 8998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.618167871008642e-06,
+      "loss": 0.2959,
+      "step": 8999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.616295303531926e-06,
+      "loss": 0.279,
+      "step": 9000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6144233051546797e-06,
+      "loss": 0.2689,
+      "step": 9001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6125518760211933e-06,
+      "loss": 0.2753,
+      "step": 9002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6106810162757046e-06,
+      "loss": 0.254,
+      "step": 9003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.608810726062412e-06,
+      "loss": 0.2568,
+      "step": 9004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6069410055254674e-06,
+      "loss": 0.2473,
+      "step": 9005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6050718548089804e-06,
+      "loss": 0.2919,
+      "step": 9006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6032032740570177e-06,
+      "loss": 0.2626,
+      "step": 9007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6013352634135957e-06,
+      "loss": 0.3216,
+      "step": 9008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.599467823022691e-06,
+      "loss": 0.6083,
+      "step": 9009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5976009530282455e-06,
+      "loss": 0.2357,
+      "step": 9010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5957346535741378e-06,
+      "loss": 0.2939,
+      "step": 9011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.593868924804218e-06,
+      "loss": 0.3043,
+      "step": 9012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.592003766862288e-06,
+      "loss": 0.2481,
+      "step": 9013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5901391798921018e-06,
+      "loss": 0.2776,
+      "step": 9014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5882751640373783e-06,
+      "loss": 0.2607,
+      "step": 9015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.586411719441777e-06,
+      "loss": 0.2996,
+      "step": 9016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5845488462489344e-06,
+      "loss": 0.3062,
+      "step": 9017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.582686544602423e-06,
+      "loss": 0.2727,
+      "step": 9018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5808248146457825e-06,
+      "loss": 0.2786,
+      "step": 9019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5789636565225063e-06,
+      "loss": 0.3326,
+      "step": 9020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5771030703760434e-06,
+      "loss": 0.3009,
+      "step": 9021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.575243056349801e-06,
+      "loss": 0.2975,
+      "step": 9022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5733836145871305e-06,
+      "loss": 0.2295,
+      "step": 9023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.571524745231361e-06,
+      "loss": 0.2505,
+      "step": 9024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569666448425756e-06,
+      "loss": 0.2752,
+      "step": 9025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5678087243135476e-06,
+      "loss": 0.2639,
+      "step": 9026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565951573037919e-06,
+      "loss": 0.2641,
+      "step": 9027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.56409499474201e-06,
+      "loss": 0.2936,
+      "step": 9028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.562238989568917e-06,
+      "loss": 0.2559,
+      "step": 9029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.560383557661692e-06,
+      "loss": 0.2814,
+      "step": 9030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.558528699163344e-06,
+      "loss": 0.2956,
+      "step": 9031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.556674414216833e-06,
+      "loss": 0.2402,
+      "step": 9032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5548207029650784e-06,
+      "loss": 0.2813,
+      "step": 9033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5529675655509567e-06,
+      "loss": 0.2469,
+      "step": 9034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5511150021172993e-06,
+      "loss": 0.3143,
+      "step": 9035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5492630128068895e-06,
+      "loss": 0.2601,
+      "step": 9036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5474115977624724e-06,
+      "loss": 0.2759,
+      "step": 9037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5455607571267484e-06,
+      "loss": 0.3003,
+      "step": 9038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5437104910423647e-06,
+      "loss": 0.2761,
+      "step": 9039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.541860799651934e-06,
+      "loss": 0.2429,
+      "step": 9040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5400116830980203e-06,
+      "loss": 0.2861,
+      "step": 9041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5381631415231455e-06,
+      "loss": 0.2895,
+      "step": 9042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5363151750697856e-06,
+      "loss": 0.2639,
+      "step": 9043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.534467783880373e-06,
+      "loss": 0.2704,
+      "step": 9044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532620968097299e-06,
+      "loss": 0.2745,
+      "step": 9045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.530774727862899e-06,
+      "loss": 0.2701,
+      "step": 9046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.528929063319475e-06,
+      "loss": 0.2693,
+      "step": 9047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5270839746092878e-06,
+      "loss": 0.2646,
+      "step": 9048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5252394618745415e-06,
+      "loss": 0.2801,
+      "step": 9049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5233955252574027e-06,
+      "loss": 0.2787,
+      "step": 9050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.521552164899994e-06,
+      "loss": 0.2725,
+      "step": 9051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5197093809443917e-06,
+      "loss": 0.2574,
+      "step": 9052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5178671735326333e-06,
+      "loss": 0.2852,
+      "step": 9053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.516025542806696e-06,
+      "loss": 0.2609,
+      "step": 9054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5141844889085365e-06,
+      "loss": 0.552,
+      "step": 9055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.512344011980045e-06,
+      "loss": 0.2188,
+      "step": 9056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.51050411216308e-06,
+      "loss": 0.2703,
+      "step": 9057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.508664789599451e-06,
+      "loss": 0.2961,
+      "step": 9058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5068260444309245e-06,
+      "loss": 0.3074,
+      "step": 9059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.504987876799224e-06,
+      "loss": 0.2694,
+      "step": 9060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503150286846019e-06,
+      "loss": 0.2668,
+      "step": 9061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.501313274712952e-06,
+      "loss": 0.2484,
+      "step": 9062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4994768405416024e-06,
+      "loss": 0.2653,
+      "step": 9063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.497640984473518e-06,
+      "loss": 0.2391,
+      "step": 9064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4958057066501952e-06,
+      "loss": 0.2624,
+      "step": 9065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4939710072130895e-06,
+      "loss": 0.2863,
+      "step": 9066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.492136886303611e-06,
+      "loss": 0.2831,
+      "step": 9067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.490303344063123e-06,
+      "loss": 0.2546,
+      "step": 9068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.48847038063295e-06,
+      "loss": 0.2824,
+      "step": 9069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.486637996154362e-06,
+      "loss": 0.2989,
+      "step": 9070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4848061907685915e-06,
+      "loss": 0.2344,
+      "step": 9071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4829749646168268e-06,
+      "loss": 0.5498,
+      "step": 9072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4811443178402097e-06,
+      "loss": 0.2698,
+      "step": 9073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4793142505798363e-06,
+      "loss": 0.2719,
+      "step": 9074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4774847629767594e-06,
+      "loss": 0.2679,
+      "step": 9075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.47565585517199e-06,
+      "loss": 0.2991,
+      "step": 9076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.473827527306486e-06,
+      "loss": 0.2684,
+      "step": 9077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4719997795211683e-06,
+      "loss": 0.2932,
+      "step": 9078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.470172611956909e-06,
+      "loss": 0.2335,
+      "step": 9079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.468346024754541e-06,
+      "loss": 0.2741,
+      "step": 9080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4665200180548454e-06,
+      "loss": 0.2979,
+      "step": 9081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.464694591998563e-06,
+      "loss": 0.275,
+      "step": 9082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4628697467263916e-06,
+      "loss": 0.3026,
+      "step": 9083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4610454823789742e-06,
+      "loss": 0.2519,
+      "step": 9084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.459221799096918e-06,
+      "loss": 0.2508,
+      "step": 9085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4573986970207906e-06,
+      "loss": 0.2575,
+      "step": 9086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4555761762911e-06,
+      "loss": 0.287,
+      "step": 9087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4537542370483203e-06,
+      "loss": 0.3323,
+      "step": 9088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4519328794328755e-06,
+      "loss": 0.2597,
+      "step": 9089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4501121035851494e-06,
+      "loss": 0.2785,
+      "step": 9090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4482919096454803e-06,
+      "loss": 0.2682,
+      "step": 9091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.44647229775415e-06,
+      "loss": 0.2406,
+      "step": 9092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.444653268051418e-06,
+      "loss": 0.5818,
+      "step": 9093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4428348206774775e-06,
+      "loss": 0.3411,
+      "step": 9094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441016955772487e-06,
+      "loss": 0.2684,
+      "step": 9095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.439199673476561e-06,
+      "loss": 0.2643,
+      "step": 9096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.437382973929764e-06,
+      "loss": 0.2584,
+      "step": 9097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4355668572721224e-06,
+      "loss": 0.2581,
+      "step": 9098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.433751323643606e-06,
+      "loss": 0.2538,
+      "step": 9099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.431936373184156e-06,
+      "loss": 0.2744,
+      "step": 9100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.430122006033653e-06,
+      "loss": 0.2372,
+      "step": 9101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.428308222331942e-06,
+      "loss": 0.2873,
+      "step": 9102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42649502221882e-06,
+      "loss": 0.2692,
+      "step": 9103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42468240583404e-06,
+      "loss": 0.2466,
+      "step": 9104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422870373317313e-06,
+      "loss": 0.2924,
+      "step": 9105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4210589248082914e-06,
+      "loss": 0.2752,
+      "step": 9106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4192480604466052e-06,
+      "loss": 0.2638,
+      "step": 9107
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4174377803718184e-06,
+      "loss": 0.2562,
+      "step": 9108
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.41562808472346e-06,
+      "loss": 0.3356,
+      "step": 9109
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4138189736410144e-06,
+      "loss": 0.2842,
+      "step": 9110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.412010447263917e-06,
+      "loss": 0.2661,
+      "step": 9111
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4102025057315615e-06,
+      "loss": 0.3003,
+      "step": 9112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4083951491832947e-06,
+      "loss": 0.3044,
+      "step": 9113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406588377758421e-06,
+      "loss": 0.2915,
+      "step": 9114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4047821915961923e-06,
+      "loss": 0.3181,
+      "step": 9115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.402976590835824e-06,
+      "loss": 0.2797,
+      "step": 9116
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.401171575616481e-06,
+      "loss": 0.2449,
+      "step": 9117
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.399367146077286e-06,
+      "loss": 0.2392,
+      "step": 9118
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3975633023573164e-06,
+      "loss": 0.2612,
+      "step": 9119
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.395760044595602e-06,
+      "loss": 0.2234,
+      "step": 9120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3939573729311325e-06,
+      "loss": 0.2739,
+      "step": 9121
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3921552875028443e-06,
+      "loss": 0.2234,
+      "step": 9122
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.390353788449631e-06,
+      "loss": 0.2551,
+      "step": 9123
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.388552875910354e-06,
+      "loss": 0.2613,
+      "step": 9124
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3867525500238086e-06,
+      "loss": 0.3028,
+      "step": 9125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.384952810928759e-06,
+      "loss": 0.2954,
+      "step": 9126
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3831536587639193e-06,
+      "loss": 0.278,
+      "step": 9127
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.38135509366796e-06,
+      "loss": 0.2799,
+      "step": 9128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.379557115779507e-06,
+      "loss": 0.2349,
+      "step": 9129
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.377759725237133e-06,
+      "loss": 0.2949,
+      "step": 9130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3759629221793823e-06,
+      "loss": 0.2809,
+      "step": 9131
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.374166706744735e-06,
+      "loss": 0.2585,
+      "step": 9132
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.372371079071638e-06,
+      "loss": 0.2952,
+      "step": 9133
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3705760392984887e-06,
+      "loss": 0.2566,
+      "step": 9134
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.368781587563641e-06,
+      "loss": 0.2382,
+      "step": 9135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.366987724005404e-06,
+      "loss": 0.2659,
+      "step": 9136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3651944487620315e-06,
+      "loss": 0.272,
+      "step": 9137
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.363401761971752e-06,
+      "loss": 0.2944,
+      "step": 9138
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3616096637727282e-06,
+      "loss": 0.2318,
+      "step": 9139
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3598181543030885e-06,
+      "loss": 0.3537,
+      "step": 9140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.358027233700915e-06,
+      "loss": 0.3098,
+      "step": 9141
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.356236902104242e-06,
+      "loss": 0.3176,
+      "step": 9142
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3544471596510633e-06,
+      "loss": 0.3326,
+      "step": 9143
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3526580064793125e-06,
+      "loss": 0.2855,
+      "step": 9144
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.350869442726903e-06,
+      "loss": 0.241,
+      "step": 9145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3490814685316777e-06,
+      "loss": 0.2601,
+      "step": 9146
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3472940840314496e-06,
+      "loss": 0.2604,
+      "step": 9147
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3455072893639784e-06,
+      "loss": 0.259,
+      "step": 9148
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3437210846669854e-06,
+      "loss": 0.2193,
+      "step": 9149
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3419354700781393e-06,
+      "loss": 0.3079,
+      "step": 9150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3401504457350677e-06,
+      "loss": 0.2879,
+      "step": 9151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.338366011775355e-06,
+      "loss": 0.255,
+      "step": 9152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3365821683365286e-06,
+      "loss": 0.2527,
+      "step": 9153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3347989155560835e-06,
+      "loss": 0.3129,
+      "step": 9154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.333016253571463e-06,
+      "loss": 0.2556,
+      "step": 9155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331234182520066e-06,
+      "loss": 0.254,
+      "step": 9156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3294527025392454e-06,
+      "loss": 0.2884,
+      "step": 9157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.32767181376631e-06,
+      "loss": 0.2582,
+      "step": 9158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3258915163385233e-06,
+      "loss": 0.2702,
+      "step": 9159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.324111810393097e-06,
+      "loss": 0.2562,
+      "step": 9160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322332696067202e-06,
+      "loss": 0.2828,
+      "step": 9161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.320554173497972e-06,
+      "loss": 0.3104,
+      "step": 9162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.318776242822478e-06,
+      "loss": 0.2874,
+      "step": 9163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3169989041777565e-06,
+      "loss": 0.2489,
+      "step": 9164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.315222157700797e-06,
+      "loss": 0.3329,
+      "step": 9165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3134460035285433e-06,
+      "loss": 0.2841,
+      "step": 9166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.311670441797893e-06,
+      "loss": 0.2308,
+      "step": 9167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.30989547264569e-06,
+      "loss": 0.2709,
+      "step": 9168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.308121096208752e-06,
+      "loss": 0.2532,
+      "step": 9169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.30634731262383e-06,
+      "loss": 0.2775,
+      "step": 9170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3045741220276428e-06,
+      "loss": 0.3373,
+      "step": 9171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.302801524556857e-06,
+      "loss": 0.2536,
+      "step": 9172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.301029520348097e-06,
+      "loss": 0.2801,
+      "step": 9173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.299258109537943e-06,
+      "loss": 0.2907,
+      "step": 9174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.297487292262918e-06,
+      "loss": 0.2947,
+      "step": 9175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2957170686595176e-06,
+      "loss": 0.2836,
+      "step": 9176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2939474388641757e-06,
+      "loss": 0.2401,
+      "step": 9177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2921784030132886e-06,
+      "loss": 0.2247,
+      "step": 9178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290409961243204e-06,
+      "loss": 0.2394,
+      "step": 9179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2886421136902257e-06,
+      "loss": 0.2419,
+      "step": 9180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2868748604906145e-06,
+      "loss": 0.3106,
+      "step": 9181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2851082017805704e-06,
+      "loss": 0.3028,
+      "step": 9182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2833421376962718e-06,
+      "loss": 0.2422,
+      "step": 9183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.281576668373828e-06,
+      "loss": 0.2794,
+      "step": 9184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.279811793949318e-06,
+      "loss": 0.2437,
+      "step": 9185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.278047514558769e-06,
+      "loss": 0.2793,
+      "step": 9186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2762838303381607e-06,
+      "loss": 0.2706,
+      "step": 9187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.274520741423435e-06,
+      "loss": 0.2377,
+      "step": 9188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2727582479504704e-06,
+      "loss": 0.2642,
+      "step": 9189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.270996350055126e-06,
+      "loss": 0.2374,
+      "step": 9190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2692350478731893e-06,
+      "loss": 0.2606,
+      "step": 9191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2674743415404166e-06,
+      "loss": 0.2974,
+      "step": 9192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265714231192514e-06,
+      "loss": 0.2502,
+      "step": 9193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2639547169651423e-06,
+      "loss": 0.2722,
+      "step": 9194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262195798993916e-06,
+      "loss": 0.3166,
+      "step": 9195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.260437477414403e-06,
+      "loss": 0.2901,
+      "step": 9196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2586797523621306e-06,
+      "loss": 0.2933,
+      "step": 9197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2569226239725695e-06,
+      "loss": 0.2832,
+      "step": 9198
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.255166092381149e-06,
+      "loss": 0.2413,
+      "step": 9199
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2534101577232647e-06,
+      "loss": 0.251,
+      "step": 9200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2516548201342445e-06,
+      "loss": 0.2373,
+      "step": 9201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249900079749385e-06,
+      "loss": 0.2757,
+      "step": 9202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.248145936703934e-06,
+      "loss": 0.3094,
+      "step": 9203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.246392391133091e-06,
+      "loss": 0.248,
+      "step": 9204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.244639443172013e-06,
+      "loss": 0.2707,
+      "step": 9205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2428870929558012e-06,
+      "loss": 0.2359,
+      "step": 9206
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.241135340619528e-06,
+      "loss": 0.2504,
+      "step": 9207
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2393841862982036e-06,
+      "loss": 0.2808,
+      "step": 9208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2376336301267985e-06,
+      "loss": 0.2327,
+      "step": 9209
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.235883672240239e-06,
+      "loss": 0.3121,
+      "step": 9210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2341343127734028e-06,
+      "loss": 0.2801,
+      "step": 9211
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2323855518611227e-06,
+      "loss": 0.2717,
+      "step": 9212
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2306373896381795e-06,
+      "loss": 0.2492,
+      "step": 9213
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2288898262393212e-06,
+      "loss": 0.2594,
+      "step": 9214
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.227142861799235e-06,
+      "loss": 0.2529,
+      "step": 9215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.22539649645257e-06,
+      "loss": 0.2261,
+      "step": 9216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2236507303339273e-06,
+      "loss": 0.2546,
+      "step": 9217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2219055635778618e-06,
+      "loss": 0.2586,
+      "step": 9218
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.220160996318886e-06,
+      "loss": 0.2361,
+      "step": 9219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2184170286914543e-06,
+      "loss": 0.2383,
+      "step": 9220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.216673660829992e-06,
+      "loss": 0.2538,
+      "step": 9221
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214930892868864e-06,
+      "loss": 0.2588,
+      "step": 9222
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2131887249423957e-06,
+      "loss": 0.3054,
+      "step": 9223
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.211447157184864e-06,
+      "loss": 0.2628,
+      "step": 9224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2097061897305016e-06,
+      "loss": 0.2438,
+      "step": 9225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207965822713496e-06,
+      "loss": 0.2842,
+      "step": 9226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2062260562679773e-06,
+      "loss": 0.2513,
+      "step": 9227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2044868905280504e-06,
+      "loss": 0.2938,
+      "step": 9228
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2027483256277517e-06,
+      "loss": 0.2933,
+      "step": 9229
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2010103617010836e-06,
+      "loss": 0.2712,
+      "step": 9230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1992729988820026e-06,
+      "loss": 0.2474,
+      "step": 9231
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.197536237304414e-06,
+      "loss": 0.2415,
+      "step": 9232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.195800077102178e-06,
+      "loss": 0.2503,
+      "step": 9233
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1940645184091115e-06,
+      "loss": 0.3033,
+      "step": 9234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1923295613589846e-06,
+      "loss": 0.2966,
+      "step": 9235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.190595206085513e-06,
+      "loss": 0.2696,
+      "step": 9236
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.188861452722373e-06,
+      "loss": 0.2515,
+      "step": 9237
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1871283014032007e-06,
+      "loss": 0.2637,
+      "step": 9238
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1853957522615732e-06,
+      "loss": 0.2508,
+      "step": 9239
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1836638054310265e-06,
+      "loss": 0.3,
+      "step": 9240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.181932461045053e-06,
+      "loss": 0.2824,
+      "step": 9241
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1802017192370963e-06,
+      "loss": 0.575,
+      "step": 9242
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178471580140553e-06,
+      "loss": 0.313,
+      "step": 9243
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.176742043888769e-06,
+      "loss": 0.2617,
+      "step": 9244
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1750131106150563e-06,
+      "loss": 0.247,
+      "step": 9245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.173284780452667e-06,
+      "loss": 0.2808,
+      "step": 9246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.171557053534814e-06,
+      "loss": 0.6298,
+      "step": 9247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.169829929994661e-06,
+      "loss": 0.2391,
+      "step": 9248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1681034099653287e-06,
+      "loss": 0.2638,
+      "step": 9249
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1663774935798886e-06,
+      "loss": 0.2831,
+      "step": 9250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1646521809713583e-06,
+      "loss": 0.2561,
+      "step": 9251
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.162927472272728e-06,
+      "loss": 0.272,
+      "step": 9252
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.161203367616922e-06,
+      "loss": 0.2667,
+      "step": 9253
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1594798671368265e-06,
+      "loss": 0.2907,
+      "step": 9254
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1577569709652833e-06,
+      "loss": 0.2775,
+      "step": 9255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156034679235083e-06,
+      "loss": 0.2592,
+      "step": 9256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.154312992078973e-06,
+      "loss": 0.2587,
+      "step": 9257
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1525919096296455e-06,
+      "loss": 0.2728,
+      "step": 9258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1508714320197644e-06,
+      "loss": 0.2815,
+      "step": 9259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1491515593819266e-06,
+      "loss": 0.2713,
+      "step": 9260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1474322918486956e-06,
+      "loss": 0.2749,
+      "step": 9261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1457136295525817e-06,
+      "loss": 0.3287,
+      "step": 9262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.143995572626052e-06,
+      "loss": 0.3051,
+      "step": 9263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1422781212015286e-06,
+      "loss": 0.2537,
+      "step": 9264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.140561275411377e-06,
+      "loss": 0.2739,
+      "step": 9265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.138845035387932e-06,
+      "loss": 0.3284,
+      "step": 9266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1371294012634667e-06,
+      "loss": 0.2881,
+      "step": 9267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.135414373170215e-06,
+      "loss": 0.2631,
+      "step": 9268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1336999512403633e-06,
+      "loss": 0.2816,
+      "step": 9269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.131986135606051e-06,
+      "loss": 0.2986,
+      "step": 9270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.130272926399374e-06,
+      "loss": 0.2806,
+      "step": 9271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1285603237523677e-06,
+      "loss": 0.3137,
+      "step": 9272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1268483277970442e-06,
+      "loss": 0.2766,
+      "step": 9273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1251369386653454e-06,
+      "loss": 0.2699,
+      "step": 9274
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.123426156489178e-06,
+      "loss": 0.2676,
+      "step": 9275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1217159814004096e-06,
+      "loss": 0.2808,
+      "step": 9276
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.120006413530842e-06,
+      "loss": 0.2858,
+      "step": 9277
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1182974530122435e-06,
+      "loss": 0.2707,
+      "step": 9278
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116589099976334e-06,
+      "loss": 0.2557,
+      "step": 9279
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114881354554782e-06,
+      "loss": 0.2247,
+      "step": 9280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.113174216879218e-06,
+      "loss": 0.3052,
+      "step": 9281
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.111467687081209e-06,
+      "loss": 0.2475,
+      "step": 9282
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1097617652922973e-06,
+      "loss": 0.243,
+      "step": 9283
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1080564516439605e-06,
+      "loss": 0.3114,
+      "step": 9284
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1063517462676365e-06,
+      "loss": 0.2689,
+      "step": 9285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1046476492947155e-06,
+      "loss": 0.2492,
+      "step": 9286
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1029441608565425e-06,
+      "loss": 0.2549,
+      "step": 9287
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.101241281084416e-06,
+      "loss": 0.278,
+      "step": 9288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099539010109577e-06,
+      "loss": 0.2597,
+      "step": 9289
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0978373480632386e-06,
+      "loss": 0.2695,
+      "step": 9290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0961362950765495e-06,
+      "loss": 0.2772,
+      "step": 9291
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0944358512806207e-06,
+      "loss": 0.259,
+      "step": 9292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0927360168065135e-06,
+      "loss": 0.2773,
+      "step": 9293
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0910367917852437e-06,
+      "loss": 0.3198,
+      "step": 9294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0893381763477816e-06,
+      "loss": 0.272,
+      "step": 9295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.087640170625039e-06,
+      "loss": 0.2737,
+      "step": 9296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.085942774747901e-06,
+      "loss": 0.2946,
+      "step": 9297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.084245988847188e-06,
+      "loss": 0.2377,
+      "step": 9298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0825498130536804e-06,
+      "loss": 0.238,
+      "step": 9299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.080854247498112e-06,
+      "loss": 0.2417,
+      "step": 9300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.079159292311169e-06,
+      "loss": 0.2924,
+      "step": 9301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.077464947623492e-06,
+      "loss": 0.2801,
+      "step": 9302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.075771213565665e-06,
+      "loss": 0.2786,
+      "step": 9303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0740780902682444e-06,
+      "loss": 0.2521,
+      "step": 9304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0723855778617175e-06,
+      "loss": 0.2524,
+      "step": 9305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0706936764765393e-06,
+      "loss": 0.2787,
+      "step": 9306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.069002386243113e-06,
+      "loss": 0.6023,
+      "step": 9307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067311707291794e-06,
+      "loss": 0.2877,
+      "step": 9308
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.065621639752895e-06,
+      "loss": 0.2395,
+      "step": 9309
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0639321837566696e-06,
+      "loss": 0.3077,
+      "step": 9310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0622433394333443e-06,
+      "loss": 0.2843,
+      "step": 9311
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0605551069130767e-06,
+      "loss": 0.2504,
+      "step": 9312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0588674863259907e-06,
+      "loss": 0.2914,
+      "step": 9313
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.057180477802164e-06,
+      "loss": 0.2488,
+      "step": 9314
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0554940814716174e-06,
+      "loss": 0.2402,
+      "step": 9315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0538082974643325e-06,
+      "loss": 0.2429,
+      "step": 9316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0521231259102404e-06,
+      "loss": 0.2727,
+      "step": 9317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0504385669392268e-06,
+      "loss": 0.3358,
+      "step": 9318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0487546206811304e-06,
+      "loss": 0.2772,
+      "step": 9319
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047071287265735e-06,
+      "loss": 0.2646,
+      "step": 9320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0453885668227923e-06,
+      "loss": 0.261,
+      "step": 9321
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043706459481992e-06,
+      "loss": 0.2285,
+      "step": 9322
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.042024965372985e-06,
+      "loss": 0.2607,
+      "step": 9323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040344084625372e-06,
+      "loss": 0.265,
+      "step": 9324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0386638173687067e-06,
+      "loss": 0.2697,
+      "step": 9325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0369841637324992e-06,
+      "loss": 0.2578,
+      "step": 9326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0353051238462006e-06,
+      "loss": 0.2763,
+      "step": 9327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033626697839234e-06,
+      "loss": 0.6049,
+      "step": 9328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0319488858409552e-06,
+      "loss": 0.5861,
+      "step": 9329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030271687980685e-06,
+      "loss": 0.315,
+      "step": 9330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0285951043876937e-06,
+      "loss": 0.2921,
+      "step": 9331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0269191351912042e-06,
+      "loss": 0.313,
+      "step": 9332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.025243780520394e-06,
+      "loss": 0.2839,
+      "step": 9333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.023569040504384e-06,
+      "loss": 0.2872,
+      "step": 9334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0218949152722643e-06,
+      "loss": 0.2852,
+      "step": 9335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.020221404953061e-06,
+      "loss": 0.2509,
+      "step": 9336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.018548509675763e-06,
+      "loss": 0.2587,
+      "step": 9337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.016876229569308e-06,
+      "loss": 0.2901,
+      "step": 9338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0152045647625874e-06,
+      "loss": 0.2545,
+      "step": 9339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.013533515384447e-06,
+      "loss": 0.2796,
+      "step": 9340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0118630815636763e-06,
+      "loss": 0.2828,
+      "step": 9341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0101932634290345e-06,
+      "loss": 0.3207,
+      "step": 9342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0085240611092137e-06,
+      "loss": 0.2821,
+      "step": 9343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006855474732872e-06,
+      "loss": 0.2629,
+      "step": 9344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0051875044286138e-06,
+      "loss": 0.2532,
+      "step": 9345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.003520150325e-06,
+      "loss": 0.2896,
+      "step": 9346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.001853412550544e-06,
+      "loss": 0.2838,
+      "step": 9347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0001872912337016e-06,
+      "loss": 0.2639,
+      "step": 9348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9985217865029005e-06,
+      "loss": 0.2825,
+      "step": 9349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9968568984865e-06,
+      "loss": 0.2417,
+      "step": 9350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.995192627312823e-06,
+      "loss": 0.2795,
+      "step": 9351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9935289731101503e-06,
+      "loss": 0.3243,
+      "step": 9352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9918659360067005e-06,
+      "loss": 0.2706,
+      "step": 9353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9902035161306574e-06,
+      "loss": 0.2656,
+      "step": 9354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9885417136101446e-06,
+      "loss": 0.2835,
+      "step": 9355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9868805285732538e-06,
+      "loss": 0.2835,
+      "step": 9356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9852199611480207e-06,
+      "loss": 0.2807,
+      "step": 9357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983560011462425e-06,
+      "loss": 0.265,
+      "step": 9358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9819006796444185e-06,
+      "loss": 0.2463,
+      "step": 9359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9802419658218873e-06,
+      "loss": 0.2498,
+      "step": 9360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978583870122678e-06,
+      "loss": 0.2531,
+      "step": 9361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9769263926745886e-06,
+      "loss": 0.2675,
+      "step": 9362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9752695336053697e-06,
+      "loss": 0.26,
+      "step": 9363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9736132930427263e-06,
+      "loss": 0.2839,
+      "step": 9364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.971957671114306e-06,
+      "loss": 0.2706,
+      "step": 9365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9703026679477253e-06,
+      "loss": 0.2814,
+      "step": 9366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.968648283670538e-06,
+      "loss": 0.2225,
+      "step": 9367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9669945184102555e-06,
+      "loss": 0.2918,
+      "step": 9368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9653413722943437e-06,
+      "loss": 0.2851,
+      "step": 9369
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.963688845450218e-06,
+      "loss": 0.2845,
+      "step": 9370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9620369380052507e-06,
+      "loss": 0.2684,
+      "step": 9371
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9603856500867537e-06,
+      "loss": 0.2231,
+      "step": 9372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9587349818220113e-06,
+      "loss": 0.2791,
+      "step": 9373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.957084933338241e-06,
+      "loss": 0.2742,
+      "step": 9374
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.955435504762624e-06,
+      "loss": 0.265,
+      "step": 9375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.953786696222287e-06,
+      "loss": 0.2646,
+      "step": 9376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9521385078443156e-06,
+      "loss": 0.2541,
+      "step": 9377
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9504909397557436e-06,
+      "loss": 0.2821,
+      "step": 9378
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.948843992083551e-06,
+      "loss": 0.2827,
+      "step": 9379
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9471976649546876e-06,
+      "loss": 0.2851,
+      "step": 9380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945551958496035e-06,
+      "loss": 0.2683,
+      "step": 9381
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94390687283444e-06,
+      "loss": 0.316,
+      "step": 9382
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9422624080966956e-06,
+      "loss": 0.2656,
+      "step": 9383
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940618564409551e-06,
+      "loss": 0.3015,
+      "step": 9384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938975341899708e-06,
+      "loss": 0.2662,
+      "step": 9385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.937332740693809e-06,
+      "loss": 0.3258,
+      "step": 9386
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9356907609184695e-06,
+      "loss": 0.3202,
+      "step": 9387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9340494027002365e-06,
+      "loss": 0.3433,
+      "step": 9388
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.932408666165617e-06,
+      "loss": 0.2412,
+      "step": 9389
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9307685514410803e-06,
+      "loss": 0.2311,
+      "step": 9390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.92912905865303e-06,
+      "loss": 0.237,
+      "step": 9391
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9274901879278342e-06,
+      "loss": 0.3076,
+      "step": 9392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925851939391803e-06,
+      "loss": 0.2809,
+      "step": 9393
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.924214313171211e-06,
+      "loss": 0.2503,
+      "step": 9394
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9225773093922785e-06,
+      "loss": 0.2345,
+      "step": 9395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.920940928181171e-06,
+      "loss": 0.2635,
+      "step": 9396
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919305169664021e-06,
+      "loss": 0.257,
+      "step": 9397
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9176700339668986e-06,
+      "loss": 0.3221,
+      "step": 9398
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9160355212158345e-06,
+      "loss": 0.2471,
+      "step": 9399
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9144016315368075e-06,
+      "loss": 0.2194,
+      "step": 9400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9127683650557505e-06,
+      "loss": 0.2868,
+      "step": 9401
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9111357218985504e-06,
+      "loss": 0.5947,
+      "step": 9402
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9095037021910366e-06,
+      "loss": 0.2792,
+      "step": 9403
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9078723060590053e-06,
+      "loss": 0.2946,
+      "step": 9404
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9062415336281904e-06,
+      "loss": 0.2424,
+      "step": 9405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9046113850242843e-06,
+      "loss": 0.5575,
+      "step": 9406
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9029818603729332e-06,
+      "loss": 0.2314,
+      "step": 9407
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9013529597997315e-06,
+      "loss": 0.2533,
+      "step": 9408
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8997246834302297e-06,
+      "loss": 0.2859,
+      "step": 9409
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8980970313899193e-06,
+      "loss": 0.2224,
+      "step": 9410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8964700038042628e-06,
+      "loss": 0.2674,
+      "step": 9411
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.894843600798655e-06,
+      "loss": 0.2587,
+      "step": 9412
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8932178224984533e-06,
+      "loss": 0.261,
+      "step": 9413
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8915926690289643e-06,
+      "loss": 0.2709,
+      "step": 9414
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8899681405154491e-06,
+      "loss": 0.2723,
+      "step": 9415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8883442370831183e-06,
+      "loss": 0.2507,
+      "step": 9416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8867209588571288e-06,
+      "loss": 0.2488,
+      "step": 9417
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8850983059626026e-06,
+      "loss": 0.2751,
+      "step": 9418
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8834762785246007e-06,
+      "loss": 0.2551,
+      "step": 9419
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.881854876668142e-06,
+      "loss": 0.282,
+      "step": 9420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8802341005181957e-06,
+      "loss": 0.2819,
+      "step": 9421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8786139501996847e-06,
+      "loss": 0.5848,
+      "step": 9422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.876994425837484e-06,
+      "loss": 0.2703,
+      "step": 9423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8753755275564112e-06,
+      "loss": 0.281,
+      "step": 9424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8737572554812522e-06,
+      "loss": 0.244,
+      "step": 9425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8721396097367294e-06,
+      "loss": 0.2943,
+      "step": 9426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.870522590447521e-06,
+      "loss": 0.2612,
+      "step": 9427
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8689061977382684e-06,
+      "loss": 0.2703,
+      "step": 9428
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.867290431733546e-06,
+      "loss": 0.2958,
+      "step": 9429
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8656752925578948e-06,
+      "loss": 0.2273,
+      "step": 9430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8640607803357936e-06,
+      "loss": 0.2496,
+      "step": 9431
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8624468951916896e-06,
+      "loss": 0.2704,
+      "step": 9432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8608336372499736e-06,
+      "loss": 0.2464,
+      "step": 9433
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8592210066349781e-06,
+      "loss": 0.2734,
+      "step": 9434
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.857609003471007e-06,
+      "loss": 0.279,
+      "step": 9435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8559976278823e-06,
+      "loss": 0.2687,
+      "step": 9436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8543868799930542e-06,
+      "loss": 0.2455,
+      "step": 9437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8527767599274193e-06,
+      "loss": 0.2799,
+      "step": 9438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8511672678094949e-06,
+      "loss": 0.2882,
+      "step": 9439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8495584037633364e-06,
+      "loss": 0.3014,
+      "step": 9440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8479501679129375e-06,
+      "loss": 0.2403,
+      "step": 9441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846342560382265e-06,
+      "loss": 0.2645,
+      "step": 9442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.844735581295216e-06,
+      "loss": 0.263,
+      "step": 9443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8431292307756532e-06,
+      "loss": 0.2413,
+      "step": 9444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8415235089473848e-06,
+      "loss": 0.2513,
+      "step": 9445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.839918415934171e-06,
+      "loss": 0.2727,
+      "step": 9446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8383139518597293e-06,
+      "loss": 0.347,
+      "step": 9447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8367101168477152e-06,
+      "loss": 0.2689,
+      "step": 9448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8351069110217535e-06,
+      "loss": 0.2823,
+      "step": 9449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8335043345054048e-06,
+      "loss": 0.2725,
+      "step": 9450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.831902387422191e-06,
+      "loss": 0.278,
+      "step": 9451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8303010698955803e-06,
+      "loss": 0.265,
+      "step": 9452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8287003820489956e-06,
+      "loss": 0.2759,
+      "step": 9453
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8271003240058127e-06,
+      "loss": 0.2891,
+      "step": 9454
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8255008958893483e-06,
+      "loss": 0.2639,
+      "step": 9455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8239020978228894e-06,
+      "loss": 0.248,
+      "step": 9456
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.822303929929654e-06,
+      "loss": 0.2601,
+      "step": 9457
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.820706392332824e-06,
+      "loss": 0.6052,
+      "step": 9458
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8191094851555314e-06,
+      "loss": 0.2992,
+      "step": 9459
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8175132085208558e-06,
+      "loss": 0.2775,
+      "step": 9460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8159175625518344e-06,
+      "loss": 0.2957,
+      "step": 9461
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.814322547371443e-06,
+      "loss": 0.2314,
+      "step": 9462
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8127281631026284e-06,
+      "loss": 0.2336,
+      "step": 9463
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8111344098682703e-06,
+      "loss": 0.3176,
+      "step": 9464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8095412877912056e-06,
+      "loss": 0.265,
+      "step": 9465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8079487969942344e-06,
+      "loss": 0.3464,
+      "step": 9466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.806356937600089e-06,
+      "loss": 0.2669,
+      "step": 9467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8047657097314675e-06,
+      "loss": 0.2624,
+      "step": 9468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8031751135110065e-06,
+      "loss": 0.2947,
+      "step": 9469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8015851490613079e-06,
+      "loss": 0.275,
+      "step": 9470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799995816504919e-06,
+      "loss": 0.2711,
+      "step": 9471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7984071159643312e-06,
+      "loss": 0.251,
+      "step": 9472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7968190475620018e-06,
+      "loss": 0.2641,
+      "step": 9473
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.795231611420325e-06,
+      "loss": 0.313,
+      "step": 9474
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7936448076616542e-06,
+      "loss": 0.3232,
+      "step": 9475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7920586364082926e-06,
+      "loss": 0.3124,
+      "step": 9476
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7904730977824958e-06,
+      "loss": 0.2873,
+      "step": 9477
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7888881919064694e-06,
+      "loss": 0.2629,
+      "step": 9478
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7873039189023644e-06,
+      "loss": 0.2428,
+      "step": 9479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7857202788922977e-06,
+      "loss": 0.2529,
+      "step": 9480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.784137271998323e-06,
+      "loss": 0.2679,
+      "step": 9481
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.78255489834245e-06,
+      "loss": 0.2695,
+      "step": 9482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7809731580466427e-06,
+      "loss": 0.2533,
+      "step": 9483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7793920512328122e-06,
+      "loss": 0.2829,
+      "step": 9484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7778115780228267e-06,
+      "loss": 0.2685,
+      "step": 9485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.776231738538492e-06,
+      "loss": 0.3014,
+      "step": 9486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7746525329015852e-06,
+      "loss": 0.263,
+      "step": 9487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7730739612338166e-06,
+      "loss": 0.2858,
+      "step": 9488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7714960236568556e-06,
+      "loss": 0.2525,
+      "step": 9489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7699187202923241e-06,
+      "loss": 0.282,
+      "step": 9490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.76834205126179e-06,
+      "loss": 0.3347,
+      "step": 9491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7667660166867806e-06,
+      "loss": 0.2742,
+      "step": 9492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.76519061668876e-06,
+      "loss": 0.2836,
+      "step": 9493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.763615851389161e-06,
+      "loss": 0.2911,
+      "step": 9494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7620417209093544e-06,
+      "loss": 0.2878,
+      "step": 9495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7604682253706652e-06,
+      "loss": 0.2726,
+      "step": 9496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7588953648943742e-06,
+      "loss": 0.5618,
+      "step": 9497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7573231396017064e-06,
+      "loss": 0.3093,
+      "step": 9498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7557515496138455e-06,
+      "loss": 0.2517,
+      "step": 9499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7541805950519154e-06,
+      "loss": 0.2678,
+      "step": 9500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7526102760370056e-06,
+      "loss": 0.2991,
+      "step": 9501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7510405926901408e-06,
+      "loss": 0.2891,
+      "step": 9502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7494715451323063e-06,
+      "loss": 0.2471,
+      "step": 9503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7479031334844421e-06,
+      "loss": 0.2979,
+      "step": 9504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.746335357867428e-06,
+      "loss": 0.2501,
+      "step": 9505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7447682184021042e-06,
+      "loss": 0.2697,
+      "step": 9506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7432017152092507e-06,
+      "loss": 0.2668,
+      "step": 9507
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7416358484096141e-06,
+      "loss": 0.2687,
+      "step": 9508
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7400706181238824e-06,
+      "loss": 0.2621,
+      "step": 9509
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7385060244726882e-06,
+      "loss": 0.2627,
+      "step": 9510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7369420675766347e-06,
+      "loss": 0.2925,
+      "step": 9511
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7353787475562544e-06,
+      "loss": 0.3026,
+      "step": 9512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7338160645320435e-06,
+      "loss": 0.2543,
+      "step": 9513
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7322540186244462e-06,
+      "loss": 0.2983,
+      "step": 9514
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.730692609953858e-06,
+      "loss": 0.2759,
+      "step": 9515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7291318386406241e-06,
+      "loss": 0.338,
+      "step": 9516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7275717048050367e-06,
+      "loss": 0.2865,
+      "step": 9517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7260122085673525e-06,
+      "loss": 0.3047,
+      "step": 9518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7244533500477612e-06,
+      "loss": 0.2775,
+      "step": 9519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7228951293664142e-06,
+      "loss": 0.285,
+      "step": 9520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7213375466434134e-06,
+      "loss": 0.2306,
+      "step": 9521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7197806019988084e-06,
+      "loss": 0.2451,
+      "step": 9522
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7182242955526029e-06,
+      "loss": 0.2715,
+      "step": 9523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7166686274247424e-06,
+      "loss": 0.238,
+      "step": 9524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7151135977351397e-06,
+      "loss": 0.2604,
+      "step": 9525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.713559206603642e-06,
+      "loss": 0.3048,
+      "step": 9526
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7120054541500552e-06,
+      "loss": 0.5493,
+      "step": 9527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7104523404941365e-06,
+      "loss": 0.2971,
+      "step": 9528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088998657555922e-06,
+      "loss": 0.2821,
+      "step": 9529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7073480300540802e-06,
+      "loss": 0.2471,
+      "step": 9530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7057968335092024e-06,
+      "loss": 0.2377,
+      "step": 9531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7042462762405265e-06,
+      "loss": 0.2661,
+      "step": 9532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7026963583675549e-06,
+      "loss": 0.2929,
+      "step": 9533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7011470800097496e-06,
+      "loss": 0.3049,
+      "step": 9534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6995984412865218e-06,
+      "loss": 0.2688,
+      "step": 9535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6980504423172317e-06,
+      "loss": 0.2785,
+      "step": 9536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.696503083221196e-06,
+      "loss": 0.2885,
+      "step": 9537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.694956364117668e-06,
+      "loss": 0.2571,
+      "step": 9538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6934102851258726e-06,
+      "loss": 0.285,
+      "step": 9539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6918648463649668e-06,
+      "loss": 0.2803,
+      "step": 9540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6903200479540627e-06,
+      "loss": 0.2862,
+      "step": 9541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6887758900122352e-06,
+      "loss": 0.301,
+      "step": 9542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6872323726584938e-06,
+      "loss": 0.2347,
+      "step": 9543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6856894960118087e-06,
+      "loss": 0.2665,
+      "step": 9544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6841472601910892e-06,
+      "loss": 0.2335,
+      "step": 9545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6826056653152122e-06,
+      "loss": 0.5646,
+      "step": 9546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6810647115029954e-06,
+      "loss": 0.3072,
+      "step": 9547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6795243988732e-06,
+      "loss": 0.3416,
+      "step": 9548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.677984727544557e-06,
+      "loss": 0.2332,
+      "step": 9549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6764456976357279e-06,
+      "loss": 0.3188,
+      "step": 9550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.674907309265338e-06,
+      "loss": 0.287,
+      "step": 9551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6733695625519553e-06,
+      "loss": 0.286,
+      "step": 9552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6718324576141043e-06,
+      "loss": 0.3089,
+      "step": 9553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.67029599457026e-06,
+      "loss": 0.2725,
+      "step": 9554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6687601735388358e-06,
+      "loss": 0.2493,
+      "step": 9555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6672249946382179e-06,
+      "loss": 0.2916,
+      "step": 9556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6656904579867205e-06,
+      "loss": 0.2604,
+      "step": 9557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6641565637026225e-06,
+      "loss": 0.2554,
+      "step": 9558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6626233119041468e-06,
+      "loss": 0.2863,
+      "step": 9559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6610907027094714e-06,
+      "loss": 0.2711,
+      "step": 9560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6595587362367226e-06,
+      "loss": 0.2813,
+      "step": 9561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6580274126039698e-06,
+      "loss": 0.2722,
+      "step": 9562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6564967319292502e-06,
+      "loss": 0.3035,
+      "step": 9563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6549666943305342e-06,
+      "loss": 0.2395,
+      "step": 9564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.653437299925751e-06,
+      "loss": 0.2979,
+      "step": 9565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.651908548832779e-06,
+      "loss": 0.2621,
+      "step": 9566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6503804411694468e-06,
+      "loss": 0.2891,
+      "step": 9567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6488529770535367e-06,
+      "loss": 0.2832,
+      "step": 9568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6473261566027687e-06,
+      "loss": 0.2579,
+      "step": 9569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6457999799348345e-06,
+      "loss": 0.3221,
+      "step": 9570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6442744471673566e-06,
+      "loss": 0.2507,
+      "step": 9571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6427495584179165e-06,
+      "loss": 0.2961,
+      "step": 9572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6412253138040467e-06,
+      "loss": 0.2866,
+      "step": 9573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6397017134432281e-06,
+      "loss": 0.2695,
+      "step": 9574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.638178757452894e-06,
+      "loss": 0.28,
+      "step": 9575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6366564459504186e-06,
+      "loss": 0.2577,
+      "step": 9576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6351347790531457e-06,
+      "loss": 0.2874,
+      "step": 9577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6336137568783495e-06,
+      "loss": 0.6031,
+      "step": 9578
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6320933795432626e-06,
+      "loss": 0.2919,
+      "step": 9579
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6305736471650756e-06,
+      "loss": 0.2827,
+      "step": 9580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6290545598609165e-06,
+      "loss": 0.3348,
+      "step": 9581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.627536117747871e-06,
+      "loss": 0.2628,
+      "step": 9582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.626018320942967e-06,
+      "loss": 0.2819,
+      "step": 9583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6245011695631962e-06,
+      "loss": 0.2634,
+      "step": 9584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6229846637254932e-06,
+      "loss": 0.2584,
+      "step": 9585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6214688035467363e-06,
+      "loss": 0.2779,
+      "step": 9586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6199535891437678e-06,
+      "loss": 0.2803,
+      "step": 9587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6184390206333688e-06,
+      "loss": 0.3201,
+      "step": 9588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.616925098132275e-06,
+      "loss": 0.2339,
+      "step": 9589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6154118217571723e-06,
+      "loss": 0.2957,
+      "step": 9590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.613899191624697e-06,
+      "loss": 0.2483,
+      "step": 9591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.612387207851437e-06,
+      "loss": 0.2757,
+      "step": 9592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.610875870553923e-06,
+      "loss": 0.2422,
+      "step": 9593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6093651798486487e-06,
+      "loss": 0.263,
+      "step": 9594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6078551358520456e-06,
+      "loss": 0.2716,
+      "step": 9595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6063457386805004e-06,
+      "loss": 0.2653,
+      "step": 9596
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6048369884503524e-06,
+      "loss": 0.2645,
+      "step": 9597
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6033288852778882e-06,
+      "loss": 0.2669,
+      "step": 9598
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6018214292793455e-06,
+      "loss": 0.2472,
+      "step": 9599
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6003146205709064e-06,
+      "loss": 0.2523,
+      "step": 9600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5988084592687169e-06,
+      "loss": 0.2884,
+      "step": 9601
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5973029454888578e-06,
+      "loss": 0.2417,
+      "step": 9602
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5957980793473682e-06,
+      "loss": 0.2799,
+      "step": 9603
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5942938609602365e-06,
+      "loss": 0.28,
+      "step": 9604
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5927902904434e-06,
+      "loss": 0.2823,
+      "step": 9605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5912873679127495e-06,
+      "loss": 0.2827,
+      "step": 9606
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.589785093484114e-06,
+      "loss": 0.5929,
+      "step": 9607
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5882834672732939e-06,
+      "loss": 0.2723,
+      "step": 9608
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.586782489396017e-06,
+      "loss": 0.2827,
+      "step": 9609
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5852821599679747e-06,
+      "loss": 0.3037,
+      "step": 9610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5837824791048062e-06,
+      "loss": 0.2631,
+      "step": 9611
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5822834469220982e-06,
+      "loss": 0.2571,
+      "step": 9612
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5807850635353906e-06,
+      "loss": 0.2388,
+      "step": 9613
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5792873290601662e-06,
+      "loss": 0.2974,
+      "step": 9614
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5777902436118708e-06,
+      "loss": 0.3185,
+      "step": 9615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5762938073058853e-06,
+      "loss": 0.2598,
+      "step": 9616
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5747980202575475e-06,
+      "loss": 0.295,
+      "step": 9617
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.573302882582154e-06,
+      "loss": 0.2463,
+      "step": 9618
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5718083943949337e-06,
+      "loss": 0.2769,
+      "step": 9619
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.57031455581108e-06,
+      "loss": 0.2896,
+      "step": 9620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5688213669457243e-06,
+      "loss": 0.3014,
+      "step": 9621
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5673288279139586e-06,
+      "loss": 0.2529,
+      "step": 9622
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5658369388308238e-06,
+      "loss": 0.2572,
+      "step": 9623
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5643456998112971e-06,
+      "loss": 0.2864,
+      "step": 9624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5628551109703282e-06,
+      "loss": 0.283,
+      "step": 9625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.561365172422795e-06,
+      "loss": 0.2617,
+      "step": 9626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5598758842835382e-06,
+      "loss": 0.2762,
+      "step": 9627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5583872466673433e-06,
+      "loss": 0.2802,
+      "step": 9628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5568992596889487e-06,
+      "loss": 0.2702,
+      "step": 9629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5554119234630438e-06,
+      "loss": 0.2714,
+      "step": 9630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.553925238104257e-06,
+      "loss": 0.3066,
+      "step": 9631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5524392037271828e-06,
+      "loss": 0.2478,
+      "step": 9632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5509538204463536e-06,
+      "loss": 0.2829,
+      "step": 9633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5494690883762553e-06,
+      "loss": 0.2643,
+      "step": 9634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5479850076313241e-06,
+      "loss": 0.2995,
+      "step": 9635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5465015783259463e-06,
+      "loss": 0.2699,
+      "step": 9636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5450188005744593e-06,
+      "loss": 0.2903,
+      "step": 9637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5435366744911406e-06,
+      "loss": 0.2469,
+      "step": 9638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5420552001902355e-06,
+      "loss": 0.2745,
+      "step": 9639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5405743777859206e-06,
+      "loss": 0.2694,
+      "step": 9640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5390942073923343e-06,
+      "loss": 0.2482,
+      "step": 9641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.53761468912356e-06,
+      "loss": 0.2903,
+      "step": 9642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5361358230936308e-06,
+      "loss": 0.2616,
+      "step": 9643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5346576094165343e-06,
+      "loss": 0.2466,
+      "step": 9644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5331800482061954e-06,
+      "loss": 0.2435,
+      "step": 9645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5317031395765081e-06,
+      "loss": 0.2365,
+      "step": 9646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.530226883641297e-06,
+      "loss": 0.2845,
+      "step": 9647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5287512805143467e-06,
+      "loss": 0.2715,
+      "step": 9648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5272763303093907e-06,
+      "loss": 0.2733,
+      "step": 9649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5258020331401102e-06,
+      "loss": 0.2758,
+      "step": 9650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5243283891201388e-06,
+      "loss": 0.2772,
+      "step": 9651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.522855398363051e-06,
+      "loss": 0.2529,
+      "step": 9652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5213830609823877e-06,
+      "loss": 0.2728,
+      "step": 9653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5199113770916207e-06,
+      "loss": 0.3022,
+      "step": 9654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.518440346804182e-06,
+      "loss": 0.2502,
+      "step": 9655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5169699702334562e-06,
+      "loss": 0.2434,
+      "step": 9656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5155002474927683e-06,
+      "loss": 0.2771,
+      "step": 9657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5140311786953986e-06,
+      "loss": 0.2565,
+      "step": 9658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5125627639545725e-06,
+      "loss": 0.2664,
+      "step": 9659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5110950033834726e-06,
+      "loss": 0.2642,
+      "step": 9660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5096278970952272e-06,
+      "loss": 0.2405,
+      "step": 9661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.508161445202906e-06,
+      "loss": 0.2526,
+      "step": 9662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.506695647819546e-06,
+      "loss": 0.2799,
+      "step": 9663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5052305050581173e-06,
+      "loss": 0.2782,
+      "step": 9664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.503766017031547e-06,
+      "loss": 0.2469,
+      "step": 9665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5023021838527108e-06,
+      "loss": 0.2711,
+      "step": 9666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5008390056344347e-06,
+      "loss": 0.2717,
+      "step": 9667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.499376482489494e-06,
+      "loss": 0.2559,
+      "step": 9668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4979146145306068e-06,
+      "loss": 0.2724,
+      "step": 9669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4964534018704558e-06,
+      "loss": 0.2732,
+      "step": 9670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4949928446216567e-06,
+      "loss": 0.29,
+      "step": 9671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.493532942896785e-06,
+      "loss": 0.5996,
+      "step": 9672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4920736968083616e-06,
+      "loss": 0.2831,
+      "step": 9673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4906151064688602e-06,
+      "loss": 0.2847,
+      "step": 9674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4891571719907016e-06,
+      "loss": 0.2447,
+      "step": 9675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4876998934862497e-06,
+      "loss": 0.2653,
+      "step": 9676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4862432710678355e-06,
+      "loss": 0.2581,
+      "step": 9677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4847873048477191e-06,
+      "loss": 0.275,
+      "step": 9678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4833319949381232e-06,
+      "loss": 0.2882,
+      "step": 9679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4818773414512134e-06,
+      "loss": 0.2963,
+      "step": 9680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4804233444991102e-06,
+      "loss": 0.309,
+      "step": 9681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4789700041938816e-06,
+      "loss": 0.2344,
+      "step": 9682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4775173206475357e-06,
+      "loss": 0.269,
+      "step": 9683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4760652939720488e-06,
+      "loss": 0.3092,
+      "step": 9684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.474613924279329e-06,
+      "loss": 0.2766,
+      "step": 9685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4731632116812434e-06,
+      "loss": 0.2602,
+      "step": 9686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4717131562896047e-06,
+      "loss": 0.2472,
+      "step": 9687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4702637582161761e-06,
+      "loss": 0.2795,
+      "step": 9688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4688150175726724e-06,
+      "loss": 0.2886,
+      "step": 9689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4673669344707498e-06,
+      "loss": 0.2604,
+      "step": 9690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4659195090220258e-06,
+      "loss": 0.2692,
+      "step": 9691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4644727413380566e-06,
+      "loss": 0.2769,
+      "step": 9692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.463026631530351e-06,
+      "loss": 0.2602,
+      "step": 9693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4615811797103751e-06,
+      "loss": 0.5746,
+      "step": 9694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4601363859895301e-06,
+      "loss": 0.2729,
+      "step": 9695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4586922504791767e-06,
+      "loss": 0.2681,
+      "step": 9696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.457248773290617e-06,
+      "loss": 0.2784,
+      "step": 9697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4558059545351144e-06,
+      "loss": 0.5823,
+      "step": 9698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.454363794323872e-06,
+      "loss": 0.253,
+      "step": 9699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4529222927680375e-06,
+      "loss": 0.2785,
+      "step": 9700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4514814499787266e-06,
+      "loss": 0.3043,
+      "step": 9701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4500412660669828e-06,
+      "loss": 0.3049,
+      "step": 9702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4486017411438114e-06,
+      "loss": 0.2595,
+      "step": 9703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.447162875320165e-06,
+      "loss": 0.2455,
+      "step": 9704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4457246687069427e-06,
+      "loss": 0.3316,
+      "step": 9705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.444287121414998e-06,
+      "loss": 0.2393,
+      "step": 9706
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.442850233555122e-06,
+      "loss": 0.2883,
+      "step": 9707
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4414140052380721e-06,
+      "loss": 0.3138,
+      "step": 9708
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4399784365745396e-06,
+      "loss": 0.3044,
+      "step": 9709
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4385435276751724e-06,
+      "loss": 0.3032,
+      "step": 9710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.437109278650567e-06,
+      "loss": 0.3209,
+      "step": 9711
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4356756896112678e-06,
+      "loss": 0.2537,
+      "step": 9712
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4342427606677712e-06,
+      "loss": 0.2808,
+      "step": 9713
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.432810491930514e-06,
+      "loss": 0.272,
+      "step": 9714
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4313788835098964e-06,
+      "loss": 0.2477,
+      "step": 9715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4299479355162526e-06,
+      "loss": 0.2949,
+      "step": 9716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4285176480598772e-06,
+      "loss": 0.2582,
+      "step": 9717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4270880212510086e-06,
+      "loss": 0.3104,
+      "step": 9718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.425659055199835e-06,
+      "loss": 0.2592,
+      "step": 9719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4242307500164964e-06,
+      "loss": 0.2753,
+      "step": 9720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4228031058110725e-06,
+      "loss": 0.2289,
+      "step": 9721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4213761226936095e-06,
+      "loss": 0.2394,
+      "step": 9722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4199498007740841e-06,
+      "loss": 0.2959,
+      "step": 9723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4185241401624327e-06,
+      "loss": 0.2681,
+      "step": 9724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4170991409685386e-06,
+      "loss": 0.2711,
+      "step": 9725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4156748033022328e-06,
+      "loss": 0.2444,
+      "step": 9726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4142511272732994e-06,
+      "loss": 0.2516,
+      "step": 9727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4128281129914611e-06,
+      "loss": 0.3023,
+      "step": 9728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4114057605664066e-06,
+      "loss": 0.2648,
+      "step": 9729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.409984070107755e-06,
+      "loss": 0.2441,
+      "step": 9730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4085630417250873e-06,
+      "loss": 0.2982,
+      "step": 9731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4071426755279293e-06,
+      "loss": 0.29,
+      "step": 9732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4057229716257548e-06,
+      "loss": 0.2644,
+      "step": 9733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4043039301279904e-06,
+      "loss": 0.2658,
+      "step": 9734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.402885551144002e-06,
+      "loss": 0.2925,
+      "step": 9735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4014678347831178e-06,
+      "loss": 0.3234,
+      "step": 9736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4000507811546094e-06,
+      "loss": 0.2839,
+      "step": 9737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.398634390367688e-06,
+      "loss": 0.2759,
+      "step": 9738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.397218662531532e-06,
+      "loss": 0.2896,
+      "step": 9739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3958035977552509e-06,
+      "loss": 0.2549,
+      "step": 9740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.394389196147915e-06,
+      "loss": 0.2703,
+      "step": 9741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3929754578185373e-06,
+      "loss": 0.29,
+      "step": 9742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3915623828760837e-06,
+      "loss": 0.2773,
+      "step": 9743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3901499714294675e-06,
+      "loss": 0.2894,
+      "step": 9744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3887382235875446e-06,
+      "loss": 0.2537,
+      "step": 9745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3873271394591348e-06,
+      "loss": 0.2374,
+      "step": 9746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.38591671915299e-06,
+      "loss": 0.2666,
+      "step": 9747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3845069627778218e-06,
+      "loss": 0.3292,
+      "step": 9748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.383097870442286e-06,
+      "loss": 0.2768,
+      "step": 9749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3816894422549888e-06,
+      "loss": 0.2569,
+      "step": 9750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3802816783244877e-06,
+      "loss": 0.246,
+      "step": 9751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3788745787592784e-06,
+      "loss": 0.2312,
+      "step": 9752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.377468143667824e-06,
+      "loss": 0.2919,
+      "step": 9753
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3760623731585165e-06,
+      "loss": 0.2836,
+      "step": 9754
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3746572673397096e-06,
+      "loss": 0.2731,
+      "step": 9755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.373252826319701e-06,
+      "loss": 0.2968,
+      "step": 9756
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3718490502067393e-06,
+      "loss": 0.2946,
+      "step": 9757
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.370445939109022e-06,
+      "loss": 0.3139,
+      "step": 9758
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3690434931346874e-06,
+      "loss": 0.2404,
+      "step": 9759
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3676417123918374e-06,
+      "loss": 0.3181,
+      "step": 9760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3662405969885084e-06,
+      "loss": 0.2812,
+      "step": 9761
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3648401470326932e-06,
+      "loss": 0.2633,
+      "step": 9762
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3634403626323334e-06,
+      "loss": 0.2422,
+      "step": 9763
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3620412438953145e-06,
+      "loss": 0.2888,
+      "step": 9764
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3606427909294784e-06,
+      "loss": 0.2874,
+      "step": 9765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359245003842602e-06,
+      "loss": 0.3331,
+      "step": 9766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.35784788274243e-06,
+      "loss": 0.3184,
+      "step": 9767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3564514277366403e-06,
+      "loss": 0.2883,
+      "step": 9768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.355055638932864e-06,
+      "loss": 0.2527,
+      "step": 9769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.353660516438684e-06,
+      "loss": 0.2435,
+      "step": 9770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.352266060361629e-06,
+      "loss": 0.2354,
+      "step": 9771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.350872270809177e-06,
+      "loss": 0.2676,
+      "step": 9772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3494791478887504e-06,
+      "loss": 0.2786,
+      "step": 9773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3480866917077294e-06,
+      "loss": 0.2835,
+      "step": 9774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3466949023734387e-06,
+      "loss": 0.298,
+      "step": 9775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3453037799931435e-06,
+      "loss": 0.2686,
+      "step": 9776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.343913324674072e-06,
+      "loss": 0.2469,
+      "step": 9777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3425235365233892e-06,
+      "loss": 0.2936,
+      "step": 9778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3411344156482142e-06,
+      "loss": 0.3016,
+      "step": 9779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.2621,
+      "step": 9780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3383581761526022e-06,
+      "loss": 0.2589,
+      "step": 9781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.336971057746147e-06,
+      "loss": 0.311,
+      "step": 9782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3355846070431533e-06,
+      "loss": 0.2666,
+      "step": 9783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.33419882415049e-06,
+      "loss": 0.3063,
+      "step": 9784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3328137091749594e-06,
+      "loss": 0.2723,
+      "step": 9785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3314292622233227e-06,
+      "loss": 0.2581,
+      "step": 9786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3300454834022857e-06,
+      "loss": 0.3113,
+      "step": 9787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3286623728185044e-06,
+      "loss": 0.3043,
+      "step": 9788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3272799305785822e-06,
+      "loss": 0.2677,
+      "step": 9789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.325898156789066e-06,
+      "loss": 0.2315,
+      "step": 9790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.324517051556463e-06,
+      "loss": 0.601,
+      "step": 9791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3231366149872183e-06,
+      "loss": 0.3023,
+      "step": 9792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3217568471877284e-06,
+      "loss": 0.2868,
+      "step": 9793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.320377748264341e-06,
+      "loss": 0.2831,
+      "step": 9794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3189993183233496e-06,
+      "loss": 0.2608,
+      "step": 9795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3176215574709982e-06,
+      "loss": 0.2668,
+      "step": 9796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3162444658134731e-06,
+      "loss": 0.2589,
+      "step": 9797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3148680434569206e-06,
+      "loss": 0.2736,
+      "step": 9798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.313492290507422e-06,
+      "loss": 0.291,
+      "step": 9799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3121172070710165e-06,
+      "loss": 0.2738,
+      "step": 9800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3107427932536886e-06,
+      "loss": 0.2787,
+      "step": 9801
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.309369049161372e-06,
+      "loss": 0.2728,
+      "step": 9802
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3079959748999494e-06,
+      "loss": 0.251,
+      "step": 9803
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3066235705752439e-06,
+      "loss": 0.2325,
+      "step": 9804
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3052518362930433e-06,
+      "loss": 0.2521,
+      "step": 9805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3038807721590663e-06,
+      "loss": 0.2484,
+      "step": 9806
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3025103782789906e-06,
+      "loss": 0.2568,
+      "step": 9807
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3011406547584392e-06,
+      "loss": 0.2615,
+      "step": 9808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2997716017029849e-06,
+      "loss": 0.2695,
+      "step": 9809
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2984032192181473e-06,
+      "loss": 0.2481,
+      "step": 9810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2970355074093898e-06,
+      "loss": 0.2468,
+      "step": 9811
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2956684663821363e-06,
+      "loss": 0.2677,
+      "step": 9812
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2943020962417485e-06,
+      "loss": 0.322,
+      "step": 9813
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2929363970935371e-06,
+      "loss": 0.2497,
+      "step": 9814
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2915713690427655e-06,
+      "loss": 0.3096,
+      "step": 9815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2902070121946441e-06,
+      "loss": 0.2708,
+      "step": 9816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2888433266543288e-06,
+      "loss": 0.3307,
+      "step": 9817
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2874803125269274e-06,
+      "loss": 0.2576,
+      "step": 9818
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.286117969917493e-06,
+      "loss": 0.2819,
+      "step": 9819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2847562989310313e-06,
+      "loss": 0.2889,
+      "step": 9820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2833952996724864e-06,
+      "loss": 0.2562,
+      "step": 9821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2820349722467663e-06,
+      "loss": 0.294,
+      "step": 9822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2806753167587117e-06,
+      "loss": 0.3049,
+      "step": 9823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2793163333131208e-06,
+      "loss": 0.289,
+      "step": 9824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.277958022014736e-06,
+      "loss": 0.2167,
+      "step": 9825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2766003829682504e-06,
+      "loss": 0.2513,
+      "step": 9826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2752434162783056e-06,
+      "loss": 0.2759,
+      "step": 9827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.273887122049483e-06,
+      "loss": 0.5486,
+      "step": 9828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2725315003863292e-06,
+      "loss": 0.2589,
+      "step": 9829
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2711765513933216e-06,
+      "loss": 0.2759,
+      "step": 9830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2698222751748946e-06,
+      "loss": 0.2731,
+      "step": 9831
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.26846867183543e-06,
+      "loss": 0.2802,
+      "step": 9832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2671157414792567e-06,
+      "loss": 0.3047,
+      "step": 9833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2657634842106526e-06,
+      "loss": 0.2815,
+      "step": 9834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2644119001338385e-06,
+      "loss": 0.2588,
+      "step": 9835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2630609893529956e-06,
+      "loss": 0.306,
+      "step": 9836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2617107519722393e-06,
+      "loss": 0.3174,
+      "step": 9837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.26036118809564e-06,
+      "loss": 0.3262,
+      "step": 9838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2590122978272178e-06,
+      "loss": 0.2402,
+      "step": 9839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2576640812709363e-06,
+      "loss": 0.6742,
+      "step": 9840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.256316538530713e-06,
+      "loss": 0.3113,
+      "step": 9841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.254969669710402e-06,
+      "loss": 0.3033,
+      "step": 9842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2536234749138232e-06,
+      "loss": 0.248,
+      "step": 9843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2522779542447272e-06,
+      "loss": 0.2856,
+      "step": 9844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2509331078068231e-06,
+      "loss": 0.2316,
+      "step": 9845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.249588935703765e-06,
+      "loss": 0.2607,
+      "step": 9846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2482454380391552e-06,
+      "loss": 0.2594,
+      "step": 9847
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.246902614916544e-06,
+      "loss": 0.2625,
+      "step": 9848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.245560466439425e-06,
+      "loss": 0.2535,
+      "step": 9849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2442189927112514e-06,
+      "loss": 0.588,
+      "step": 9850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.242878193835415e-06,
+      "loss": 0.2385,
+      "step": 9851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2415380699152568e-06,
+      "loss": 0.2433,
+      "step": 9852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.240198621054066e-06,
+      "loss": 0.2903,
+      "step": 9853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2388598473550828e-06,
+      "loss": 0.3208,
+      "step": 9854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.237521748921492e-06,
+      "loss": 0.2714,
+      "step": 9855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2361843258564277e-06,
+      "loss": 0.2559,
+      "step": 9856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2348475782629733e-06,
+      "loss": 0.2958,
+      "step": 9857
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2335115062441593e-06,
+      "loss": 0.2708,
+      "step": 9858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2321761099029571e-06,
+      "loss": 0.2709,
+      "step": 9859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2308413893423021e-06,
+      "loss": 0.2973,
+      "step": 9860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.229507344665062e-06,
+      "loss": 0.603,
+      "step": 9861
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2281739759740575e-06,
+      "loss": 0.239,
+      "step": 9862
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2268412833720611e-06,
+      "loss": 0.2968,
+      "step": 9863
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2255092669617897e-06,
+      "loss": 0.2596,
+      "step": 9864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2241779268459098e-06,
+      "loss": 0.2218,
+      "step": 9865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2228472631270272e-06,
+      "loss": 0.2832,
+      "step": 9866
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2215172759077143e-06,
+      "loss": 0.2881,
+      "step": 9867
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2201879652904714e-06,
+      "loss": 0.2374,
+      "step": 9868
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2188593313777575e-06,
+      "loss": 0.2803,
+      "step": 9869
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2175313742719775e-06,
+      "loss": 0.2687,
+      "step": 9870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2162040940754826e-06,
+      "loss": 0.2388,
+      "step": 9871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2148774908905782e-06,
+      "loss": 0.2557,
+      "step": 9872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2135515648195029e-06,
+      "loss": 0.2516,
+      "step": 9873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.212226315964462e-06,
+      "loss": 0.2314,
+      "step": 9874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.210901744427594e-06,
+      "loss": 0.2977,
+      "step": 9875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.20957785031099e-06,
+      "loss": 0.3063,
+      "step": 9876
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.208254633716691e-06,
+      "loss": 0.2479,
+      "step": 9877
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2069320947466845e-06,
+      "loss": 0.2745,
+      "step": 9878
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2056102335029052e-06,
+      "loss": 0.2602,
+      "step": 9879
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2042890500872306e-06,
+      "loss": 0.2932,
+      "step": 9880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2029685446015005e-06,
+      "loss": 0.3196,
+      "step": 9881
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2016487171474844e-06,
+      "loss": 0.2668,
+      "step": 9882
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2003295678269112e-06,
+      "loss": 0.2781,
+      "step": 9883
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1990110967414548e-06,
+      "loss": 0.3168,
+      "step": 9884
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1976933039927363e-06,
+      "loss": 0.5594,
+      "step": 9885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1963761896823255e-06,
+      "loss": 0.2618,
+      "step": 9886
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1950597539117348e-06,
+      "loss": 0.2421,
+      "step": 9887
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1937439967824338e-06,
+      "loss": 0.2948,
+      "step": 9888
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1924289183958349e-06,
+      "loss": 0.2701,
+      "step": 9889
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1911145188532936e-06,
+      "loss": 0.2894,
+      "step": 9890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1898007982561177e-06,
+      "loss": 0.2553,
+      "step": 9891
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1884877567055653e-06,
+      "loss": 0.2543,
+      "step": 9892
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1871753943028375e-06,
+      "loss": 0.2953,
+      "step": 9893
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1858637111490845e-06,
+      "loss": 0.3027,
+      "step": 9894
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1845527073454045e-06,
+      "loss": 0.2673,
+      "step": 9895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1832423829928452e-06,
+      "loss": 0.2372,
+      "step": 9896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1819327381923972e-06,
+      "loss": 0.2414,
+      "step": 9897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1806237730450009e-06,
+      "loss": 0.2679,
+      "step": 9898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1793154876515477e-06,
+      "loss": 0.2697,
+      "step": 9899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1780078821128716e-06,
+      "loss": 0.2605,
+      "step": 9900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1767009565297583e-06,
+      "loss": 0.3094,
+      "step": 9901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1753947110029373e-06,
+      "loss": 0.2665,
+      "step": 9902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1740891456330894e-06,
+      "loss": 0.2715,
+      "step": 9903
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1727842605208373e-06,
+      "loss": 0.2929,
+      "step": 9904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1714800557667616e-06,
+      "loss": 0.2645,
+      "step": 9905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1701765314713786e-06,
+      "loss": 0.2672,
+      "step": 9906
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.168873687735158e-06,
+      "loss": 0.2319,
+      "step": 9907
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1675715246585184e-06,
+      "loss": 0.249,
+      "step": 9908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1662700423418239e-06,
+      "loss": 0.2432,
+      "step": 9909
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1649692408853875e-06,
+      "loss": 0.2649,
+      "step": 9910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1636691203894623e-06,
+      "loss": 0.2471,
+      "step": 9911
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.162369680954264e-06,
+      "loss": 0.3283,
+      "step": 9912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.16107092267994e-06,
+      "loss": 0.3055,
+      "step": 9913
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1597728456665958e-06,
+      "loss": 0.2349,
+      "step": 9914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1584754500142792e-06,
+      "loss": 0.2845,
+      "step": 9915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.157178735822988e-06,
+      "loss": 0.2772,
+      "step": 9916
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1558827031926679e-06,
+      "loss": 0.5519,
+      "step": 9917
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1545873522232055e-06,
+      "loss": 0.3152,
+      "step": 9918
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1532926830144475e-06,
+      "loss": 0.2501,
+      "step": 9919
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1519986956661744e-06,
+      "loss": 0.5618,
+      "step": 9920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1507053902781217e-06,
+      "loss": 0.2558,
+      "step": 9921
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1494127669499732e-06,
+      "loss": 0.2357,
+      "step": 9922
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1481208257813558e-06,
+      "loss": 0.2667,
+      "step": 9923
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1468295668718498e-06,
+      "loss": 0.231,
+      "step": 9924
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1455389903209713e-06,
+      "loss": 0.2867,
+      "step": 9925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1442490962281983e-06,
+      "loss": 0.3174,
+      "step": 9926
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.142959884692949e-06,
+      "loss": 0.2521,
+      "step": 9927
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1416713558145854e-06,
+      "loss": 0.283,
+      "step": 9928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1403835096924244e-06,
+      "loss": 0.2653,
+      "step": 9929
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1390963464257254e-06,
+      "loss": 0.2654,
+      "step": 9930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1378098661136993e-06,
+      "loss": 0.2322,
+      "step": 9931
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1365240688554958e-06,
+      "loss": 0.3051,
+      "step": 9932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.135238954750223e-06,
+      "loss": 0.2597,
+      "step": 9933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1339545238969308e-06,
+      "loss": 0.2883,
+      "step": 9934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1326707763946143e-06,
+      "loss": 0.2836,
+      "step": 9935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.131387712342219e-06,
+      "loss": 0.3225,
+      "step": 9936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.130105331838638e-06,
+      "loss": 0.3378,
+      "step": 9937
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1288236349827108e-06,
+      "loss": 0.6455,
+      "step": 9938
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1275426218732233e-06,
+      "loss": 0.2538,
+      "step": 9939
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1262622926089118e-06,
+      "loss": 0.2444,
+      "step": 9940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1249826472884574e-06,
+      "loss": 0.2537,
+      "step": 9941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1237036860104833e-06,
+      "loss": 0.259,
+      "step": 9942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1224254088735752e-06,
+      "loss": 0.2936,
+      "step": 9943
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.121147815976248e-06,
+      "loss": 0.2749,
+      "step": 9944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1198709074169766e-06,
+      "loss": 0.2701,
+      "step": 9945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1185946832941774e-06,
+      "loss": 0.3221,
+      "step": 9946
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1173191437062147e-06,
+      "loss": 0.2545,
+      "step": 9947
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1160442887514045e-06,
+      "loss": 0.2734,
+      "step": 9948
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1147701185280002e-06,
+      "loss": 0.3171,
+      "step": 9949
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1134966331342157e-06,
+      "loss": 0.2708,
+      "step": 9950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1122238326681978e-06,
+      "loss": 0.2604,
+      "step": 9951
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1109517172280525e-06,
+      "loss": 0.2535,
+      "step": 9952
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.109680286911826e-06,
+      "loss": 0.2397,
+      "step": 9953
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1084095418175156e-06,
+      "loss": 0.2753,
+      "step": 9954
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1071394820430647e-06,
+      "loss": 0.2865,
+      "step": 9955
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1058701076863575e-06,
+      "loss": 0.2385,
+      "step": 9956
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1046014188452392e-06,
+      "loss": 0.2606,
+      "step": 9957
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.103333415617488e-06,
+      "loss": 0.2532,
+      "step": 9958
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.102066098100838e-06,
+      "loss": 0.2868,
+      "step": 9959
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1007994663929656e-06,
+      "loss": 0.3349,
+      "step": 9960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0995335205914993e-06,
+      "loss": 0.2753,
+      "step": 9961
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0982682607940131e-06,
+      "loss": 0.3004,
+      "step": 9962
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0970036870980195e-06,
+      "loss": 0.2977,
+      "step": 9963
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0957397996009934e-06,
+      "loss": 0.3046,
+      "step": 9964
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.094476598400348e-06,
+      "loss": 0.2841,
+      "step": 9965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0932140835934414e-06,
+      "loss": 0.2584,
+      "step": 9966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0919522552775829e-06,
+      "loss": 0.3019,
+      "step": 9967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0906911135500298e-06,
+      "loss": 0.2882,
+      "step": 9968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0894306585079838e-06,
+      "loss": 0.2058,
+      "step": 9969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.088170890248591e-06,
+      "loss": 0.2933,
+      "step": 9970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0869118088689535e-06,
+      "loss": 0.3075,
+      "step": 9971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0856534144661146e-06,
+      "loss": 0.2606,
+      "step": 9972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0843957071370626e-06,
+      "loss": 0.2866,
+      "step": 9973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0831386869787353e-06,
+      "loss": 0.2972,
+      "step": 9974
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0818823540880174e-06,
+      "loss": 0.277,
+      "step": 9975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.080626708561744e-06,
+      "loss": 0.2919,
+      "step": 9976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0793717504966906e-06,
+      "loss": 0.2162,
+      "step": 9977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0781174799895844e-06,
+      "loss": 0.3007,
+      "step": 9978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0768638971371014e-06,
+      "loss": 0.2684,
+      "step": 9979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0756110020358568e-06,
+      "loss": 0.2422,
+      "step": 9980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0743587947824186e-06,
+      "loss": 0.3024,
+      "step": 9981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0731072754733019e-06,
+      "loss": 0.2855,
+      "step": 9982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0718564442049672e-06,
+      "loss": 0.2949,
+      "step": 9983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0706063010738232e-06,
+      "loss": 0.2958,
+      "step": 9984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0693568461762238e-06,
+      "loss": 0.2711,
+      "step": 9985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.068108079608473e-06,
+      "loss": 0.2485,
+      "step": 9986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.066860001466813e-06,
+      "loss": 0.239,
+      "step": 9987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0656126118474485e-06,
+      "loss": 0.2626,
+      "step": 9988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0643659108465166e-06,
+      "loss": 0.2715,
+      "step": 9989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0631198985601077e-06,
+      "loss": 0.2724,
+      "step": 9990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0618745750842585e-06,
+      "loss": 0.2841,
+      "step": 9991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0606299405149522e-06,
+      "loss": 0.5852,
+      "step": 9992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.059385994948121e-06,
+      "loss": 0.301,
+      "step": 9993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0581427384796372e-06,
+      "loss": 0.2637,
+      "step": 9994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0569001712053317e-06,
+      "loss": 0.2565,
+      "step": 9995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0556582932209703e-06,
+      "loss": 0.2871,
+      "step": 9996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0544171046222717e-06,
+      "loss": 0.2781,
+      "step": 9997
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.053176605504902e-06,
+      "loss": 0.6038,
+      "step": 9998
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.051936795964471e-06,
+      "loss": 0.2552,
+      "step": 9999
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0506976760965414e-06,
+      "loss": 0.2501,
+      "step": 10000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0494592459966102e-06,
+      "loss": 0.2786,
+      "step": 10001
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0482215057601364e-06,
+      "loss": 0.2624,
+      "step": 10002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0469844554825192e-06,
+      "loss": 0.2753,
+      "step": 10003
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0457480952591005e-06,
+      "loss": 0.2574,
+      "step": 10004
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0445124251851735e-06,
+      "loss": 0.2845,
+      "step": 10005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.043277445355978e-06,
+      "loss": 0.2184,
+      "step": 10006
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0420431558667033e-06,
+      "loss": 0.2621,
+      "step": 10007
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0408095568124765e-06,
+      "loss": 0.2249,
+      "step": 10008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0395766482883806e-06,
+      "loss": 0.2835,
+      "step": 10009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0383444303894453e-06,
+      "loss": 0.2427,
+      "step": 10010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0371129032106375e-06,
+      "loss": 0.3246,
+      "step": 10011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0358820668468805e-06,
+      "loss": 0.279,
+      "step": 10012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0346519213930417e-06,
+      "loss": 0.2651,
+      "step": 10013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.033422466943933e-06,
+      "loss": 0.3018,
+      "step": 10014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0321937035943153e-06,
+      "loss": 0.2598,
+      "step": 10015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0309656314388949e-06,
+      "loss": 0.2408,
+      "step": 10016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0297382505723297e-06,
+      "loss": 0.5775,
+      "step": 10017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0285115610892138e-06,
+      "loss": 0.2997,
+      "step": 10018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0272855630840982e-06,
+      "loss": 0.2479,
+      "step": 10019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0260602566514755e-06,
+      "loss": 0.2786,
+      "step": 10020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0248356418857863e-06,
+      "loss": 0.2689,
+      "step": 10021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0236117188814187e-06,
+      "loss": 0.2477,
+      "step": 10022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0223884877327062e-06,
+      "loss": 0.2961,
+      "step": 10023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0211659485339308e-06,
+      "loss": 0.2662,
+      "step": 10024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0199441013793155e-06,
+      "loss": 0.2689,
+      "step": 10025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.01872294636304e-06,
+      "loss": 0.2758,
+      "step": 10026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0175024835792202e-06,
+      "loss": 0.2993,
+      "step": 10027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0162827131219255e-06,
+      "loss": 0.2344,
+      "step": 10028
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0150636350851695e-06,
+      "loss": 0.2725,
+      "step": 10029
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0138452495629125e-06,
+      "loss": 0.262,
+      "step": 10030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0126275566490628e-06,
+      "loss": 0.2758,
+      "step": 10031
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.011410556437471e-06,
+      "loss": 0.2301,
+      "step": 10032
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0101942490219418e-06,
+      "loss": 0.2752,
+      "step": 10033
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0089786344962194e-06,
+      "loss": 0.2966,
+      "step": 10034
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0077637129539963e-06,
+      "loss": 0.2946,
+      "step": 10035
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0065494844889156e-06,
+      "loss": 0.2461,
+      "step": 10036
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0053359491945624e-06,
+      "loss": 0.2614,
+      "step": 10037
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.004123107164472e-06,
+      "loss": 0.2887,
+      "step": 10038
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0029109584921193e-06,
+      "loss": 0.2658,
+      "step": 10039
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0016995032709354e-06,
+      "loss": 0.2348,
+      "step": 10040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0004887415942943e-06,
+      "loss": 0.2816,
+      "step": 10041
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.992786735555104e-07,
+      "loss": 0.3228,
+      "step": 10042
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.980692992478524e-07,
+      "loss": 0.287,
+      "step": 10043
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.968606187645336e-07,
+      "loss": 0.2502,
+      "step": 10044
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.956526321987147e-07,
+      "loss": 0.2922,
+      "step": 10045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.94445339643495e-07,
+      "loss": 0.2318,
+      "step": 10046
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.93238741191932e-07,
+      "loss": 0.3,
+      "step": 10047
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.920328369370258e-07,
+      "loss": 0.2662,
+      "step": 10048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.908276269717166e-07,
+      "loss": 0.2729,
+      "step": 10049
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.896231113888988e-07,
+      "loss": 0.2654,
+      "step": 10050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.884192902814094e-07,
+      "loss": 0.5657,
+      "step": 10051
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.87216163742033e-07,
+      "loss": 0.2981,
+      "step": 10052
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.860137318635021e-07,
+      "loss": 0.3122,
+      "step": 10053
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.848119947384937e-07,
+      "loss": 0.2991,
+      "step": 10054
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.836109524596326e-07,
+      "loss": 0.2426,
+      "step": 10055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.82410605119486e-07,
+      "loss": 0.314,
+      "step": 10056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.812109528105728e-07,
+      "loss": 0.265,
+      "step": 10057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.800119956253574e-07,
+      "loss": 0.5778,
+      "step": 10058
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.788137336562464e-07,
+      "loss": 0.2259,
+      "step": 10059
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.776161669955997e-07,
+      "loss": 0.3255,
+      "step": 10060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.764192957357166e-07,
+      "loss": 0.2555,
+      "step": 10061
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.7522311996885e-07,
+      "loss": 0.2515,
+      "step": 10062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.740276397871906e-07,
+      "loss": 0.2311,
+      "step": 10063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.728328552828814e-07,
+      "loss": 0.2904,
+      "step": 10064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.71638766548012e-07,
+      "loss": 0.2338,
+      "step": 10065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.704453736746156e-07,
+      "loss": 0.2412,
+      "step": 10066
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.692526767546727e-07,
+      "loss": 0.305,
+      "step": 10067
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.680606758801126e-07,
+      "loss": 0.2756,
+      "step": 10068
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.66869371142809e-07,
+      "loss": 0.2956,
+      "step": 10069
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.656787626345765e-07,
+      "loss": 0.2383,
+      "step": 10070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.644888504471883e-07,
+      "loss": 0.2682,
+      "step": 10071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.632996346723522e-07,
+      "loss": 0.2812,
+      "step": 10072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.62111115401728e-07,
+      "loss": 0.2468,
+      "step": 10073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.60923292726923e-07,
+      "loss": 0.2803,
+      "step": 10074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.597361667394867e-07,
+      "loss": 0.2731,
+      "step": 10075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.585497375309183e-07,
+      "loss": 0.2899,
+      "step": 10076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.57364005192658e-07,
+      "loss": 0.2965,
+      "step": 10077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.561789698161007e-07,
+      "loss": 0.2524,
+      "step": 10078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.54994631492584e-07,
+      "loss": 0.2448,
+      "step": 10079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.538109903133862e-07,
+      "loss": 0.2676,
+      "step": 10080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.526280463697391e-07,
+      "loss": 0.2374,
+      "step": 10081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.514457997528171e-07,
+      "loss": 0.2627,
+      "step": 10082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.502642505537452e-07,
+      "loss": 0.2521,
+      "step": 10083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.49083398863585e-07,
+      "loss": 0.2778,
+      "step": 10084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.479032447733561e-07,
+      "loss": 0.257,
+      "step": 10085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.467237883740199e-07,
+      "loss": 0.2679,
+      "step": 10086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.455450297564773e-07,
+      "loss": 0.2492,
+      "step": 10087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.443669690115853e-07,
+      "loss": 0.2881,
+      "step": 10088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.431896062301427e-07,
+      "loss": 0.2505,
+      "step": 10089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.420129415028934e-07,
+      "loss": 0.2717,
+      "step": 10090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.408369749205303e-07,
+      "loss": 0.2823,
+      "step": 10091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.396617065736902e-07,
+      "loss": 0.2504,
+      "step": 10092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.384871365529591e-07,
+      "loss": 0.2822,
+      "step": 10093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.373132649488636e-07,
+      "loss": 0.2186,
+      "step": 10094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.361400918518814e-07,
+      "loss": 0.3019,
+      "step": 10095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.349676173524347e-07,
+      "loss": 0.2479,
+      "step": 10096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.337958415408932e-07,
+      "loss": 0.2894,
+      "step": 10097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.326247645075703e-07,
+      "loss": 0.2569,
+      "step": 10098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.314543863427272e-07,
+      "loss": 0.2856,
+      "step": 10099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.302847071365739e-07,
+      "loss": 0.2401,
+      "step": 10100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.291157269792572e-07,
+      "loss": 0.2794,
+      "step": 10101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.279474459608806e-07,
+      "loss": 0.2654,
+      "step": 10102
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.267798641714887e-07,
+      "loss": 0.6188,
+      "step": 10103
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.256129817010728e-07,
+      "loss": 0.2567,
+      "step": 10104
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.244467986395699e-07,
+      "loss": 0.2742,
+      "step": 10105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.23281315076865e-07,
+      "loss": 0.2898,
+      "step": 10106
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.221165311027879e-07,
+      "loss": 0.2709,
+      "step": 10107
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.209524468071096e-07,
+      "loss": 0.3021,
+      "step": 10108
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.197890622795603e-07,
+      "loss": 0.278,
+      "step": 10109
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.186263776098014e-07,
+      "loss": 0.2543,
+      "step": 10110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.174643928874483e-07,
+      "loss": 0.312,
+      "step": 10111
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.163031082020623e-07,
+      "loss": 0.2413,
+      "step": 10112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.151425236431489e-07,
+      "loss": 0.2113,
+      "step": 10113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.139826393001617e-07,
+      "loss": 0.2538,
+      "step": 10114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.128234552624937e-07,
+      "loss": 0.3054,
+      "step": 10115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.116649716194948e-07,
+      "loss": 0.241,
+      "step": 10116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.105071884604543e-07,
+      "loss": 0.2518,
+      "step": 10117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.093501058746057e-07,
+      "loss": 0.2545,
+      "step": 10118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.081937239511329e-07,
+      "loss": 0.2581,
+      "step": 10119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.070380427791636e-07,
+      "loss": 0.3092,
+      "step": 10120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.058830624477754e-07,
+      "loss": 0.2772,
+      "step": 10121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.047287830459806e-07,
+      "loss": 0.3005,
+      "step": 10122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.035752046627522e-07,
+      "loss": 0.2443,
+      "step": 10123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.024223273870014e-07,
+      "loss": 0.3227,
+      "step": 10124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.012701513075839e-07,
+      "loss": 0.2647,
+      "step": 10125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.001186765133052e-07,
+      "loss": 0.2624,
+      "step": 10126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.989679030929155e-07,
+      "loss": 0.2267,
+      "step": 10127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.978178311351094e-07,
+      "loss": 0.3022,
+      "step": 10128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.966684607285304e-07,
+      "loss": 0.2643,
+      "step": 10129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.955197919617653e-07,
+      "loss": 0.2537,
+      "step": 10130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.943718249233491e-07,
+      "loss": 0.2786,
+      "step": 10131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.932245597017597e-07,
+      "loss": 0.2454,
+      "step": 10132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.920779963854231e-07,
+      "loss": 0.2853,
+      "step": 10133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.909321350627109e-07,
+      "loss": 0.2805,
+      "step": 10134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.897869758219401e-07,
+      "loss": 0.286,
+      "step": 10135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.886425187513748e-07,
+      "loss": 0.2698,
+      "step": 10136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.87498763939223e-07,
+      "loss": 0.2537,
+      "step": 10137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.863557114736432e-07,
+      "loss": 0.2933,
+      "step": 10138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.852133614427306e-07,
+      "loss": 0.2758,
+      "step": 10139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.840717139345345e-07,
+      "loss": 0.2505,
+      "step": 10140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.829307690370481e-07,
+      "loss": 0.2435,
+      "step": 10141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.817905268382088e-07,
+      "loss": 0.2628,
+      "step": 10142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.806509874259017e-07,
+      "loss": 0.2493,
+      "step": 10143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.795121508879567e-07,
+      "loss": 0.2852,
+      "step": 10144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.783740173121502e-07,
+      "loss": 0.2282,
+      "step": 10145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.772365867862021e-07,
+      "loss": 0.262,
+      "step": 10146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.76099859397781e-07,
+      "loss": 0.2187,
+      "step": 10147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.749638352345002e-07,
+      "loss": 0.2646,
+      "step": 10148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.738285143839198e-07,
+      "loss": 0.256,
+      "step": 10149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.726938969335419e-07,
+      "loss": 0.287,
+      "step": 10150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.715599829708199e-07,
+      "loss": 0.2524,
+      "step": 10151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.704267725831517e-07,
+      "loss": 0.3007,
+      "step": 10152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.69294265857874e-07,
+      "loss": 0.253,
+      "step": 10153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.681624628822794e-07,
+      "loss": 0.2729,
+      "step": 10154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.670313637436012e-07,
+      "loss": 0.2668,
+      "step": 10155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.659009685290177e-07,
+      "loss": 0.3208,
+      "step": 10156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.647712773256545e-07,
+      "loss": 0.2519,
+      "step": 10157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.636422902205821e-07,
+      "loss": 0.2219,
+      "step": 10158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.625140073008209e-07,
+      "loss": 0.2719,
+      "step": 10159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.613864286533257e-07,
+      "loss": 0.2439,
+      "step": 10160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.602595543650117e-07,
+      "loss": 0.28,
+      "step": 10161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.591333845227312e-07,
+      "loss": 0.2411,
+      "step": 10162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.580079192132829e-07,
+      "loss": 0.2894,
+      "step": 10163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.568831585234116e-07,
+      "loss": 0.2717,
+      "step": 10164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.557591025398082e-07,
+      "loss": 0.2485,
+      "step": 10165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.54635751349111e-07,
+      "loss": 0.2521,
+      "step": 10166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.535131050379019e-07,
+      "loss": 0.2629,
+      "step": 10167
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.523911636927074e-07,
+      "loss": 0.2896,
+      "step": 10168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.512699274000047e-07,
+      "loss": 0.2623,
+      "step": 10169
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.501493962462092e-07,
+      "loss": 0.3315,
+      "step": 10170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.490295703176876e-07,
+      "loss": 0.2733,
+      "step": 10171
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.479104497007496e-07,
+      "loss": 0.2975,
+      "step": 10172
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.467920344816538e-07,
+      "loss": 0.2535,
+      "step": 10173
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.456743247465992e-07,
+      "loss": 0.2791,
+      "step": 10174
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.445573205817359e-07,
+      "loss": 0.2787,
+      "step": 10175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.434410220731571e-07,
+      "loss": 0.2979,
+      "step": 10176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.423254293068983e-07,
+      "loss": 0.2758,
+      "step": 10177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.412105423689465e-07,
+      "loss": 0.2809,
+      "step": 10178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.400963613452307e-07,
+      "loss": 0.2643,
+      "step": 10179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.389828863216276e-07,
+      "loss": 0.2905,
+      "step": 10180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.378701173839577e-07,
+      "loss": 0.2662,
+      "step": 10181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.367580546179877e-07,
+      "loss": 0.2819,
+      "step": 10182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.356466981094313e-07,
+      "loss": 0.2352,
+      "step": 10183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.345360479439435e-07,
+      "loss": 0.2591,
+      "step": 10184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.33426104207129e-07,
+      "loss": 0.2668,
+      "step": 10185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.323168669845383e-07,
+      "loss": 0.2708,
+      "step": 10186
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.31208336361663e-07,
+      "loss": 0.2935,
+      "step": 10187
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.301005124239459e-07,
+      "loss": 0.2203,
+      "step": 10188
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.289933952567708e-07,
+      "loss": 0.2464,
+      "step": 10189
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.278869849454718e-07,
+      "loss": 0.2709,
+      "step": 10190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.267812815753195e-07,
+      "loss": 0.236,
+      "step": 10191
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.256762852315414e-07,
+      "loss": 0.2982,
+      "step": 10192
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.245719959993049e-07,
+      "loss": 0.3003,
+      "step": 10193
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.234684139637205e-07,
+      "loss": 0.2906,
+      "step": 10194
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.223655392098484e-07,
+      "loss": 0.2953,
+      "step": 10195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.212633718226926e-07,
+      "loss": 0.2431,
+      "step": 10196
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.201619118872039e-07,
+      "loss": 0.2515,
+      "step": 10197
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.190611594882736e-07,
+      "loss": 0.2704,
+      "step": 10198
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.179611147107458e-07,
+      "loss": 0.2585,
+      "step": 10199
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.168617776394072e-07,
+      "loss": 0.2499,
+      "step": 10200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.157631483589856e-07,
+      "loss": 0.2742,
+      "step": 10201
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.146652269541599e-07,
+      "loss": 0.283,
+      "step": 10202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.135680135095525e-07,
+      "loss": 0.2634,
+      "step": 10203
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.124715081097323e-07,
+      "loss": 0.2778,
+      "step": 10204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.113757108392095e-07,
+      "loss": 0.3146,
+      "step": 10205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.102806217824455e-07,
+      "loss": 0.2761,
+      "step": 10206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.09186241023846e-07,
+      "loss": 0.3047,
+      "step": 10207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.080925686477548e-07,
+      "loss": 0.3096,
+      "step": 10208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.06999604738471e-07,
+      "loss": 0.597,
+      "step": 10209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.059073493802327e-07,
+      "loss": 0.2451,
+      "step": 10210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.048158026572272e-07,
+      "loss": 0.2445,
+      "step": 10211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.037249646535838e-07,
+      "loss": 0.2629,
+      "step": 10212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.026348354533808e-07,
+      "loss": 0.3062,
+      "step": 10213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.01545415140641e-07,
+      "loss": 0.3006,
+      "step": 10214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.004567037993282e-07,
+      "loss": 0.2589,
+      "step": 10215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.993687015133567e-07,
+      "loss": 0.2894,
+      "step": 10216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.982814083665825e-07,
+      "loss": 0.2858,
+      "step": 10217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.971948244428118e-07,
+      "loss": 0.3172,
+      "step": 10218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.961089498257912e-07,
+      "loss": 0.2963,
+      "step": 10219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.950237845992148e-07,
+      "loss": 0.2882,
+      "step": 10220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.939393288467234e-07,
+      "loss": 0.3141,
+      "step": 10221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.928555826518991e-07,
+      "loss": 0.2718,
+      "step": 10222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.917725460982717e-07,
+      "loss": 0.2129,
+      "step": 10223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.906902192693177e-07,
+      "loss": 0.273,
+      "step": 10224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.896086022484572e-07,
+      "loss": 0.2366,
+      "step": 10225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.885276951190568e-07,
+      "loss": 0.2632,
+      "step": 10226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.874474979644254e-07,
+      "loss": 0.2374,
+      "step": 10227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.863680108678218e-07,
+      "loss": 0.3135,
+      "step": 10228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.85289233912444e-07,
+      "loss": 0.2545,
+      "step": 10229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.842111671814401e-07,
+      "loss": 0.2639,
+      "step": 10230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.831338107579056e-07,
+      "loss": 0.267,
+      "step": 10231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.820571647248732e-07,
+      "loss": 0.2482,
+      "step": 10232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.809812291653285e-07,
+      "loss": 0.262,
+      "step": 10233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.799060041621975e-07,
+      "loss": 0.2926,
+      "step": 10234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.788314897983551e-07,
+      "loss": 0.2347,
+      "step": 10235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.777576861566149e-07,
+      "loss": 0.2864,
+      "step": 10236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.766845933197464e-07,
+      "loss": 0.2698,
+      "step": 10237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.756122113704567e-07,
+      "loss": 0.2802,
+      "step": 10238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.745405403913975e-07,
+      "loss": 0.3302,
+      "step": 10239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.734695804651693e-07,
+      "loss": 0.2688,
+      "step": 10240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.72399331674315e-07,
+      "loss": 0.2608,
+      "step": 10241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.713297941013264e-07,
+      "loss": 0.2874,
+      "step": 10242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.702609678286366e-07,
+      "loss": 0.2923,
+      "step": 10243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.691928529386261e-07,
+      "loss": 0.2902,
+      "step": 10244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.681254495136214e-07,
+      "loss": 0.2681,
+      "step": 10245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.670587576358889e-07,
+      "loss": 0.3006,
+      "step": 10246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.659927773876474e-07,
+      "loss": 0.2382,
+      "step": 10247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.649275088510544e-07,
+      "loss": 0.295,
+      "step": 10248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.638629521082186e-07,
+      "loss": 0.2173,
+      "step": 10249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.627991072411889e-07,
+      "loss": 0.3004,
+      "step": 10250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.617359743319608e-07,
+      "loss": 0.2554,
+      "step": 10251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.606735534624787e-07,
+      "loss": 0.2717,
+      "step": 10252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.59611844714625e-07,
+      "loss": 0.3234,
+      "step": 10253
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.585508481702308e-07,
+      "loss": 0.2795,
+      "step": 10254
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.574905639110752e-07,
+      "loss": 0.3218,
+      "step": 10255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.564309920188784e-07,
+      "loss": 0.2847,
+      "step": 10256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.553721325753061e-07,
+      "loss": 0.2843,
+      "step": 10257
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.543139856619708e-07,
+      "loss": 0.3107,
+      "step": 10258
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.532565513604306e-07,
+      "loss": 0.27,
+      "step": 10259
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.521998297521848e-07,
+      "loss": 0.2729,
+      "step": 10260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.511438209186817e-07,
+      "loss": 0.2521,
+      "step": 10261
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.500885249413126e-07,
+      "loss": 0.267,
+      "step": 10262
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.490339419014147e-07,
+      "loss": 0.2605,
+      "step": 10263
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.47980071880271e-07,
+      "loss": 0.2557,
+      "step": 10264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.469269149591074e-07,
+      "loss": 0.2662,
+      "step": 10265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.45874471219098e-07,
+      "loss": 0.2609,
+      "step": 10266
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.448227407413566e-07,
+      "loss": 0.2703,
+      "step": 10267
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.437717236069453e-07,
+      "loss": 0.2714,
+      "step": 10268
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.427214198968769e-07,
+      "loss": 0.2731,
+      "step": 10269
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.416718296920977e-07,
+      "loss": 0.233,
+      "step": 10270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.406229530735076e-07,
+      "loss": 0.3277,
+      "step": 10271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.395747901219474e-07,
+      "loss": 0.3146,
+      "step": 10272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.385273409182081e-07,
+      "loss": 0.2561,
+      "step": 10273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.37480605543015e-07,
+      "loss": 0.3088,
+      "step": 10274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.364345840770515e-07,
+      "loss": 0.2448,
+      "step": 10275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.353892766009396e-07,
+      "loss": 0.2888,
+      "step": 10276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.343446831952428e-07,
+      "loss": 0.6007,
+      "step": 10277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.333008039404743e-07,
+      "loss": 0.2688,
+      "step": 10278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.322576389170921e-07,
+      "loss": 0.2797,
+      "step": 10279
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.312151882054974e-07,
+      "loss": 0.3083,
+      "step": 10280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.301734518860381e-07,
+      "loss": 0.3824,
+      "step": 10281
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.291324300390057e-07,
+      "loss": 0.2395,
+      "step": 10282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.280921227446381e-07,
+      "loss": 0.3444,
+      "step": 10283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.270525300831133e-07,
+      "loss": 0.3143,
+      "step": 10284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.260136521345618e-07,
+      "loss": 0.272,
+      "step": 10285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.249754889790539e-07,
+      "loss": 0.2831,
+      "step": 10286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.239380406966057e-07,
+      "loss": 0.5676,
+      "step": 10287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.229013073671787e-07,
+      "loss": 0.2584,
+      "step": 10288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.218652890706801e-07,
+      "loss": 0.2538,
+      "step": 10289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.208299858869616e-07,
+      "loss": 0.2994,
+      "step": 10290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.197953978958161e-07,
+      "loss": 0.2445,
+      "step": 10291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.187615251769875e-07,
+      "loss": 0.2707,
+      "step": 10292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.177283678101598e-07,
+      "loss": 0.3119,
+      "step": 10293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.16695925874964e-07,
+      "loss": 0.3141,
+      "step": 10294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.156641994509772e-07,
+      "loss": 0.2675,
+      "step": 10295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.146331886177183e-07,
+      "loss": 0.291,
+      "step": 10296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.136028934546546e-07,
+      "loss": 0.2991,
+      "step": 10297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.125733140411928e-07,
+      "loss": 0.545,
+      "step": 10298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.115444504566882e-07,
+      "loss": 0.2623,
+      "step": 10299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.105163027804429e-07,
+      "loss": 0.2701,
+      "step": 10300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.094888710917003e-07,
+      "loss": 0.3021,
+      "step": 10301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.084621554696502e-07,
+      "loss": 0.2534,
+      "step": 10302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.074361559934251e-07,
+      "loss": 0.2833,
+      "step": 10303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.064108727421082e-07,
+      "loss": 0.2531,
+      "step": 10304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.053863057947175e-07,
+      "loss": 0.2766,
+      "step": 10305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.043624552302231e-07,
+      "loss": 0.2597,
+      "step": 10306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.03339321127543e-07,
+      "loss": 0.6469,
+      "step": 10307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.023169035655297e-07,
+      "loss": 0.2501,
+      "step": 10308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.012952026229892e-07,
+      "loss": 0.267,
+      "step": 10309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.002742183786671e-07,
+      "loss": 0.275,
+      "step": 10310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.992539509112595e-07,
+      "loss": 0.3052,
+      "step": 10311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.98234400299399e-07,
+      "loss": 0.269,
+      "step": 10312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.972155666216684e-07,
+      "loss": 0.2802,
+      "step": 10313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.961974499565982e-07,
+      "loss": 0.2859,
+      "step": 10314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.951800503826556e-07,
+      "loss": 0.2609,
+      "step": 10315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.941633679782578e-07,
+      "loss": 0.2773,
+      "step": 10316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.931474028217666e-07,
+      "loss": 0.2495,
+      "step": 10317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.921321549914872e-07,
+      "loss": 0.2581,
+      "step": 10318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.911176245656693e-07,
+      "loss": 0.3005,
+      "step": 10319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.90103811622509e-07,
+      "loss": 0.2834,
+      "step": 10320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.890907162401473e-07,
+      "loss": 0.2974,
+      "step": 10321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.880783384966638e-07,
+      "loss": 0.2846,
+      "step": 10322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.870666784700908e-07,
+      "loss": 0.2733,
+      "step": 10323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.860557362384024e-07,
+      "loss": 0.3167,
+      "step": 10324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.850455118795152e-07,
+      "loss": 0.2252,
+      "step": 10325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.840360054712946e-07,
+      "loss": 0.2889,
+      "step": 10326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.830272170915464e-07,
+      "loss": 0.283,
+      "step": 10327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.82019146818026e-07,
+      "loss": 0.2692,
+      "step": 10328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.810117947284256e-07,
+      "loss": 0.306,
+      "step": 10329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.800051609003911e-07,
+      "loss": 0.2507,
+      "step": 10330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.789992454115058e-07,
+      "loss": 0.2523,
+      "step": 10331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.779940483393033e-07,
+      "loss": 0.2793,
+      "step": 10332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.769895697612572e-07,
+      "loss": 0.2893,
+      "step": 10333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7598580975479e-07,
+      "loss": 0.2636,
+      "step": 10334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.749827683972654e-07,
+      "loss": 0.2209,
+      "step": 10335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.739804457659927e-07,
+      "loss": 0.2809,
+      "step": 10336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.729788419382255e-07,
+      "loss": 0.254,
+      "step": 10337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.719779569911622e-07,
+      "loss": 0.2605,
+      "step": 10338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.709777910019477e-07,
+      "loss": 0.2521,
+      "step": 10339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.699783440476681e-07,
+      "loss": 0.2465,
+      "step": 10340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.689796162053575e-07,
+      "loss": 0.2467,
+      "step": 10341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.67981607551994e-07,
+      "loss": 0.6031,
+      "step": 10342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.669843181644953e-07,
+      "loss": 0.265,
+      "step": 10343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.659877481197285e-07,
+      "loss": 0.2589,
+      "step": 10344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.64991897494508e-07,
+      "loss": 0.2261,
+      "step": 10345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.639967663655844e-07,
+      "loss": 0.2673,
+      "step": 10346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.630023548096609e-07,
+      "loss": 0.2745,
+      "step": 10347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.620086629033795e-07,
+      "loss": 0.296,
+      "step": 10348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.610156907233312e-07,
+      "loss": 0.2382,
+      "step": 10349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.600234383460469e-07,
+      "loss": 0.2868,
+      "step": 10350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.590319058480044e-07,
+      "loss": 0.3128,
+      "step": 10351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.580410933056314e-07,
+      "loss": 0.2762,
+      "step": 10352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.57051000795288e-07,
+      "loss": 0.3168,
+      "step": 10353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.560616283932897e-07,
+      "loss": 0.2469,
+      "step": 10354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.550729761758901e-07,
+      "loss": 0.2615,
+      "step": 10355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.540850442192926e-07,
+      "loss": 0.2931,
+      "step": 10356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.530978325996385e-07,
+      "loss": 0.2956,
+      "step": 10357
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.521113413930202e-07,
+      "loss": 0.2905,
+      "step": 10358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.511255706754715e-07,
+      "loss": 0.2384,
+      "step": 10359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.501405205229683e-07,
+      "loss": 0.2676,
+      "step": 10360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.491561910114352e-07,
+      "loss": 0.3158,
+      "step": 10361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.481725822167384e-07,
+      "loss": 0.2765,
+      "step": 10362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.471896942146905e-07,
+      "loss": 0.2473,
+      "step": 10363
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.462075270810475e-07,
+      "loss": 0.263,
+      "step": 10364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4522608089151e-07,
+      "loss": 0.2588,
+      "step": 10365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.442453557217243e-07,
+      "loss": 0.2292,
+      "step": 10366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.432653516472765e-07,
+      "loss": 0.3415,
+      "step": 10367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.422860687437027e-07,
+      "loss": 0.2875,
+      "step": 10368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.413075070864805e-07,
+      "loss": 0.2634,
+      "step": 10369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.403296667510339e-07,
+      "loss": 0.3031,
+      "step": 10370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.393525478127282e-07,
+      "loss": 0.2927,
+      "step": 10371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.383761503468766e-07,
+      "loss": 0.2823,
+      "step": 10372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.374004744287354e-07,
+      "loss": 0.2841,
+      "step": 10373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.364255201335013e-07,
+      "loss": 0.2617,
+      "step": 10374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.354512875363228e-07,
+      "loss": 0.3224,
+      "step": 10375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.344777767122867e-07,
+      "loss": 0.2996,
+      "step": 10376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.335049877364264e-07,
+      "loss": 0.2654,
+      "step": 10377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.325329206837217e-07,
+      "loss": 0.2483,
+      "step": 10378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.315615756290927e-07,
+      "loss": 0.2231,
+      "step": 10379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.305909526474085e-07,
+      "loss": 0.265,
+      "step": 10380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.296210518134771e-07,
+      "loss": 0.2485,
+      "step": 10381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.286518732020519e-07,
+      "loss": 0.238,
+      "step": 10382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.276834168878388e-07,
+      "loss": 0.2448,
+      "step": 10383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.267156829454768e-07,
+      "loss": 0.3331,
+      "step": 10384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.257486714495542e-07,
+      "loss": 0.2851,
+      "step": 10385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.247823824746058e-07,
+      "loss": 0.2515,
+      "step": 10386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.238168160951075e-07,
+      "loss": 0.3083,
+      "step": 10387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.228519723854787e-07,
+      "loss": 0.2543,
+      "step": 10388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.218878514200843e-07,
+      "loss": 0.2869,
+      "step": 10389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.209244532732394e-07,
+      "loss": 0.259,
+      "step": 10390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.199617780191924e-07,
+      "loss": 0.2486,
+      "step": 10391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.189998257321428e-07,
+      "loss": 0.2713,
+      "step": 10392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.180385964862346e-07,
+      "loss": 0.3004,
+      "step": 10393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.170780903555529e-07,
+      "loss": 0.237,
+      "step": 10394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.161183074141319e-07,
+      "loss": 0.2733,
+      "step": 10395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.151592477359413e-07,
+      "loss": 0.5713,
+      "step": 10396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.142009113949076e-07,
+      "loss": 0.2364,
+      "step": 10397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.132432984648895e-07,
+      "loss": 0.2567,
+      "step": 10398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.122864090196956e-07,
+      "loss": 0.6224,
+      "step": 10399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.113302431330803e-07,
+      "loss": 0.2811,
+      "step": 10400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.10374800878738e-07,
+      "loss": 0.3021,
+      "step": 10401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.09420082330312e-07,
+      "loss": 0.3235,
+      "step": 10402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.084660875613846e-07,
+      "loss": 0.2883,
+      "step": 10403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.07512816645488e-07,
+      "loss": 0.2834,
+      "step": 10404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.065602696560924e-07,
+      "loss": 0.2847,
+      "step": 10405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.056084466666167e-07,
+      "loss": 0.2787,
+      "step": 10406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.046573477504236e-07,
+      "loss": 0.2586,
+      "step": 10407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.037069729808188e-07,
+      "loss": 0.2845,
+      "step": 10408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.027573224310501e-07,
+      "loss": 0.2635,
+      "step": 10409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.01808396174316e-07,
+      "loss": 0.2975,
+      "step": 10410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.008601942837533e-07,
+      "loss": 0.2719,
+      "step": 10411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.999127168324425e-07,
+      "loss": 0.3004,
+      "step": 10412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.989659638934131e-07,
+      "loss": 0.2484,
+      "step": 10413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.980199355396343e-07,
+      "loss": 0.2873,
+      "step": 10414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.970746318440235e-07,
+      "loss": 0.2653,
+      "step": 10415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.961300528794378e-07,
+      "loss": 0.267,
+      "step": 10416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.951861987186824e-07,
+      "loss": 0.2244,
+      "step": 10417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.942430694345058e-07,
+      "loss": 0.267,
+      "step": 10418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.933006650995965e-07,
+      "loss": 0.2414,
+      "step": 10419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.923589857865908e-07,
+      "loss": 0.258,
+      "step": 10420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.914180315680729e-07,
+      "loss": 0.2464,
+      "step": 10421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.904778025165614e-07,
+      "loss": 0.2617,
+      "step": 10422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.895382987045284e-07,
+      "loss": 0.2484,
+      "step": 10423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.885995202043848e-07,
+      "loss": 0.2515,
+      "step": 10424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.876614670884873e-07,
+      "loss": 0.2789,
+      "step": 10425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.867241394291356e-07,
+      "loss": 0.2703,
+      "step": 10426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.857875372985733e-07,
+      "loss": 0.2812,
+      "step": 10427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.848516607689947e-07,
+      "loss": 0.2932,
+      "step": 10428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.839165099125254e-07,
+      "loss": 0.3059,
+      "step": 10429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.829820848012457e-07,
+      "loss": 0.3168,
+      "step": 10430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.820483855071768e-07,
+      "loss": 0.2957,
+      "step": 10431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.811154121022821e-07,
+      "loss": 0.2262,
+      "step": 10432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.801831646584732e-07,
+      "loss": 0.2581,
+      "step": 10433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.79251643247598e-07,
+      "loss": 0.2501,
+      "step": 10434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.783208479414603e-07,
+      "loss": 0.2808,
+      "step": 10435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.77390778811796e-07,
+      "loss": 0.6025,
+      "step": 10436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.764614359302912e-07,
+      "loss": 0.2513,
+      "step": 10437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.755328193685772e-07,
+      "loss": 0.2607,
+      "step": 10438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.746049291982247e-07,
+      "loss": 0.2401,
+      "step": 10439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.736777654907532e-07,
+      "loss": 0.278,
+      "step": 10440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.727513283176222e-07,
+      "loss": 0.2461,
+      "step": 10441
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.718256177502379e-07,
+      "loss": 0.2544,
+      "step": 10442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.709006338599488e-07,
+      "loss": 0.2806,
+      "step": 10443
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.699763767180477e-07,
+      "loss": 0.2455,
+      "step": 10444
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.690528463957734e-07,
+      "loss": 0.2767,
+      "step": 10445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.681300429643044e-07,
+      "loss": 0.2718,
+      "step": 10446
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.672079664947683e-07,
+      "loss": 0.6139,
+      "step": 10447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.662866170582338e-07,
+      "loss": 0.2974,
+      "step": 10448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.653659947257139e-07,
+      "loss": 0.2699,
+      "step": 10449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.644460995681644e-07,
+      "loss": 0.2922,
+      "step": 10450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.635269316564873e-07,
+      "loss": 0.2414,
+      "step": 10451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.62608491061527e-07,
+      "loss": 0.2637,
+      "step": 10452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.616907778540726e-07,
+      "loss": 0.2783,
+      "step": 10453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.607737921048573e-07,
+      "loss": 0.2968,
+      "step": 10454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.598575338845569e-07,
+      "loss": 0.2773,
+      "step": 10455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.589420032637949e-07,
+      "loss": 0.2729,
+      "step": 10456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.580272003131315e-07,
+      "loss": 0.2513,
+      "step": 10457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.57113125103077e-07,
+      "loss": 0.2886,
+      "step": 10458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.561997777040862e-07,
+      "loss": 0.2676,
+      "step": 10459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.552871581865515e-07,
+      "loss": 0.2918,
+      "step": 10460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.543752666208158e-07,
+      "loss": 0.2942,
+      "step": 10461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.534641030771615e-07,
+      "loss": 0.2472,
+      "step": 10462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.525536676258203e-07,
+      "loss": 0.31,
+      "step": 10463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.516439603369583e-07,
+      "loss": 0.2626,
+      "step": 10464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.507349812806939e-07,
+      "loss": 0.2842,
+      "step": 10465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.498267305270888e-07,
+      "loss": 0.302,
+      "step": 10466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.489192081461436e-07,
+      "loss": 0.3217,
+      "step": 10467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.48012414207807e-07,
+      "loss": 0.2433,
+      "step": 10468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.471063487819695e-07,
+      "loss": 0.2498,
+      "step": 10469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.462010119384665e-07,
+      "loss": 0.2684,
+      "step": 10470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.452964037470776e-07,
+      "loss": 0.2736,
+      "step": 10471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.443925242775227e-07,
+      "loss": 0.2201,
+      "step": 10472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.43489373599474e-07,
+      "loss": 0.2712,
+      "step": 10473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.425869517825366e-07,
+      "loss": 0.2578,
+      "step": 10474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.416852588962662e-07,
+      "loss": 0.2773,
+      "step": 10475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.407842950101605e-07,
+      "loss": 0.3067,
+      "step": 10476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.398840601936628e-07,
+      "loss": 0.27,
+      "step": 10477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.389845545161598e-07,
+      "loss": 0.2852,
+      "step": 10478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.380857780469762e-07,
+      "loss": 0.2975,
+      "step": 10479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.371877308553896e-07,
+      "loss": 0.2757,
+      "step": 10480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.36290413010615e-07,
+      "loss": 0.2927,
+      "step": 10481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.353938245818147e-07,
+      "loss": 0.2905,
+      "step": 10482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.34497965638091e-07,
+      "loss": 0.3113,
+      "step": 10483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.336028362484947e-07,
+      "loss": 0.2873,
+      "step": 10484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.327084364820168e-07,
+      "loss": 0.2693,
+      "step": 10485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.318147664075923e-07,
+      "loss": 0.2186,
+      "step": 10486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.30921826094104e-07,
+      "loss": 0.3048,
+      "step": 10487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.300296156103723e-07,
+      "loss": 0.6042,
+      "step": 10488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.291381350251645e-07,
+      "loss": 0.2502,
+      "step": 10489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.282473844071933e-07,
+      "loss": 0.2612,
+      "step": 10490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.273573638251117e-07,
+      "loss": 0.2348,
+      "step": 10491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.264680733475192e-07,
+      "loss": 0.2571,
+      "step": 10492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.255795130429575e-07,
+      "loss": 0.286,
+      "step": 10493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.246916829799132e-07,
+      "loss": 0.2637,
+      "step": 10494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.238045832268146e-07,
+      "loss": 0.271,
+      "step": 10495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.229182138520339e-07,
+      "loss": 0.2603,
+      "step": 10496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.220325749238919e-07,
+      "loss": 0.3203,
+      "step": 10497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.211476665106463e-07,
+      "loss": 0.2745,
+      "step": 10498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.202634886805013e-07,
+      "loss": 0.2832,
+      "step": 10499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.193800415016071e-07,
+      "loss": 0.3075,
+      "step": 10500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.184973250420544e-07,
+      "loss": 0.3088,
+      "step": 10501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.17615339369878e-07,
+      "loss": 0.2941,
+      "step": 10502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.167340845530544e-07,
+      "loss": 0.2606,
+      "step": 10503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.158535606595138e-07,
+      "loss": 0.2651,
+      "step": 10504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.149737677571164e-07,
+      "loss": 0.2852,
+      "step": 10505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.140947059136736e-07,
+      "loss": 0.2871,
+      "step": 10506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.132163751969399e-07,
+      "loss": 0.5829,
+      "step": 10507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.123387756746123e-07,
+      "loss": 0.2967,
+      "step": 10508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.114619074143335e-07,
+      "loss": 0.2571,
+      "step": 10509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.105857704836836e-07,
+      "loss": 0.5992,
+      "step": 10510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.097103649501977e-07,
+      "loss": 0.3119,
+      "step": 10511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.088356908813419e-07,
+      "loss": 0.2571,
+      "step": 10512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.079617483445332e-07,
+      "loss": 0.2983,
+      "step": 10513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.070885374071321e-07,
+      "loss": 0.2593,
+      "step": 10514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.062160581364406e-07,
+      "loss": 0.2964,
+      "step": 10515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.053443105997069e-07,
+      "loss": 0.2898,
+      "step": 10516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.044732948641162e-07,
+      "loss": 0.2579,
+      "step": 10517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.036030109968082e-07,
+      "loss": 0.2892,
+      "step": 10518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.027334590648547e-07,
+      "loss": 0.2431,
+      "step": 10519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.018646391352799e-07,
+      "loss": 0.2501,
+      "step": 10520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.00996551275047e-07,
+      "loss": 0.2723,
+      "step": 10521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.001291955510634e-07,
+      "loss": 0.2573,
+      "step": 10522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.992625720301814e-07,
+      "loss": 0.3287,
+      "step": 10523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.983966807791951e-07,
+      "loss": 0.2615,
+      "step": 10524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.975315218648457e-07,
+      "loss": 0.2998,
+      "step": 10525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.966670953538133e-07,
+      "loss": 0.2562,
+      "step": 10526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.958034013127222e-07,
+      "loss": 0.3052,
+      "step": 10527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.949404398081448e-07,
+      "loss": 0.2487,
+      "step": 10528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.940782109065911e-07,
+      "loss": 0.2537,
+      "step": 10529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.932167146745193e-07,
+      "loss": 0.2241,
+      "step": 10530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.923559511783282e-07,
+      "loss": 0.256,
+      "step": 10531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.914959204843639e-07,
+      "loss": 0.2524,
+      "step": 10532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.906366226589099e-07,
+      "loss": 0.2492,
+      "step": 10533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.897780577681954e-07,
+      "loss": 0.2391,
+      "step": 10534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.889202258784009e-07,
+      "loss": 0.2836,
+      "step": 10535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.88063127055638e-07,
+      "loss": 0.2707,
+      "step": 10536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.872067613659692e-07,
+      "loss": 0.5853,
+      "step": 10537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.863511288753986e-07,
+      "loss": 0.2502,
+      "step": 10538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.854962296498766e-07,
+      "loss": 0.3179,
+      "step": 10539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.846420637552918e-07,
+      "loss": 0.2991,
+      "step": 10540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.83788631257478e-07,
+      "loss": 0.3018,
+      "step": 10541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.829359322222182e-07,
+      "loss": 0.2626,
+      "step": 10542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.820839667152299e-07,
+      "loss": 0.3276,
+      "step": 10543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.812327348021805e-07,
+      "loss": 0.2945,
+      "step": 10544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.803822365486787e-07,
+      "loss": 0.3055,
+      "step": 10545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.795324720202754e-07,
+      "loss": 0.2872,
+      "step": 10546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.786834412824692e-07,
+      "loss": 0.2974,
+      "step": 10547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.778351444006946e-07,
+      "loss": 0.2582,
+      "step": 10548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.76987581440338e-07,
+      "loss": 0.2956,
+      "step": 10549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.761407524667239e-07,
+      "loss": 0.3148,
+      "step": 10550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7529465754512226e-07,
+      "loss": 0.3064,
+      "step": 10551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7444929674074435e-07,
+      "loss": 0.2509,
+      "step": 10552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.73604670118748e-07,
+      "loss": 0.2363,
+      "step": 10553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7276077774423334e-07,
+      "loss": 0.2545,
+      "step": 10554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7191761968224057e-07,
+      "loss": 0.2847,
+      "step": 10555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7107519599775997e-07,
+      "loss": 0.2504,
+      "step": 10556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7023350675571846e-07,
+      "loss": 0.6125,
+      "step": 10557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.693925520209908e-07,
+      "loss": 0.2773,
+      "step": 10558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6855233185839175e-07,
+      "loss": 0.2301,
+      "step": 10559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.677128463326841e-07,
+      "loss": 0.3303,
+      "step": 10560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.668740955085704e-07,
+      "loss": 0.2538,
+      "step": 10561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.660360794506946e-07,
+      "loss": 0.261,
+      "step": 10562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.651987982236517e-07,
+      "loss": 0.2473,
+      "step": 10563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6436225189197103e-07,
+      "loss": 0.2572,
+      "step": 10564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.635264405201312e-07,
+      "loss": 0.3335,
+      "step": 10565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6269136417255167e-07,
+      "loss": 0.2767,
+      "step": 10566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.618570229135966e-07,
+      "loss": 0.2702,
+      "step": 10567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.610234168075733e-07,
+      "loss": 0.3082,
+      "step": 10568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.601905459187317e-07,
+      "loss": 0.2458,
+      "step": 10569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5935841031126693e-07,
+      "loss": 0.2302,
+      "step": 10570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.585270100493122e-07,
+      "loss": 0.2791,
+      "step": 10571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.576963451969485e-07,
+      "loss": 0.2523,
+      "step": 10572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5686641581820345e-07,
+      "loss": 0.2899,
+      "step": 10573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5603722197703925e-07,
+      "loss": 0.2668,
+      "step": 10574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.552087637373681e-07,
+      "loss": 0.2677,
+      "step": 10575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.543810411630434e-07,
+      "loss": 0.2477,
+      "step": 10576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.535540543178629e-07,
+      "loss": 0.2773,
+      "step": 10577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5272780326556466e-07,
+      "loss": 0.2397,
+      "step": 10578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5190228806983205e-07,
+      "loss": 0.2968,
+      "step": 10579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5107750879429424e-07,
+      "loss": 0.2383,
+      "step": 10580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5025346550251815e-07,
+      "loss": 0.2869,
+      "step": 10581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.494301582580185e-07,
+      "loss": 0.2935,
+      "step": 10582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4860758712425124e-07,
+      "loss": 0.2528,
+      "step": 10583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4778575216461673e-07,
+      "loss": 0.2806,
+      "step": 10584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4696465344245874e-07,
+      "loss": 0.2936,
+      "step": 10585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4614429102105893e-07,
+      "loss": 0.2203,
+      "step": 10586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4532466496365334e-07,
+      "loss": 0.2712,
+      "step": 10587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4450577533341035e-07,
+      "loss": 0.2574,
+      "step": 10588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.436876221934472e-07,
+      "loss": 0.295,
+      "step": 10589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4287020560682345e-07,
+      "loss": 0.263,
+      "step": 10590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.42053525636541e-07,
+      "loss": 0.2476,
+      "step": 10591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4123758234554615e-07,
+      "loss": 0.2755,
+      "step": 10592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.404223757967252e-07,
+      "loss": 0.5533,
+      "step": 10593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.396079060529146e-07,
+      "loss": 0.2665,
+      "step": 10594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.387941731768852e-07,
+      "loss": 0.267,
+      "step": 10595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3798117723135915e-07,
+      "loss": 0.2541,
+      "step": 10596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.371689182789962e-07,
+      "loss": 0.2609,
+      "step": 10597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.363573963824008e-07,
+      "loss": 0.2586,
+      "step": 10598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3554661160412406e-07,
+      "loss": 0.2878,
+      "step": 10599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3473656400665256e-07,
+      "loss": 0.2692,
+      "step": 10600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.339272536524253e-07,
+      "loss": 0.2986,
+      "step": 10601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.331186806038179e-07,
+      "loss": 0.2589,
+      "step": 10602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.323108449231506e-07,
+      "loss": 0.2644,
+      "step": 10603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.315037466726879e-07,
+      "loss": 0.269,
+      "step": 10604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.306973859146368e-07,
+      "loss": 0.2852,
+      "step": 10605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.298917627111476e-07,
+      "loss": 0.2578,
+      "step": 10606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.290868771243151e-07,
+      "loss": 0.2289,
+      "step": 10607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.282827292161762e-07,
+      "loss": 0.241,
+      "step": 10608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2747931904870697e-07,
+      "loss": 0.2812,
+      "step": 10609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.266766466838335e-07,
+      "loss": 0.3018,
+      "step": 10610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2587471218342057e-07,
+      "loss": 0.6152,
+      "step": 10611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.250735156092778e-07,
+      "loss": 0.2538,
+      "step": 10612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.242730570231568e-07,
+      "loss": 0.2151,
+      "step": 10613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2347333648675383e-07,
+      "loss": 0.2544,
+      "step": 10614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2267435406170845e-07,
+      "loss": 0.3024,
+      "step": 10615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2187610980959916e-07,
+      "loss": 0.3073,
+      "step": 10616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.210786037919523e-07,
+      "loss": 0.2786,
+      "step": 10617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2028183607023766e-07,
+      "loss": 0.3167,
+      "step": 10618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.194858067058627e-07,
+      "loss": 0.2816,
+      "step": 10619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.186905157601839e-07,
+      "loss": 0.2636,
+      "step": 10620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1789596329449676e-07,
+      "loss": 0.6168,
+      "step": 10621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1710214937004223e-07,
+      "loss": 0.2958,
+      "step": 10622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1630907404800467e-07,
+      "loss": 0.2322,
+      "step": 10623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.155167373895075e-07,
+      "loss": 0.3427,
+      "step": 10624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.14725139455624e-07,
+      "loss": 0.2393,
+      "step": 10625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.139342803073632e-07,
+      "loss": 0.2492,
+      "step": 10626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.131441600056818e-07,
+      "loss": 0.295,
+      "step": 10627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.12354778611479e-07,
+      "loss": 0.2583,
+      "step": 10628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.11566136185596e-07,
+      "loss": 0.2529,
+      "step": 10629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1077823278881767e-07,
+      "loss": 0.2549,
+      "step": 10630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.099910684818698e-07,
+      "loss": 0.2022,
+      "step": 10631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.092046433254271e-07,
+      "loss": 0.3221,
+      "step": 10632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.084189573801001e-07,
+      "loss": 0.2886,
+      "step": 10633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.076340107064458e-07,
+      "loss": 0.274,
+      "step": 10634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0684980336496484e-07,
+      "loss": 0.282,
+      "step": 10635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0606633541609983e-07,
+      "loss": 0.2231,
+      "step": 10636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.052836069202382e-07,
+      "loss": 0.2781,
+      "step": 10637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.045016179377048e-07,
+      "loss": 0.2819,
+      "step": 10638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0372036852877607e-07,
+      "loss": 0.2317,
+      "step": 10639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.029398587536637e-07,
+      "loss": 0.2639,
+      "step": 10640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.021600886725263e-07,
+      "loss": 0.2709,
+      "step": 10641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.013810583454647e-07,
+      "loss": 0.3183,
+      "step": 10642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.006027678325242e-07,
+      "loss": 0.2614,
+      "step": 10643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9982521719369003e-07,
+      "loss": 0.2722,
+      "step": 10644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.990484064888911e-07,
+      "loss": 0.2424,
+      "step": 10645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.982723357780027e-07,
+      "loss": 0.2834,
+      "step": 10646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9749700512083824e-07,
+      "loss": 0.24,
+      "step": 10647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9672241457715755e-07,
+      "loss": 0.2673,
+      "step": 10648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.959485642066618e-07,
+      "loss": 0.2567,
+      "step": 10649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.951754540689956e-07,
+      "loss": 0.2541,
+      "step": 10650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.944030842237467e-07,
+      "loss": 0.2701,
+      "step": 10651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.936314547304454e-07,
+      "loss": 0.2679,
+      "step": 10652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.928605656485662e-07,
+      "loss": 0.2813,
+      "step": 10653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.920904170375239e-07,
+      "loss": 0.2947,
+      "step": 10654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.913210089566766e-07,
+      "loss": 0.2719,
+      "step": 10655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.905523414653301e-07,
+      "loss": 0.2612,
+      "step": 10656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.897844146227259e-07,
+      "loss": 0.2624,
+      "step": 10657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8901722848805443e-07,
+      "loss": 0.2831,
+      "step": 10658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8825078312044515e-07,
+      "loss": 0.2256,
+      "step": 10659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8748507857897187e-07,
+      "loss": 0.2902,
+      "step": 10660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8672011492265404e-07,
+      "loss": 0.2784,
+      "step": 10661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8595589221044674e-07,
+      "loss": 0.277,
+      "step": 10662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8519241050125724e-07,
+      "loss": 0.2922,
+      "step": 10663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.844296698539274e-07,
+      "loss": 0.2989,
+      "step": 10664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8366767032724685e-07,
+      "loss": 0.2772,
+      "step": 10665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8290641197994526e-07,
+      "loss": 0.2678,
+      "step": 10666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.821458948706991e-07,
+      "loss": 0.323,
+      "step": 10667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8138611905812584e-07,
+      "loss": 0.253,
+      "step": 10668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.806270846007798e-07,
+      "loss": 0.308,
+      "step": 10669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7986879155717084e-07,
+      "loss": 0.2569,
+      "step": 10670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7911123998573995e-07,
+      "loss": 0.3113,
+      "step": 10671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.78354429944876e-07,
+      "loss": 0.2806,
+      "step": 10672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.775983614929102e-07,
+      "loss": 0.2773,
+      "step": 10673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.768430346881169e-07,
+      "loss": 0.2744,
+      "step": 10674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.760884495887152e-07,
+      "loss": 0.2924,
+      "step": 10675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7533460625285955e-07,
+      "loss": 0.2761,
+      "step": 10676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7458150473865806e-07,
+      "loss": 0.2402,
+      "step": 10677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7382914510415316e-07,
+      "loss": 0.3066,
+      "step": 10678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7307752740733283e-07,
+      "loss": 0.2853,
+      "step": 10679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7232665170612857e-07,
+      "loss": 0.249,
+      "step": 10680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.715765180584141e-07,
+      "loss": 0.287,
+      "step": 10681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.708271265220087e-07,
+      "loss": 0.2442,
+      "step": 10682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7007847715466506e-07,
+      "loss": 0.2762,
+      "step": 10683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6933057001409366e-07,
+      "loss": 0.2545,
+      "step": 10684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.685834051579329e-07,
+      "loss": 0.3158,
+      "step": 10685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.678369826437733e-07,
+      "loss": 0.2856,
+      "step": 10686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.670913025291456e-07,
+      "loss": 0.2534,
+      "step": 10687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.663463648715226e-07,
+      "loss": 0.2485,
+      "step": 10688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.656021697283196e-07,
+      "loss": 0.283,
+      "step": 10689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6485871715689735e-07,
+      "loss": 0.2891,
+      "step": 10690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6411600721455776e-07,
+      "loss": 0.2845,
+      "step": 10691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.633740399585428e-07,
+      "loss": 0.2684,
+      "step": 10692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6263281544603903e-07,
+      "loss": 0.2914,
+      "step": 10693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6189233373418064e-07,
+      "loss": 0.2717,
+      "step": 10694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.611525948800376e-07,
+      "loss": 0.2533,
+      "step": 10695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6041359894062544e-07,
+      "loss": 0.2678,
+      "step": 10696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.596753459729019e-07,
+      "loss": 0.2552,
+      "step": 10697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.589378360337692e-07,
+      "loss": 0.2858,
+      "step": 10698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.582010691800708e-07,
+      "loss": 0.2289,
+      "step": 10699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.574650454685902e-07,
+      "loss": 0.2962,
+      "step": 10700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5672976495606084e-07,
+      "loss": 0.2483,
+      "step": 10701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5599522769915074e-07,
+      "loss": 0.24,
+      "step": 10702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5526143375447684e-07,
+      "loss": 0.2428,
+      "step": 10703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5452838317859615e-07,
+      "loss": 0.3191,
+      "step": 10704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.537960760280068e-07,
+      "loss": 0.2846,
+      "step": 10705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5306451235915475e-07,
+      "loss": 0.2684,
+      "step": 10706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.523336922284204e-07,
+      "loss": 0.2854,
+      "step": 10707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5160361569213766e-07,
+      "loss": 0.3141,
+      "step": 10708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5087428280657144e-07,
+      "loss": 0.2583,
+      "step": 10709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.50145693627939e-07,
+      "loss": 0.2971,
+      "step": 10710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4941784821239445e-07,
+      "loss": 0.2782,
+      "step": 10711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.486907466160372e-07,
+      "loss": 0.2786,
+      "step": 10712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4796438889491025e-07,
+      "loss": 0.2624,
+      "step": 10713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472387751049944e-07,
+      "loss": 0.2983,
+      "step": 10714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4651390530221927e-07,
+      "loss": 0.2867,
+      "step": 10715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.457897795424525e-07,
+      "loss": 0.303,
+      "step": 10716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.450663978815061e-07,
+      "loss": 0.3206,
+      "step": 10717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.443437603751354e-07,
+      "loss": 0.2963,
+      "step": 10718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.436218670790381e-07,
+      "loss": 0.2662,
+      "step": 10719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4290071804885526e-07,
+      "loss": 0.2805,
+      "step": 10720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4218031334016465e-07,
+      "loss": 0.2844,
+      "step": 10721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.414606530084974e-07,
+      "loss": 0.2551,
+      "step": 10722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": 0.2351,
+      "step": 10723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4002356569803775e-07,
+      "loss": 0.2902,
+      "step": 10724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3930613883000897e-07,
+      "loss": 0.2559,
+      "step": 10725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3858945656052855e-07,
+      "loss": 0.238,
+      "step": 10726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3787351894483566e-07,
+      "loss": 0.2661,
+      "step": 10727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3715832603810727e-07,
+      "loss": 0.2454,
+      "step": 10728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3644387789547264e-07,
+      "loss": 0.2573,
+      "step": 10729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.357301745719932e-07,
+      "loss": 0.2639,
+      "step": 10730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3501721612267833e-07,
+      "loss": 0.2828,
+      "step": 10731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.34305002602483e-07,
+      "loss": 0.3252,
+      "step": 10732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.335935340662966e-07,
+      "loss": 0.2589,
+      "step": 10733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3288281056895746e-07,
+      "loss": 0.2573,
+      "step": 10734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.321728321652451e-07,
+      "loss": 0.345,
+      "step": 10735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.314635989098802e-07,
+      "loss": 0.3053,
+      "step": 10736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.307551108575291e-07,
+      "loss": 0.2269,
+      "step": 10737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.300473680627947e-07,
+      "loss": 0.2351,
+      "step": 10738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2934037058023115e-07,
+      "loss": 0.2972,
+      "step": 10739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.28634118464326e-07,
+      "loss": 0.2173,
+      "step": 10740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2792861176951465e-07,
+      "loss": 0.2764,
+      "step": 10741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2722385055017567e-07,
+      "loss": 0.262,
+      "step": 10742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.26519834860628e-07,
+      "loss": 0.3237,
+      "step": 10743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.258165647551337e-07,
+      "loss": 0.2888,
+      "step": 10744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2511404028789604e-07,
+      "loss": 0.2547,
+      "step": 10745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2441226151306403e-07,
+      "loss": 0.2758,
+      "step": 10746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2371122848472655e-07,
+      "loss": 0.222,
+      "step": 10747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.230109412569149e-07,
+      "loss": 0.267,
+      "step": 10748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.223113998836036e-07,
+      "loss": 0.265,
+      "step": 10749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.216126044187118e-07,
+      "loss": 0.2225,
+      "step": 10750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.209145549160997e-07,
+      "loss": 0.2956,
+      "step": 10751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2021725142956537e-07,
+      "loss": 0.2444,
+      "step": 10752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1952069401285814e-07,
+      "loss": 0.2349,
+      "step": 10753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.188248827196616e-07,
+      "loss": 0.2723,
+      "step": 10754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.181298176036074e-07,
+      "loss": 0.5592,
+      "step": 10755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1743549871826704e-07,
+      "loss": 0.2473,
+      "step": 10756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.167419261171556e-07,
+      "loss": 0.2812,
+      "step": 10757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.160490998537313e-07,
+      "loss": 0.288,
+      "step": 10758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1535701998139045e-07,
+      "loss": 0.2789,
+      "step": 10759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.146656865534803e-07,
+      "loss": 0.2982,
+      "step": 10760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.139750996232804e-07,
+      "loss": 0.2981,
+      "step": 10761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.132852592440194e-07,
+      "loss": 0.2327,
+      "step": 10762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1259616546886804e-07,
+      "loss": 0.2523,
+      "step": 10763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.119078183509372e-07,
+      "loss": 0.3148,
+      "step": 10764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1122021794328214e-07,
+      "loss": 0.2629,
+      "step": 10765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1053336429889616e-07,
+      "loss": 0.5923,
+      "step": 10766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.098472574707245e-07,
+      "loss": 0.2656,
+      "step": 10767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0916189751164506e-07,
+      "loss": 0.3156,
+      "step": 10768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.08477284474481e-07,
+      "loss": 0.2507,
+      "step": 10769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.077934184120035e-07,
+      "loss": 0.2425,
+      "step": 10770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0711029937691704e-07,
+      "loss": 0.2481,
+      "step": 10771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.06427927421874e-07,
+      "loss": 0.2932,
+      "step": 10772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0574630259947e-07,
+      "loss": 0.2834,
+      "step": 10773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.050654249622398e-07,
+      "loss": 0.2822,
+      "step": 10774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0438529456266463e-07,
+      "loss": 0.2178,
+      "step": 10775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.037059114531604e-07,
+      "loss": 0.2725,
+      "step": 10776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0302727568609637e-07,
+      "loss": 0.3517,
+      "step": 10777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0234938731377394e-07,
+      "loss": 0.2649,
+      "step": 10778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.016722463884436e-07,
+      "loss": 0.2627,
+      "step": 10779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.009958529622958e-07,
+      "loss": 0.3549,
+      "step": 10780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0032020708746334e-07,
+      "loss": 0.308,
+      "step": 10781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.996453088160234e-07,
+      "loss": 0.2462,
+      "step": 10782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.989711581999899e-07,
+      "loss": 0.2692,
+      "step": 10783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.982977552913269e-07,
+      "loss": 0.2842,
+      "step": 10784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.97625100141935e-07,
+      "loss": 0.2784,
+      "step": 10785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.969531928036595e-07,
+      "loss": 0.2428,
+      "step": 10786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9628203332828675e-07,
+      "loss": 0.33,
+      "step": 10787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9561162176754863e-07,
+      "loss": 0.2586,
+      "step": 10788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.949419581731161e-07,
+      "loss": 0.2617,
+      "step": 10789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9427304259660117e-07,
+      "loss": 0.3091,
+      "step": 10790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9360487508956594e-07,
+      "loss": 0.275,
+      "step": 10791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9293745570350365e-07,
+      "loss": 0.2443,
+      "step": 10792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.922707844898598e-07,
+      "loss": 0.2662,
+      "step": 10793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9160486150001556e-07,
+      "loss": 0.2491,
+      "step": 10794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.909396867852987e-07,
+      "loss": 0.2719,
+      "step": 10795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9027526039697717e-07,
+      "loss": 0.2585,
+      "step": 10796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8961158238625997e-07,
+      "loss": 0.2435,
+      "step": 10797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.889486528043028e-07,
+      "loss": 0.3191,
+      "step": 10798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8828647170219937e-07,
+      "loss": 0.2316,
+      "step": 10799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.876250391309876e-07,
+      "loss": 0.2142,
+      "step": 10800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.869643551416479e-07,
+      "loss": 0.2639,
+      "step": 10801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.863044197851017e-07,
+      "loss": 0.2616,
+      "step": 10802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.85645233112215e-07,
+      "loss": 0.2632,
+      "step": 10803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8498679517379277e-07,
+      "loss": 0.2493,
+      "step": 10804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.843291060205855e-07,
+      "loss": 0.2602,
+      "step": 10805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.836721657032848e-07,
+      "loss": 0.291,
+      "step": 10806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8301597427252137e-07,
+      "loss": 0.2439,
+      "step": 10807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.823605317788769e-07,
+      "loss": 0.2838,
+      "step": 10808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8170583827286435e-07,
+      "loss": 0.2596,
+      "step": 10809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.810518938049478e-07,
+      "loss": 0.2753,
+      "step": 10810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8039869842552583e-07,
+      "loss": 0.2918,
+      "step": 10811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.797462521849481e-07,
+      "loss": 0.3076,
+      "step": 10812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.790945551335e-07,
+      "loss": 0.2723,
+      "step": 10813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.784436073214103e-07,
+      "loss": 0.2469,
+      "step": 10814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.777934087988532e-07,
+      "loss": 0.3251,
+      "step": 10815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.771439596159409e-07,
+      "loss": 0.2961,
+      "step": 10816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7649525982272996e-07,
+      "loss": 0.5551,
+      "step": 10817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7584730946921825e-07,
+      "loss": 0.2514,
+      "step": 10818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.75200108605348e-07,
+      "loss": 0.3074,
+      "step": 10819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.745536572810026e-07,
+      "loss": 0.3062,
+      "step": 10820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.739079555460056e-07,
+      "loss": 0.2678,
+      "step": 10821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.73263003450126e-07,
+      "loss": 0.286,
+      "step": 10822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.726188010430719e-07,
+      "loss": 0.3054,
+      "step": 10823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.719753483744969e-07,
+      "loss": 0.294,
+      "step": 10824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7133264549399464e-07,
+      "loss": 0.2742,
+      "step": 10825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.70690692451101e-07,
+      "loss": 0.2859,
+      "step": 10826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.700494892952954e-07,
+      "loss": 0.2513,
+      "step": 10827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.69409036075996e-07,
+      "loss": 0.2698,
+      "step": 10828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.687693328425711e-07,
+      "loss": 0.2748,
+      "step": 10829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.681303796443202e-07,
+      "loss": 0.3055,
+      "step": 10830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6749217653049385e-07,
+      "loss": 0.2891,
+      "step": 10831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6685472355028053e-07,
+      "loss": 0.2393,
+      "step": 10832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.66218020752812e-07,
+      "loss": 0.249,
+      "step": 10833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.655820681871635e-07,
+      "loss": 0.2439,
+      "step": 10834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6494686590234797e-07,
+      "loss": 0.285,
+      "step": 10835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6431241394732856e-07,
+      "loss": 0.2506,
+      "step": 10836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.636787123710016e-07,
+      "loss": 0.2686,
+      "step": 10837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6304576122221035e-07,
+      "loss": 0.2657,
+      "step": 10838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.624135605497402e-07,
+      "loss": 0.5769,
+      "step": 10839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.617821104023177e-07,
+      "loss": 0.2706,
+      "step": 10840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6115141082861396e-07,
+      "loss": 0.2601,
+      "step": 10841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.605214618772356e-07,
+      "loss": 0.2571,
+      "step": 10842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.598922635967416e-07,
+      "loss": 0.2645,
+      "step": 10843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.592638160356231e-07,
+      "loss": 0.2885,
+      "step": 10844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.586361192423181e-07,
+      "loss": 0.2585,
+      "step": 10845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5800917326521013e-07,
+      "loss": 0.5839,
+      "step": 10846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.573829781526171e-07,
+      "loss": 0.2693,
+      "step": 10847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.56757533952805e-07,
+      "loss": 0.2687,
+      "step": 10848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.561328407139785e-07,
+      "loss": 0.286,
+      "step": 10849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.555088984842868e-07,
+      "loss": 0.2788,
+      "step": 10850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.548857073118216e-07,
+      "loss": 0.2451,
+      "step": 10851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5426326724461215e-07,
+      "loss": 0.2814,
+      "step": 10852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5364157833063676e-07,
+      "loss": 0.279,
+      "step": 10853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.530206406178104e-07,
+      "loss": 0.2783,
+      "step": 10854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.524004541539904e-07,
+      "loss": 0.272,
+      "step": 10855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.517810189869796e-07,
+      "loss": 0.2605,
+      "step": 10856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5116233516452094e-07,
+      "loss": 0.2546,
+      "step": 10857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.505444027342996e-07,
+      "loss": 0.2748,
+      "step": 10858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4992722174393966e-07,
+      "loss": 0.2784,
+      "step": 10859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.493107922410165e-07,
+      "loss": 0.2602,
+      "step": 10860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.486951142730354e-07,
+      "loss": 0.3102,
+      "step": 10861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.480801878874528e-07,
+      "loss": 0.2849,
+      "step": 10862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.474660131316642e-07,
+      "loss": 0.2665,
+      "step": 10863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.468525900530061e-07,
+      "loss": 0.2656,
+      "step": 10864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4623991869875965e-07,
+      "loss": 0.27,
+      "step": 10865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.456279991161437e-07,
+      "loss": 0.2473,
+      "step": 10866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.450168313523249e-07,
+      "loss": 0.277,
+      "step": 10867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.444064154544079e-07,
+      "loss": 0.2339,
+      "step": 10868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.437967514694406e-07,
+      "loss": 0.2688,
+      "step": 10869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4318783944441314e-07,
+      "loss": 0.3087,
+      "step": 10870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4257967942625694e-07,
+      "loss": 0.3168,
+      "step": 10871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4197227146184664e-07,
+      "loss": 0.2666,
+      "step": 10872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4136561559799597e-07,
+      "loss": 0.2789,
+      "step": 10873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4075971188146754e-07,
+      "loss": 0.2799,
+      "step": 10874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.401545603589572e-07,
+      "loss": 0.2864,
+      "step": 10875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3955016107710896e-07,
+      "loss": 0.2412,
+      "step": 10876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3894651408250536e-07,
+      "loss": 0.2651,
+      "step": 10877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3834361942167484e-07,
+      "loss": 0.2861,
+      "step": 10878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3774147714108463e-07,
+      "loss": 0.2487,
+      "step": 10879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.371400872871432e-07,
+      "loss": 0.2917,
+      "step": 10880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.365394499062057e-07,
+      "loss": 0.2616,
+      "step": 10881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3593956504456396e-07,
+      "loss": 0.2626,
+      "step": 10882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.353404327484543e-07,
+      "loss": 0.2569,
+      "step": 10883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.347420530640565e-07,
+      "loss": 0.2606,
+      "step": 10884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3414442603748922e-07,
+      "loss": 0.3099,
+      "step": 10885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.335475517148167e-07,
+      "loss": 0.3089,
+      "step": 10886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.329514301420388e-07,
+      "loss": 0.3102,
+      "step": 10887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3235606136510545e-07,
+      "loss": 0.2842,
+      "step": 10888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3176144542990443e-07,
+      "loss": 0.2803,
+      "step": 10889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3116758238226233e-07,
+      "loss": 0.2562,
+      "step": 10890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3057447226795705e-07,
+      "loss": 0.2645,
+      "step": 10891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2998211513269753e-07,
+      "loss": 0.2889,
+      "step": 10892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.293905110221406e-07,
+      "loss": 0.2865,
+      "step": 10893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2879965998188646e-07,
+      "loss": 0.259,
+      "step": 10894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2820956205747312e-07,
+      "loss": 0.2379,
+      "step": 10895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2762021729438423e-07,
+      "loss": 0.2462,
+      "step": 10896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2703162573804006e-07,
+      "loss": 0.2617,
+      "step": 10897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.264437874338099e-07,
+      "loss": 0.2662,
+      "step": 10898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2585670242699975e-07,
+      "loss": 0.2913,
+      "step": 10899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2527037076286008e-07,
+      "loss": 0.2559,
+      "step": 10900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2468479248658026e-07,
+      "loss": 0.3108,
+      "step": 10901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2409996764329644e-07,
+      "loss": 0.3046,
+      "step": 10902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.235158962780837e-07,
+      "loss": 0.2772,
+      "step": 10903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2293257843595706e-07,
+      "loss": 0.2725,
+      "step": 10904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.223500141618795e-07,
+      "loss": 0.2646,
+      "step": 10905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2176820350074846e-07,
+      "loss": 0.2773,
+      "step": 10906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2118714649740912e-07,
+      "loss": 0.2639,
+      "step": 10907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.206068431966446e-07,
+      "loss": 0.2736,
+      "step": 10908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2002729364318464e-07,
+      "loss": 0.226,
+      "step": 10909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1944849788169798e-07,
+      "loss": 0.2361,
+      "step": 10910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1887045595679112e-07,
+      "loss": 0.2853,
+      "step": 10911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.182931679130218e-07,
+      "loss": 0.2546,
+      "step": 10912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1771663379488106e-07,
+      "loss": 0.2852,
+      "step": 10913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1714085364680671e-07,
+      "loss": 0.2612,
+      "step": 10914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1656582751317657e-07,
+      "loss": 0.2672,
+      "step": 10915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1599155543831074e-07,
+      "loss": 0.261,
+      "step": 10916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1541803746647272e-07,
+      "loss": 0.3041,
+      "step": 10917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1484527364186492e-07,
+      "loss": 0.2885,
+      "step": 10918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1427326400863424e-07,
+      "loss": 0.2695,
+      "step": 10919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1370200861086655e-07,
+      "loss": 0.2617,
+      "step": 10920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1313150749259216e-07,
+      "loss": 0.263,
+      "step": 10921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1256176069778367e-07,
+      "loss": 0.2766,
+      "step": 10922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1199276827035374e-07,
+      "loss": 0.2861,
+      "step": 10923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1142453025415734e-07,
+      "loss": 0.2974,
+      "step": 10924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1085704669299045e-07,
+      "loss": 0.2552,
+      "step": 10925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.102903176305926e-07,
+      "loss": 0.2514,
+      "step": 10926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.097243431106466e-07,
+      "loss": 0.3233,
+      "step": 10927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.091591231767709e-07,
+      "loss": 0.2919,
+      "step": 10928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0859465787253396e-07,
+      "loss": 0.263,
+      "step": 10929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0803094724143879e-07,
+      "loss": 0.2747,
+      "step": 10930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0746799132693506e-07,
+      "loss": 0.2407,
+      "step": 10931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.069057901724114e-07,
+      "loss": 0.2614,
+      "step": 10932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0634434382120205e-07,
+      "loss": 0.264,
+      "step": 10933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0578365231657792e-07,
+      "loss": 0.2674,
+      "step": 10934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0522371570175447e-07,
+      "loss": 0.2305,
+      "step": 10935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.046645340198905e-07,
+      "loss": 0.2805,
+      "step": 10936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0410610731408377e-07,
+      "loss": 0.2699,
+      "step": 10937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0354843562737537e-07,
+      "loss": 0.217,
+      "step": 10938
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0299151900274873e-07,
+      "loss": 0.3044,
+      "step": 10939
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0243535748312615e-07,
+      "loss": 0.2817,
+      "step": 10940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.018799511113767e-07,
+      "loss": 0.3231,
+      "step": 10941
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0132529993030392e-07,
+      "loss": 0.2599,
+      "step": 10942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0077140398266248e-07,
+      "loss": 0.2949,
+      "step": 10943
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.002182633111416e-07,
+      "loss": 0.6014,
+      "step": 10944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9966587795837377e-07,
+      "loss": 0.2759,
+      "step": 10945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9911424796693611e-07,
+      "loss": 0.3049,
+      "step": 10946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.985633733793446e-07,
+      "loss": 0.2601,
+      "step": 10947
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9801325423805862e-07,
+      "loss": 0.3292,
+      "step": 10948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9746389058547534e-07,
+      "loss": 0.2265,
+      "step": 10949
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9691528246394197e-07,
+      "loss": 0.5916,
+      "step": 10950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.963674299157403e-07,
+      "loss": 0.2798,
+      "step": 10951
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9582033298309434e-07,
+      "loss": 0.2923,
+      "step": 10952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9527399170817473e-07,
+      "loss": 0.2465,
+      "step": 10953
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9472840613308787e-07,
+      "loss": 0.2719,
+      "step": 10954
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9418357629988782e-07,
+      "loss": 0.3484,
+      "step": 10955
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.936395022505644e-07,
+      "loss": 0.2772,
+      "step": 10956
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.930961840270551e-07,
+      "loss": 0.2726,
+      "step": 10957
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9255362167123316e-07,
+      "loss": 0.2411,
+      "step": 10958
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.920118152249173e-07,
+      "loss": 0.264,
+      "step": 10959
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.914707647298697e-07,
+      "loss": 0.3056,
+      "step": 10960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.909304702277903e-07,
+      "loss": 0.2653,
+      "step": 10961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.903909317603214e-07,
+      "loss": 0.2652,
+      "step": 10962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.898521493690486e-07,
+      "loss": 0.2892,
+      "step": 10963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8931412309549867e-07,
+      "loss": 0.2734,
+      "step": 10964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8877685298114178e-07,
+      "loss": 0.2504,
+      "step": 10965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.882403390673837e-07,
+      "loss": 0.262,
+      "step": 10966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8770458139558134e-07,
+      "loss": 0.2695,
+      "step": 10967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.87169580007025e-07,
+      "loss": 0.2392,
+      "step": 10968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.866353349429506e-07,
+      "loss": 0.2555,
+      "step": 10969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.861018462445352e-07,
+      "loss": 0.2729,
+      "step": 10970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8556911395289811e-07,
+      "loss": 0.2485,
+      "step": 10971
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8503713810909984e-07,
+      "loss": 0.2606,
+      "step": 10972
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8450591875413981e-07,
+      "loss": 0.3301,
+      "step": 10973
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8397545592896527e-07,
+      "loss": 0.2673,
+      "step": 10974
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8344574967446015e-07,
+      "loss": 0.2759,
+      "step": 10975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8291680003145074e-07,
+      "loss": 0.2781,
+      "step": 10976
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.823886070407077e-07,
+      "loss": 0.2952,
+      "step": 10977
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8186117074293964e-07,
+      "loss": 0.3367,
+      "step": 10978
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8133449117880064e-07,
+      "loss": 0.2573,
+      "step": 10979
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.808085683888827e-07,
+      "loss": 0.2839,
+      "step": 10980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8028340241372345e-07,
+      "loss": 0.5588,
+      "step": 10981
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.797589932937982e-07,
+      "loss": 0.2917,
+      "step": 10982
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.792353410695269e-07,
+      "loss": 0.2467,
+      "step": 10983
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7871244578126835e-07,
+      "loss": 0.2821,
+      "step": 10984
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7819030746932696e-07,
+      "loss": 0.3131,
+      "step": 10985
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7766892617394727e-07,
+      "loss": 0.2578,
+      "step": 10986
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.771483019353104e-07,
+      "loss": 0.2864,
+      "step": 10987
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7662843479354874e-07,
+      "loss": 0.2723,
+      "step": 10988
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.761093247887269e-07,
+      "loss": 0.2858,
+      "step": 10989
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.755909719608573e-07,
+      "loss": 0.3066,
+      "step": 10990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.750733763498924e-07,
+      "loss": 0.2475,
+      "step": 10991
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.745565379957248e-07,
+      "loss": 0.2643,
+      "step": 10992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7404045693819037e-07,
+      "loss": 0.2493,
+      "step": 10993
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7352513321706621e-07,
+      "loss": 0.303,
+      "step": 10994
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7301056687207053e-07,
+      "loss": 0.2623,
+      "step": 10995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.724967579428638e-07,
+      "loss": 0.2954,
+      "step": 10996
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7198370646904773e-07,
+      "loss": 0.3277,
+      "step": 10997
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.714714124901662e-07,
+      "loss": 0.2625,
+      "step": 10998
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.709598760457043e-07,
+      "loss": 0.2752,
+      "step": 10999
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7044909717508828e-07,
+      "loss": 0.3015,
+      "step": 11000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6993907591768556e-07,
+      "loss": 0.2481,
+      "step": 11001
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6942981231280798e-07,
+      "loss": 0.3054,
+      "step": 11002
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6892130639970638e-07,
+      "loss": 0.2666,
+      "step": 11003
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6841355821757277e-07,
+      "loss": 0.2927,
+      "step": 11004
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.679065678055447e-07,
+      "loss": 0.2709,
+      "step": 11005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6740033520269538e-07,
+      "loss": 0.2939,
+      "step": 11006
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.668948604480436e-07,
+      "loss": 0.2606,
+      "step": 11007
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6639014358054927e-07,
+      "loss": 0.2576,
+      "step": 11008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6588618463911356e-07,
+      "loss": 0.2776,
+      "step": 11009
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6538298366257975e-07,
+      "loss": 0.2117,
+      "step": 11010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6488054068972914e-07,
+      "loss": 0.5712,
+      "step": 11011
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.643788557592918e-07,
+      "loss": 0.3251,
+      "step": 11012
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6387792890993238e-07,
+      "loss": 0.2996,
+      "step": 11013
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6337776018026108e-07,
+      "loss": 0.2424,
+      "step": 11014
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.628783496088271e-07,
+      "loss": 0.272,
+      "step": 11015
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6237969723412294e-07,
+      "loss": 0.265,
+      "step": 11016
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6188180309458345e-07,
+      "loss": 0.2539,
+      "step": 11017
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6138466722858237e-07,
+      "loss": 0.2629,
+      "step": 11018
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6088828967443793e-07,
+      "loss": 0.2788,
+      "step": 11019
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6039267047040728e-07,
+      "loss": 0.292,
+      "step": 11020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5989780965468994e-07,
+      "loss": 0.2114,
+      "step": 11021
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5940370726542864e-07,
+      "loss": 0.2867,
+      "step": 11022
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.589103633407052e-07,
+      "loss": 0.2458,
+      "step": 11023
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5841777791854584e-07,
+      "loss": 0.3018,
+      "step": 11024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5792595103691466e-07,
+      "loss": 0.307,
+      "step": 11025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5743488273372133e-07,
+      "loss": 0.2418,
+      "step": 11026
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5694457304681222e-07,
+      "loss": 0.3038,
+      "step": 11027
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5645502201397933e-07,
+      "loss": 0.259,
+      "step": 11028
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5596622967295584e-07,
+      "loss": 0.2399,
+      "step": 11029
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.554781960614138e-07,
+      "loss": 0.3043,
+      "step": 11030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.549909212169709e-07,
+      "loss": 0.2683,
+      "step": 11031
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5450440517717934e-07,
+      "loss": 0.2667,
+      "step": 11032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5401864797954248e-07,
+      "loss": 0.2121,
+      "step": 11033
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5353364966149697e-07,
+      "loss": 0.2802,
+      "step": 11034
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5304941026042408e-07,
+      "loss": 0.2393,
+      "step": 11035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5256592981364947e-07,
+      "loss": 0.265,
+      "step": 11036
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.520832083584345e-07,
+      "loss": 0.2571,
+      "step": 11037
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.51601245931986e-07,
+      "loss": 0.2621,
+      "step": 11038
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5112004257144986e-07,
+      "loss": 0.3093,
+      "step": 11039
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.506395983139175e-07,
+      "loss": 0.3196,
+      "step": 11040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.501599131964182e-07,
+      "loss": 0.2657,
+      "step": 11041
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4968098725592127e-07,
+      "loss": 0.3131,
+      "step": 11042
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4920282052934387e-07,
+      "loss": 0.2822,
+      "step": 11043
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.487254130535376e-07,
+      "loss": 0.2692,
+      "step": 11044
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.482487648653008e-07,
+      "loss": 0.2626,
+      "step": 11045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.477728760013697e-07,
+      "loss": 0.2756,
+      "step": 11046
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4729774649842376e-07,
+      "loss": 0.2456,
+      "step": 11047
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4682337639308486e-07,
+      "loss": 0.2608,
+      "step": 11048
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4634976572191372e-07,
+      "loss": 0.2312,
+      "step": 11049
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.458769145214145e-07,
+      "loss": 0.2748,
+      "step": 11050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4540482282803136e-07,
+      "loss": 0.2961,
+      "step": 11051
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4493349067815188e-07,
+      "loss": 0.2699,
+      "step": 11052
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4446291810810365e-07,
+      "loss": 0.2286,
+      "step": 11053
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4399310515415655e-07,
+      "loss": 0.2567,
+      "step": 11054
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4352405185252048e-07,
+      "loss": 0.3259,
+      "step": 11055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.430557582393477e-07,
+      "loss": 0.2772,
+      "step": 11056
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.425882243507337e-07,
+      "loss": 0.2618,
+      "step": 11057
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4212145022271196e-07,
+      "loss": 0.2341,
+      "step": 11058
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4165543589125918e-07,
+      "loss": 0.2669,
+      "step": 11059
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4119018139229333e-07,
+      "loss": 0.3201,
+      "step": 11060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4072568676167575e-07,
+      "loss": 0.2262,
+      "step": 11061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4026195203520666e-07,
+      "loss": 0.2884,
+      "step": 11062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3979897724862523e-07,
+      "loss": 0.2607,
+      "step": 11063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3933676243762072e-07,
+      "loss": 0.2772,
+      "step": 11064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3887530763781465e-07,
+      "loss": 0.2661,
+      "step": 11065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.384146128847741e-07,
+      "loss": 0.3054,
+      "step": 11066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3795467821400842e-07,
+      "loss": 0.2769,
+      "step": 11067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.37495503660966e-07,
+      "loss": 0.2525,
+      "step": 11068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3703708926103842e-07,
+      "loss": 0.2617,
+      "step": 11069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.365794350495564e-07,
+      "loss": 0.309,
+      "step": 11070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3612254106179723e-07,
+      "loss": 0.251,
+      "step": 11071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3566640733297166e-07,
+      "loss": 0.3071,
+      "step": 11072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.352110338982382e-07,
+      "loss": 0.2515,
+      "step": 11073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3475642079269659e-07,
+      "loss": 0.2426,
+      "step": 11074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.343025680513832e-07,
+      "loss": 0.3289,
+      "step": 11075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.338494757092812e-07,
+      "loss": 0.2855,
+      "step": 11076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.333971438013104e-07,
+      "loss": 0.2477,
+      "step": 11077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.329455723623352e-07,
+      "loss": 0.2708,
+      "step": 11078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.324947614271621e-07,
+      "loss": 0.2846,
+      "step": 11079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.320447110305345e-07,
+      "loss": 0.2825,
+      "step": 11080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3159542120714352e-07,
+      "loss": 0.2745,
+      "step": 11081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3114689199161478e-07,
+      "loss": 0.2184,
+      "step": 11082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3069912341852064e-07,
+      "loss": 0.2872,
+      "step": 11083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3025211552237127e-07,
+      "loss": 0.2186,
+      "step": 11084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2980586833762242e-07,
+      "loss": 0.3228,
+      "step": 11085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2936038189866773e-07,
+      "loss": 0.2629,
+      "step": 11086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.289156562398408e-07,
+      "loss": 0.2468,
+      "step": 11087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2847169139542204e-07,
+      "loss": 0.2712,
+      "step": 11088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2802848739962737e-07,
+      "loss": 0.2634,
+      "step": 11089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2758604428661836e-07,
+      "loss": 0.2529,
+      "step": 11090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2714436209049664e-07,
+      "loss": 0.2556,
+      "step": 11091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2670344084530384e-07,
+      "loss": 0.2754,
+      "step": 11092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2626328058502502e-07,
+      "loss": 0.271,
+      "step": 11093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2582388134358414e-07,
+      "loss": 0.2536,
+      "step": 11094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2538524315484968e-07,
+      "loss": 0.2448,
+      "step": 11095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2494736605262792e-07,
+      "loss": 0.2592,
+      "step": 11096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2451025007066963e-07,
+      "loss": 0.2485,
+      "step": 11097
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2407389524266456e-07,
+      "loss": 0.2639,
+      "step": 11098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2363830160224465e-07,
+      "loss": 0.3061,
+      "step": 11099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2320346918298644e-07,
+      "loss": 0.2926,
+      "step": 11100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2276939801839972e-07,
+      "loss": 0.3254,
+      "step": 11101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.223360881419433e-07,
+      "loss": 0.27,
+      "step": 11102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2190353958701495e-07,
+      "loss": 0.2877,
+      "step": 11103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.214717523869524e-07,
+      "loss": 0.2866,
+      "step": 11104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2104072657503573e-07,
+      "loss": 0.2492,
+      "step": 11105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2061046218448724e-07,
+      "loss": 0.221,
+      "step": 11106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.201809592484682e-07,
+      "loss": 0.258,
+      "step": 11107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1975221780008317e-07,
+      "loss": 0.2326,
+      "step": 11108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.19324237872378e-07,
+      "loss": 0.2755,
+      "step": 11109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1889701949833743e-07,
+      "loss": 0.3179,
+      "step": 11110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1847056271089174e-07,
+      "loss": 0.2632,
+      "step": 11111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1804486754290912e-07,
+      "loss": 0.2695,
+      "step": 11112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1761993402719884e-07,
+      "loss": 0.2848,
+      "step": 11113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1719576219651585e-07,
+      "loss": 0.2698,
+      "step": 11114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1677235208354842e-07,
+      "loss": 0.2657,
+      "step": 11115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.163497037209349e-07,
+      "loss": 0.2778,
+      "step": 11116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1592781714125034e-07,
+      "loss": 0.2684,
+      "step": 11117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1550669237700985e-07,
+      "loss": 0.256,
+      "step": 11118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.150863294606741e-07,
+      "loss": 0.3174,
+      "step": 11119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1466672842464055e-07,
+      "loss": 0.3174,
+      "step": 11120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1424788930125108e-07,
+      "loss": 0.598,
+      "step": 11121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1382981212278655e-07,
+      "loss": 0.2747,
+      "step": 11122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1341249692147116e-07,
+      "loss": 0.3002,
+      "step": 11123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1299594372947031e-07,
+      "loss": 0.286,
+      "step": 11124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1258015257888832e-07,
+      "loss": 0.2589,
+      "step": 11125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.12165123501774e-07,
+      "loss": 0.2772,
+      "step": 11126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1175085653011397e-07,
+      "loss": 0.2831,
+      "step": 11127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1133735169583826e-07,
+      "loss": 0.2903,
+      "step": 11128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1092460903081803e-07,
+      "loss": 0.2384,
+      "step": 11129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1051262856686673e-07,
+      "loss": 0.3017,
+      "step": 11130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1010141033573562e-07,
+      "loss": 0.2823,
+      "step": 11131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0969095436912047e-07,
+      "loss": 0.264,
+      "step": 11132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0928126069865819e-07,
+      "loss": 0.2909,
+      "step": 11133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0887232935592351e-07,
+      "loss": 0.2753,
+      "step": 11134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0846416037243678e-07,
+      "loss": 0.2318,
+      "step": 11135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.080567537796573e-07,
+      "loss": 0.324,
+      "step": 11136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0765010960898548e-07,
+      "loss": 0.2839,
+      "step": 11137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0724422789176404e-07,
+      "loss": 0.3337,
+      "step": 11138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.068391086592746e-07,
+      "loss": 0.2611,
+      "step": 11139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0643475194274444e-07,
+      "loss": 0.2267,
+      "step": 11140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.060311577733375e-07,
+      "loss": 0.2205,
+      "step": 11141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0562832618216223e-07,
+      "loss": 0.2562,
+      "step": 11142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0522625720026491e-07,
+      "loss": 0.2975,
+      "step": 11143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0482495085863631e-07,
+      "loss": 0.2576,
+      "step": 11144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0442440718820834e-07,
+      "loss": 0.308,
+      "step": 11145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0402462621984965e-07,
+      "loss": 0.2617,
+      "step": 11146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0362560798437671e-07,
+      "loss": 0.2844,
+      "step": 11147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0322735251254156e-07,
+      "loss": 0.5356,
+      "step": 11148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.028298598350408e-07,
+      "loss": 0.2559,
+      "step": 11149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0243312998251209e-07,
+      "loss": 0.2758,
+      "step": 11150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0203716298553212e-07,
+      "loss": 0.2247,
+      "step": 11151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.016419588746198e-07,
+      "loss": 0.2969,
+      "step": 11152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0124751768023633e-07,
+      "loss": 0.2437,
+      "step": 11153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0085383943278293e-07,
+      "loss": 0.3262,
+      "step": 11154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0046092416260312e-07,
+      "loss": 0.2685,
+      "step": 11155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0006877189997821e-07,
+      "loss": 0.2482,
+      "step": 11156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.967738267513737e-08,
+      "loss": 0.2437,
+      "step": 11157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.928675651824427e-08,
+      "loss": 0.284,
+      "step": 11158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.889689345940812e-08,
+      "loss": 0.2286,
+      "step": 11159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.85077935286749e-08,
+      "loss": 0.2334,
+      "step": 11160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.811945675603729e-08,
+      "loss": 0.3089,
+      "step": 11161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.773188317142579e-08,
+      "loss": 0.2933,
+      "step": 11162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.734507280471094e-08,
+      "loss": 0.2717,
+      "step": 11163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.69590256857078e-08,
+      "loss": 0.2385,
+      "step": 11164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.657374184417146e-08,
+      "loss": 0.22,
+      "step": 11165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.61892213097959e-08,
+      "loss": 0.2663,
+      "step": 11166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.580546411221858e-08,
+      "loss": 0.2419,
+      "step": 11167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.542247028101914e-08,
+      "loss": 0.2545,
+      "step": 11168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.50402398457162e-08,
+      "loss": 0.2546,
+      "step": 11169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.46587728357673e-08,
+      "loss": 0.2878,
+      "step": 11170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.427806928057893e-08,
+      "loss": 0.3085,
+      "step": 11171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.389812920949093e-08,
+      "loss": 0.2669,
+      "step": 11172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.351895265178656e-08,
+      "loss": 0.5581,
+      "step": 11173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.314053963669245e-08,
+      "loss": 0.2675,
+      "step": 11174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.276289019337415e-08,
+      "loss": 0.2682,
+      "step": 11175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.238600435094058e-08,
+      "loss": 0.2626,
+      "step": 11176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.200988213843631e-08,
+      "loss": 0.2632,
+      "step": 11177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.163452358485591e-08,
+      "loss": 0.2552,
+      "step": 11178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.125992871912626e-08,
+      "loss": 0.2536,
+      "step": 11179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.088609757012201e-08,
+      "loss": 0.2545,
+      "step": 11180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.051303016665347e-08,
+      "loss": 0.2369,
+      "step": 11181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.014072653747763e-08,
+      "loss": 0.2917,
+      "step": 11182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.97691867112882e-08,
+      "loss": 0.2599,
+      "step": 11183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.939841071672117e-08,
+      "loss": 0.249,
+      "step": 11184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.90283985823559e-08,
+      "loss": 0.3525,
+      "step": 11185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.865915033671069e-08,
+      "loss": 0.2525,
+      "step": 11186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.829066600824277e-08,
+      "loss": 0.3111,
+      "step": 11187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.79229456253572e-08,
+      "loss": 0.266,
+      "step": 11188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.755598921639241e-08,
+      "loss": 0.2385,
+      "step": 11189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.718979680963469e-08,
+      "loss": 0.3097,
+      "step": 11190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.682436843330477e-08,
+      "loss": 0.3012,
+      "step": 11191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.645970411557125e-08,
+      "loss": 0.2763,
+      "step": 11192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.609580388454052e-08,
+      "loss": 0.2753,
+      "step": 11193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.573266776825683e-08,
+      "loss": 0.2708,
+      "step": 11194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.537029579471334e-08,
+      "loss": 0.2969,
+      "step": 11195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.50086879918366e-08,
+      "loss": 0.2747,
+      "step": 11196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.464784438749985e-08,
+      "loss": 0.3044,
+      "step": 11197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.428776500951308e-08,
+      "loss": 0.2467,
+      "step": 11198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.392844988563075e-08,
+      "loss": 0.284,
+      "step": 11199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.35698990435474e-08,
+      "loss": 0.3075,
+      "step": 11200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.321211251089645e-08,
+      "loss": 0.2837,
+      "step": 11201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.285509031525696e-08,
+      "loss": 0.2766,
+      "step": 11202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.249883248414359e-08,
+      "loss": 0.2557,
+      "step": 11203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.21433390450177e-08,
+      "loss": 0.2319,
+      "step": 11204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.178861002527628e-08,
+      "loss": 0.5624,
+      "step": 11205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.143464545226298e-08,
+      "loss": 0.2836,
+      "step": 11206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.108144535325713e-08,
+      "loss": 0.2432,
+      "step": 11207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.072900975548248e-08,
+      "loss": 0.2726,
+      "step": 11208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.037733868610509e-08,
+      "loss": 0.3044,
+      "step": 11209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.002643217222661e-08,
+      "loss": 0.2598,
+      "step": 11210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.967629024089429e-08,
+      "loss": 0.3063,
+      "step": 11211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.932691291909656e-08,
+      "loss": 0.2604,
+      "step": 11212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.897830023376074e-08,
+      "loss": 0.2823,
+      "step": 11213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.863045221175647e-08,
+      "loss": 0.2502,
+      "step": 11214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.82833688798934e-08,
+      "loss": 0.304,
+      "step": 11215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.793705026492459e-08,
+      "loss": 0.2799,
+      "step": 11216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.759149639354091e-08,
+      "loss": 0.3023,
+      "step": 11217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.72467072923766e-08,
+      "loss": 0.2812,
+      "step": 11218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.690268298800596e-08,
+      "loss": 0.2649,
+      "step": 11219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.655942350694556e-08,
+      "loss": 0.2226,
+      "step": 11220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.621692887565202e-08,
+      "loss": 0.2224,
+      "step": 11221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.587519912052199e-08,
+      "loss": 0.2657,
+      "step": 11222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.553423426789664e-08,
+      "loss": 0.2788,
+      "step": 11223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.51940343440527e-08,
+      "loss": 0.2589,
+      "step": 11224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.485459937521256e-08,
+      "loss": 0.2335,
+      "step": 11225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.451592938753971e-08,
+      "loss": 0.2879,
+      "step": 11226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.417802440713439e-08,
+      "loss": 0.2693,
+      "step": 11227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.384088446004356e-08,
+      "loss": 0.2811,
+      "step": 11228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.350450957224864e-08,
+      "loss": 0.2823,
+      "step": 11229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.31688997696789e-08,
+      "loss": 0.2487,
+      "step": 11230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.283405507820141e-08,
+      "loss": 0.2932,
+      "step": 11231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.249997552362109e-08,
+      "loss": 0.2537,
+      "step": 11232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.21666611316918e-08,
+      "loss": 0.2381,
+      "step": 11233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.183411192810075e-08,
+      "loss": 0.24,
+      "step": 11234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.150232793847967e-08,
+      "loss": 0.2476,
+      "step": 11235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.117130918840032e-08,
+      "loss": 0.303,
+      "step": 11236
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.084105570337785e-08,
+      "loss": 0.2562,
+      "step": 11237
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.051156750886523e-08,
+      "loss": 0.5919,
+      "step": 11238
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.01828446302566e-08,
+      "loss": 0.366,
+      "step": 11239
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.985488709289057e-08,
+      "loss": 0.2456,
+      "step": 11240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.952769492204359e-08,
+      "loss": 0.2977,
+      "step": 11241
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.920126814293438e-08,
+      "loss": 0.274,
+      "step": 11242
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.887560678072169e-08,
+      "loss": 0.265,
+      "step": 11243
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.855071086050547e-08,
+      "loss": 0.2159,
+      "step": 11244
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.822658040732899e-08,
+      "loss": 0.2888,
+      "step": 11245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.790321544617117e-08,
+      "loss": 0.2625,
+      "step": 11246
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.758061600195986e-08,
+      "loss": 0.2703,
+      "step": 11247
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.725878209955628e-08,
+      "loss": 0.3077,
+      "step": 11248
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.693771376376612e-08,
+      "loss": 0.3311,
+      "step": 11249
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.661741101933628e-08,
+      "loss": 0.2686,
+      "step": 11250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.629787389095476e-08,
+      "loss": 0.2845,
+      "step": 11251
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.597910240324967e-08,
+      "loss": 0.3286,
+      "step": 11252
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.566109658078912e-08,
+      "loss": 0.2987,
+      "step": 11253
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.534385644808461e-08,
+      "loss": 0.2812,
+      "step": 11254
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.502738202958658e-08,
+      "loss": 0.3301,
+      "step": 11255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.471167334968887e-08,
+      "loss": 0.2943,
+      "step": 11256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.439673043272199e-08,
+      "loss": 0.2758,
+      "step": 11257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.40825533029632e-08,
+      "loss": 0.2596,
+      "step": 11258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.376914198462648e-08,
+      "loss": 0.2995,
+      "step": 11259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.345649650186691e-08,
+      "loss": 0.2594,
+      "step": 11260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.314461687878415e-08,
+      "loss": 0.2642,
+      "step": 11261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.28335031394134e-08,
+      "loss": 0.2514,
+      "step": 11262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.252315530773545e-08,
+      "loss": 0.2921,
+      "step": 11263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.22135734076712e-08,
+      "loss": 0.2561,
+      "step": 11264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.190475746307933e-08,
+      "loss": 0.2963,
+      "step": 11265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.159670749776414e-08,
+      "loss": 0.2861,
+      "step": 11266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.128942353546775e-08,
+      "loss": 0.263,
+      "step": 11267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.098290559987342e-08,
+      "loss": 0.282,
+      "step": 11268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.06771537146067e-08,
+      "loss": 0.2546,
+      "step": 11269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.037216790323319e-08,
+      "loss": 0.3136,
+      "step": 11270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.006794818926077e-08,
+      "loss": 0.2807,
+      "step": 11271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.976449459613509e-08,
+      "loss": 0.2397,
+      "step": 11272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.946180714724636e-08,
+      "loss": 0.2944,
+      "step": 11273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.915988586592481e-08,
+      "loss": 0.2356,
+      "step": 11274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8858730775438465e-08,
+      "loss": 0.2962,
+      "step": 11275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.855834189900211e-08,
+      "loss": 0.3015,
+      "step": 11276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8258719259765006e-08,
+      "loss": 0.2362,
+      "step": 11277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.795986288082422e-08,
+      "loss": 0.2847,
+      "step": 11278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7661772785211345e-08,
+      "loss": 0.2512,
+      "step": 11279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7364448995901324e-08,
+      "loss": 0.2693,
+      "step": 11280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.706789153581249e-08,
+      "loss": 0.2752,
+      "step": 11281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.677210042780212e-08,
+      "loss": 0.2554,
+      "step": 11282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.647707569466643e-08,
+      "loss": 0.2544,
+      "step": 11283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.618281735914499e-08,
+      "loss": 0.3107,
+      "step": 11284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5889325443918565e-08,
+      "loss": 0.2534,
+      "step": 11285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5596599971606823e-08,
+      "loss": 0.2894,
+      "step": 11286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.530464096477395e-08,
+      "loss": 0.2811,
+      "step": 11287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5013448445919716e-08,
+      "loss": 0.2862,
+      "step": 11288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4723022437489506e-08,
+      "loss": 0.2396,
+      "step": 11289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.443336296186874e-08,
+      "loss": 0.2404,
+      "step": 11290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.414447004138068e-08,
+      "loss": 0.2172,
+      "step": 11291
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3856343698294176e-08,
+      "loss": 0.3086,
+      "step": 11292
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3568983954813694e-08,
+      "loss": 0.2487,
+      "step": 11293
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3282390833090393e-08,
+      "loss": 0.2922,
+      "step": 11294
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.299656435521217e-08,
+      "loss": 0.2282,
+      "step": 11295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.271150454320917e-08,
+      "loss": 0.2324,
+      "step": 11296
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2427211419051605e-08,
+      "loss": 0.265,
+      "step": 11297
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.214368500465305e-08,
+      "loss": 0.278,
+      "step": 11298
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.186092532186493e-08,
+      "loss": 0.2397,
+      "step": 11299
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.157893239248202e-08,
+      "loss": 0.2766,
+      "step": 11300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.129770623823804e-08,
+      "loss": 0.2867,
+      "step": 11301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1017246880809e-08,
+      "loss": 0.3179,
+      "step": 11302
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.073755434181093e-08,
+      "loss": 0.2878,
+      "step": 11303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0458628642802156e-08,
+      "loss": 0.2625,
+      "step": 11304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.01804698052788e-08,
+      "loss": 0.3158,
+      "step": 11305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.99030778506826e-08,
+      "loss": 0.2543,
+      "step": 11306
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.962645280039202e-08,
+      "loss": 0.2661,
+      "step": 11307
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.935059467572778e-08,
+      "loss": 0.2993,
+      "step": 11308
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.907550349795287e-08,
+      "loss": 0.2518,
+      "step": 11309
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8801179288268105e-08,
+      "loss": 0.3093,
+      "step": 11310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.85276220678188e-08,
+      "loss": 0.2347,
+      "step": 11311
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.82548318576892e-08,
+      "loss": 0.2411,
+      "step": 11312
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.798280867890359e-08,
+      "loss": 0.319,
+      "step": 11313
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.771155255242854e-08,
+      "loss": 0.2524,
+      "step": 11314
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.744106349917066e-08,
+      "loss": 0.2753,
+      "step": 11315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.717134153997993e-08,
+      "loss": 0.573,
+      "step": 11316
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6902386695644174e-08,
+      "loss": 0.2864,
+      "step": 11317
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.663419898689125e-08,
+      "loss": 0.3143,
+      "step": 11318
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6366778434393524e-08,
+      "loss": 0.2847,
+      "step": 11319
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.610012505876338e-08,
+      "loss": 0.2456,
+      "step": 11320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.583423888055105e-08,
+      "loss": 0.2608,
+      "step": 11321
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.556911992025015e-08,
+      "loss": 0.2645,
+      "step": 11322
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.530476819829655e-08,
+      "loss": 0.2343,
+      "step": 11323
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.504118373506283e-08,
+      "loss": 0.2856,
+      "step": 11324
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.477836655086498e-08,
+      "loss": 0.2693,
+      "step": 11325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.451631666596123e-08,
+      "loss": 0.2758,
+      "step": 11326
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4255034100548766e-08,
+      "loss": 0.3412,
+      "step": 11327
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3994518874765914e-08,
+      "loss": 0.2557,
+      "step": 11328
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3734771008689947e-08,
+      "loss": 0.3047,
+      "step": 11329
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.347579052234374e-08,
+      "loss": 0.2413,
+      "step": 11330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3217577435686886e-08,
+      "loss": 0.2524,
+      "step": 11331
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.296013176862013e-08,
+      "loss": 0.3312,
+      "step": 11332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2703453540988704e-08,
+      "loss": 0.3398,
+      "step": 11333
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.244754277257346e-08,
+      "loss": 0.2469,
+      "step": 11334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.219239948310083e-08,
+      "loss": 0.2914,
+      "step": 11335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.193802369223399e-08,
+      "loss": 0.2906,
+      "step": 11336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.168441541958168e-08,
+      "loss": 0.2784,
+      "step": 11337
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.143157468468717e-08,
+      "loss": 0.2576,
+      "step": 11338
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.117950150704153e-08,
+      "loss": 0.2973,
+      "step": 11339
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.092819590607144e-08,
+      "loss": 0.2581,
+      "step": 11340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.067765790114697e-08,
+      "loss": 0.2953,
+      "step": 11341
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0427887511578224e-08,
+      "loss": 0.2675,
+      "step": 11342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.017888475661536e-08,
+      "loss": 0.2936,
+      "step": 11343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.99306496554519e-08,
+      "loss": 0.2623,
+      "step": 11344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.968318222722034e-08,
+      "loss": 0.3088,
+      "step": 11345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.943648249099319e-08,
+      "loss": 0.2405,
+      "step": 11346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9190550465785236e-08,
+      "loss": 0.2439,
+      "step": 11347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.894538617055243e-08,
+      "loss": 0.2798,
+      "step": 11348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8700989624189666e-08,
+      "loss": 0.2879,
+      "step": 11349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.845736084553408e-08,
+      "loss": 0.265,
+      "step": 11350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8214499853364007e-08,
+      "loss": 0.2798,
+      "step": 11351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.79724066663989e-08,
+      "loss": 0.2814,
+      "step": 11352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.773108130329495e-08,
+      "loss": 0.2737,
+      "step": 11353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.749052378265505e-08,
+      "loss": 0.2594,
+      "step": 11354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.725073412301994e-08,
+      "loss": 0.2728,
+      "step": 11355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7011712342870376e-08,
+      "loss": 0.2447,
+      "step": 11356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.677345846062941e-08,
+      "loss": 0.2601,
+      "step": 11357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.653597249466012e-08,
+      "loss": 0.2836,
+      "step": 11358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6299254463267877e-08,
+      "loss": 0.2546,
+      "step": 11359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.606330438469585e-08,
+      "loss": 0.2912,
+      "step": 11360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5828122277132836e-08,
+      "loss": 0.2618,
+      "step": 11361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.559370815870211e-08,
+      "loss": 0.2906,
+      "step": 11362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.536006204747366e-08,
+      "loss": 0.2775,
+      "step": 11363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5127183961454204e-08,
+      "loss": 0.2874,
+      "step": 11364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4895073918593814e-08,
+      "loss": 0.2669,
+      "step": 11365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.466373193678263e-08,
+      "loss": 0.2559,
+      "step": 11366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.443315803385083e-08,
+      "loss": 0.2654,
+      "step": 11367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4203352227569766e-08,
+      "loss": 0.2661,
+      "step": 11368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.397431453565192e-08,
+      "loss": 0.2279,
+      "step": 11369
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3746044975749845e-08,
+      "loss": 0.2541,
+      "step": 11370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.351854356545725e-08,
+      "loss": 0.2838,
+      "step": 11371
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3291810322311214e-08,
+      "loss": 0.2454,
+      "step": 11372
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.306584526378442e-08,
+      "loss": 0.2866,
+      "step": 11373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.284064840729406e-08,
+      "loss": 0.3113,
+      "step": 11374
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.261621977019846e-08,
+      "loss": 0.2747,
+      "step": 11375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.239255936979269e-08,
+      "loss": 0.278,
+      "step": 11376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.216966722331849e-08,
+      "loss": 0.2897,
+      "step": 11377
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1947543347953246e-08,
+      "loss": 0.3158,
+      "step": 11378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1726187760817704e-08,
+      "loss": 0.2634,
+      "step": 11379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1505600478973775e-08,
+      "loss": 0.3062,
+      "step": 11380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.128578151942119e-08,
+      "loss": 0.2733,
+      "step": 11381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.106673089910417e-08,
+      "loss": 0.2527,
+      "step": 11382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0848448634905884e-08,
+      "loss": 0.3233,
+      "step": 11383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.063093474364953e-08,
+      "loss": 0.319,
+      "step": 11384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.04141892421006e-08,
+      "loss": 0.265,
+      "step": 11385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.019821214696572e-08,
+      "loss": 0.281,
+      "step": 11386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.998300347488936e-08,
+      "loss": 0.2572,
+      "step": 11387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9768563242460468e-08,
+      "loss": 0.6128,
+      "step": 11388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9554891466205826e-08,
+      "loss": 0.2444,
+      "step": 11389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9341988162595593e-08,
+      "loss": 0.286,
+      "step": 11390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.912985334803775e-08,
+      "loss": 0.2689,
+      "step": 11391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.891848703888367e-08,
+      "loss": 0.2728,
+      "step": 11392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8707889251423647e-08,
+      "loss": 0.2765,
+      "step": 11393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.849806000189026e-08,
+      "loss": 0.2752,
+      "step": 11394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8288999306456122e-08,
+      "loss": 0.2593,
+      "step": 11395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8080707181232792e-08,
+      "loss": 0.2748,
+      "step": 11396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7873183642277423e-08,
+      "loss": 0.2015,
+      "step": 11397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.766642870558278e-08,
+      "loss": 0.2672,
+      "step": 11398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7460442387085005e-08,
+      "loss": 0.2762,
+      "step": 11399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7255224702660288e-08,
+      "loss": 0.3265,
+      "step": 11400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.705077566812708e-08,
+      "loss": 0.3335,
+      "step": 11401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6847095299241678e-08,
+      "loss": 0.3121,
+      "step": 11402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6644183611702623e-08,
+      "loss": 0.2772,
+      "step": 11403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6442040621150743e-08,
+      "loss": 0.2513,
+      "step": 11404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.62406663431658e-08,
+      "loss": 0.2474,
+      "step": 11405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6040060793268705e-08,
+      "loss": 0.2712,
+      "step": 11406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5840223986920432e-08,
+      "loss": 0.3283,
+      "step": 11407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5641155939524208e-08,
+      "loss": 0.2649,
+      "step": 11408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.544285666642221e-08,
+      "loss": 0.2557,
+      "step": 11409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5245326182899987e-08,
+      "loss": 0.2036,
+      "step": 11410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5048564504180918e-08,
+      "loss": 0.2394,
+      "step": 11411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4852571645430645e-08,
+      "loss": 0.3246,
+      "step": 11412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4657347621755977e-08,
+      "loss": 0.2924,
+      "step": 11413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4462892448202657e-08,
+      "loss": 0.275,
+      "step": 11414
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4269206139759804e-08,
+      "loss": 0.2529,
+      "step": 11415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.407628871135437e-08,
+      "loss": 0.2819,
+      "step": 11416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3884140177856675e-08,
+      "loss": 0.2571,
+      "step": 11417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.369276055407599e-08,
+      "loss": 0.2749,
+      "step": 11418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3502149854762734e-08,
+      "loss": 0.2868,
+      "step": 11419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3312308094607382e-08,
+      "loss": 0.2536,
+      "step": 11420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3123235288244895e-08,
+      "loss": 0.5137,
+      "step": 11421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2934931450245833e-08,
+      "loss": 0.2585,
+      "step": 11422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2747396595123038e-08,
+      "loss": 0.2614,
+      "step": 11423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.256063073733272e-08,
+      "loss": 0.3104,
+      "step": 11424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2374633891268928e-08,
+      "loss": 0.2795,
+      "step": 11425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.218940607126685e-08,
+      "loss": 0.2569,
+      "step": 11426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2004947291603964e-08,
+      "loss": 0.2558,
+      "step": 11427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.182125756649778e-08,
+      "loss": 0.2653,
+      "step": 11428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.163833691010475e-08,
+      "loss": 0.2581,
+      "step": 11429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1456185336524714e-08,
+      "loss": 0.3249,
+      "step": 11430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1274802859795328e-08,
+      "loss": 0.2426,
+      "step": 11431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.109418949389874e-08,
+      "loss": 0.2612,
+      "step": 11432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.091434525275382e-08,
+      "loss": 0.2803,
+      "step": 11433
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0735270150223917e-08,
+      "loss": 0.274,
+      "step": 11434
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.055696420011022e-08,
+      "loss": 0.2836,
+      "step": 11435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.037942741615617e-08,
+      "loss": 0.2904,
+      "step": 11436
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0202659812045278e-08,
+      "loss": 0.2227,
+      "step": 11437
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.002666140140108e-08,
+      "loss": 0.251,
+      "step": 11438
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.98514321977894e-08,
+      "loss": 0.3287,
+      "step": 11439
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9676972214716095e-08,
+      "loss": 0.2678,
+      "step": 11440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9503281465627077e-08,
+      "loss": 0.2937,
+      "step": 11441
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9330359963910527e-08,
+      "loss": 0.3317,
+      "step": 11442
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9158207722893564e-08,
+      "loss": 0.2131,
+      "step": 11443
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8986824755846677e-08,
+      "loss": 0.6558,
+      "step": 11444
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8816211075975976e-08,
+      "loss": 0.2593,
+      "step": 11445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.864636669643427e-08,
+      "loss": 0.2697,
+      "step": 11446
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.847729163030998e-08,
+      "loss": 0.2943,
+      "step": 11447
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8308985890637122e-08,
+      "loss": 0.2432,
+      "step": 11448
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.814144949038643e-08,
+      "loss": 0.3052,
+      "step": 11449
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7974682442470915e-08,
+      "loss": 0.2794,
+      "step": 11450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.780868475974362e-08,
+      "loss": 0.2299,
+      "step": 11451
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7643456454999875e-08,
+      "loss": 0.2826,
+      "step": 11452
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.747899754097504e-08,
+      "loss": 0.2925,
+      "step": 11453
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7315308030342314e-08,
+      "loss": 0.5677,
+      "step": 11454
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7152387935721603e-08,
+      "loss": 0.2896,
+      "step": 11455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.699023726966731e-08,
+      "loss": 0.3215,
+      "step": 11456
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.682885604467721e-08,
+      "loss": 0.2579,
+      "step": 11457
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.666824427319136e-08,
+      "loss": 0.2515,
+      "step": 11458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6508401967588738e-08,
+      "loss": 0.2817,
+      "step": 11459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6349329140188385e-08,
+      "loss": 0.2773,
+      "step": 11460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6191025803250493e-08,
+      "loss": 0.28,
+      "step": 11461
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6033491968976412e-08,
+      "loss": 0.2805,
+      "step": 11462
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.587672764950976e-08,
+      "loss": 0.285,
+      "step": 11463
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5720732856930878e-08,
+      "loss": 0.2961,
+      "step": 11464
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5565507603264585e-08,
+      "loss": 0.2628,
+      "step": 11465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.541105190047465e-08,
+      "loss": 0.2618,
+      "step": 11466
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.525736576046599e-08,
+      "loss": 0.2624,
+      "step": 11467
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5104449195082473e-08,
+      "loss": 0.3469,
+      "step": 11468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4952302216112437e-08,
+      "loss": 0.3184,
+      "step": 11469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.480092483527984e-08,
+      "loss": 0.2355,
+      "step": 11470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4650317064254238e-08,
+      "loss": 0.3027,
+      "step": 11471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4500478914644122e-08,
+      "loss": 0.2668,
+      "step": 11472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4351410397995813e-08,
+      "loss": 0.2556,
+      "step": 11473
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4203111525801228e-08,
+      "loss": 0.2491,
+      "step": 11474
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4055582309489002e-08,
+      "loss": 0.2595,
+      "step": 11475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.390882276043115e-08,
+      "loss": 0.2816,
+      "step": 11476
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3762832889937517e-08,
+      "loss": 0.3431,
+      "step": 11477
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3617612709262428e-08,
+      "loss": 0.2835,
+      "step": 11478
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3473162229596937e-08,
+      "loss": 0.2776,
+      "step": 11479
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3329481462075466e-08,
+      "loss": 0.2713,
+      "step": 11480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3186570417771383e-08,
+      "loss": 0.2608,
+      "step": 11481
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3044429107700319e-08,
+      "loss": 0.2422,
+      "step": 11482
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2903057542817954e-08,
+      "loss": 0.2609,
+      "step": 11483
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2762455734020018e-08,
+      "loss": 0.2527,
+      "step": 11484
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.262262369214451e-08,
+      "loss": 0.2543,
+      "step": 11485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.248356142796725e-08,
+      "loss": 0.5543,
+      "step": 11486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2345268952207445e-08,
+      "loss": 0.3311,
+      "step": 11487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2207746275523235e-08,
+      "loss": 0.3029,
+      "step": 11488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2070993408516141e-08,
+      "loss": 0.2455,
+      "step": 11489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1935010361724397e-08,
+      "loss": 0.2894,
+      "step": 11490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1799797145628511e-08,
+      "loss": 0.307,
+      "step": 11491
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1665353770652366e-08,
+      "loss": 0.2742,
+      "step": 11492
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1531680247156562e-08,
+      "loss": 0.2823,
+      "step": 11493
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1398776585445082e-08,
+      "loss": 0.2738,
+      "step": 11494
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.126664279575973e-08,
+      "loss": 0.2755,
+      "step": 11495
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1135278888286805e-08,
+      "loss": 0.2807,
+      "step": 11496
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1004684873149319e-08,
+      "loss": 0.2877,
+      "step": 11497
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0874860760413664e-08,
+      "loss": 0.2685,
+      "step": 11498
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0745806560086281e-08,
+      "loss": 0.2725,
+      "step": 11499
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0617522282113656e-08,
+      "loss": 0.3019,
+      "step": 11500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0490007936383439e-08,
+      "loss": 0.2753,
+      "step": 11501
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0363263532724433e-08,
+      "loss": 0.2572,
+      "step": 11502
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0237289080904377e-08,
+      "loss": 0.2715,
+      "step": 11503
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0112084590633287e-08,
+      "loss": 0.2794,
+      "step": 11504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.987650071561217e-09,
+      "loss": 0.2914,
+      "step": 11505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.863985533278275e-09,
+      "loss": 0.2257,
+      "step": 11506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.741090985316836e-09,
+      "loss": 0.2817,
+      "step": 11507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.618966437149324e-09,
+      "loss": 0.2543,
+      "step": 11508
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.497611898188209e-09,
+      "loss": 0.2716,
+      "step": 11509
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.377027377786007e-09,
+      "loss": 0.2936,
+      "step": 11510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.25721288523751e-09,
+      "loss": 0.276,
+      "step": 11511
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.13816842977755e-09,
+      "loss": 0.25,
+      "step": 11512
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.01989402058101e-09,
+      "loss": 0.2637,
+      "step": 11513
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.902389666765044e-09,
+      "loss": 0.2573,
+      "step": 11514
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.785655377384628e-09,
+      "loss": 0.3189,
+      "step": 11515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.66969116143701e-09,
+      "loss": 0.2859,
+      "step": 11516
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.554497027862818e-09,
+      "loss": 0.3009,
+      "step": 11517
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.440072985537174e-09,
+      "loss": 0.2725,
+      "step": 11518
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.326419043281909e-09,
+      "loss": 0.3018,
+      "step": 11519
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.213535209855571e-09,
+      "loss": 0.259,
+      "step": 11520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.101421493958984e-09,
+      "loss": 0.2982,
+      "step": 11521
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.990077904234117e-09,
+      "loss": 0.2548,
+      "step": 11522
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.879504449261887e-09,
+      "loss": 0.2206,
+      "step": 11523
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.769701137564368e-09,
+      "loss": 0.5447,
+      "step": 11524
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.660667977605896e-09,
+      "loss": 0.3079,
+      "step": 11525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.552404977788641e-09,
+      "loss": 0.2728,
+      "step": 11526
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.44491214645815e-09,
+      "loss": 0.3044,
+      "step": 11527
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.338189491900016e-09,
+      "loss": 0.3257,
+      "step": 11528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.232237022338773e-09,
+      "loss": 0.2479,
+      "step": 11529
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.12705474594011e-09,
+      "loss": 0.2343,
+      "step": 11530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.0226426708119855e-09,
+      "loss": 0.2923,
+      "step": 11531
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.919000805002407e-09,
+      "loss": 0.2963,
+      "step": 11532
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.816129156499429e-09,
+      "loss": 0.2526,
+      "step": 11533
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.714027733230044e-09,
+      "loss": 0.2718,
+      "step": 11534
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.612696543066843e-09,
+      "loss": 0.2905,
+      "step": 11535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.512135593816915e-09,
+      "loss": 0.2753,
+      "step": 11536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.412344893232947e-09,
+      "loss": 0.2878,
+      "step": 11537
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3133244490043434e-09,
+      "loss": 0.2676,
+      "step": 11538
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.215074268766108e-09,
+      "loss": 0.2621,
+      "step": 11539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.117594360088852e-09,
+      "loss": 0.2385,
+      "step": 11540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.020884730485455e-09,
+      "loss": 0.306,
+      "step": 11541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.924945387411063e-09,
+      "loss": 0.3141,
+      "step": 11542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8297763382597625e-09,
+      "loss": 0.2732,
+      "step": 11543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.735377590366797e-09,
+      "loss": 0.2704,
+      "step": 11544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6417491510074584e-09,
+      "loss": 0.3116,
+      "step": 11545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.548891027398195e-09,
+      "loss": 0.2813,
+      "step": 11546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.456803226696616e-09,
+      "loss": 0.3231,
+      "step": 11547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.365485756000377e-09,
+      "loss": 0.2432,
+      "step": 11548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.274938622348291e-09,
+      "loss": 0.2744,
+      "step": 11549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.185161832718111e-09,
+      "loss": 0.2561,
+      "step": 11550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.096155394028746e-09,
+      "loss": 0.2561,
+      "step": 11551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.007919313142484e-09,
+      "loss": 0.2926,
+      "step": 11552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.920453596859443e-09,
+      "loss": 0.2908,
+      "step": 11553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.833758251919785e-09,
+      "loss": 0.2725,
+      "step": 11554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.747833285007053e-09,
+      "loss": 0.2717,
+      "step": 11555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.662678702742618e-09,
+      "loss": 0.2628,
+      "step": 11556
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.578294511691228e-09,
+      "loss": 0.29,
+      "step": 11557
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.494680718355459e-09,
+      "loss": 0.2753,
+      "step": 11558
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.411837329181268e-09,
+      "loss": 0.2796,
+      "step": 11559
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.329764350552434e-09,
+      "loss": 0.3345,
+      "step": 11560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.248461788795011e-09,
+      "loss": 0.2124,
+      "step": 11561
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.167929650176206e-09,
+      "loss": 0.265,
+      "step": 11562
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.088167940902166e-09,
+      "loss": 0.2834,
+      "step": 11563
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.009176667121306e-09,
+      "loss": 0.5439,
+      "step": 11564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.93095583492209e-09,
+      "loss": 0.2127,
+      "step": 11565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.853505450331918e-09,
+      "loss": 0.2473,
+      "step": 11566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7768255193204595e-09,
+      "loss": 0.2492,
+      "step": 11567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.700916047799652e-09,
+      "loss": 0.2236,
+      "step": 11568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6257770416192606e-09,
+      "loss": 0.2706,
+      "step": 11569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5514085065690984e-09,
+      "loss": 0.2792,
+      "step": 11570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4778104483834676e-09,
+      "loss": 0.2814,
+      "step": 11571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4049828727333866e-09,
+      "loss": 0.306,
+      "step": 11572
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3329257852332543e-09,
+      "loss": 0.2882,
+      "step": 11573
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2616391914364056e-09,
+      "loss": 0.2967,
+      "step": 11574
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1911230968362238e-09,
+      "loss": 0.2654,
+      "step": 11575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1213775068683617e-09,
+      "loss": 0.2779,
+      "step": 11576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0524024269096286e-09,
+      "loss": 0.2856,
+      "step": 11577
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9841978622746624e-09,
+      "loss": 0.3293,
+      "step": 11578
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9167638182214796e-09,
+      "loss": 0.2673,
+      "step": 11579
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.850100299947034e-09,
+      "loss": 0.2715,
+      "step": 11580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7842073125894377e-09,
+      "loss": 0.2838,
+      "step": 11581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7190848612279606e-09,
+      "loss": 0.2588,
+      "step": 11582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.654732950880812e-09,
+      "loss": 0.3109,
+      "step": 11583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.591151586508467e-09,
+      "loss": 0.3022,
+      "step": 11584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5283407730114506e-09,
+      "loss": 0.3181,
+      "step": 11585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4663005152314455e-09,
+      "loss": 0.2777,
+      "step": 11586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.405030817949072e-09,
+      "loss": 0.2429,
+      "step": 11587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3445316858883295e-09,
+      "loss": 0.2407,
+      "step": 11588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2848031237099332e-09,
+      "loss": 0.2783,
+      "step": 11589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.225845136019089e-09,
+      "loss": 0.2467,
+      "step": 11590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1676577273610498e-09,
+      "loss": 0.2602,
+      "step": 11591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.110240902217786e-09,
+      "loss": 0.2925,
+      "step": 11592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.053594665016867e-09,
+      "loss": 0.2975,
+      "step": 11593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9977190201225793e-09,
+      "loss": 0.2503,
+      "step": 11594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9426139718436986e-09,
+      "loss": 0.2598,
+      "step": 11595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8882795244257177e-09,
+      "loss": 0.2744,
+      "step": 11596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8347156820563983e-09,
+      "loss": 0.2524,
+      "step": 11597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7819224488657695e-09,
+      "loss": 0.31,
+      "step": 11598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7298998289216884e-09,
+      "loss": 0.2989,
+      "step": 11599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.678647826234281e-09,
+      "loss": 0.3439,
+      "step": 11600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6281664447526103e-09,
+      "loss": 0.28,
+      "step": 11601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5784556883691183e-09,
+      "loss": 0.5886,
+      "step": 11602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5295155609151845e-09,
+      "loss": 0.2851,
+      "step": 11603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4813460661611267e-09,
+      "loss": 0.2918,
+      "step": 11604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.433947207821751e-09,
+      "loss": 0.2743,
+      "step": 11605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3873189895485806e-09,
+      "loss": 0.2662,
+      "step": 11606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3414614149365179e-09,
+      "loss": 0.2792,
+      "step": 11607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2963744875205132e-09,
+      "loss": 0.2714,
+      "step": 11608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2520582107733436e-09,
+      "loss": 0.2455,
+      "step": 11609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2085125881133863e-09,
+      "loss": 0.2542,
+      "step": 11610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.165737622895735e-09,
+      "loss": 0.2796,
+      "step": 11611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1237333184177523e-09,
+      "loss": 0.2532,
+      "step": 11612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.082499677915738e-09,
+      "loss": 0.2457,
+      "step": 11613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.042036704568261e-09,
+      "loss": 0.2759,
+      "step": 11614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.002344401495048e-09,
+      "loss": 0.2508,
+      "step": 11615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.634227717547629e-10,
+      "loss": 0.5614,
+      "step": 11616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.252718183472287e-10,
+      "loss": 0.2728,
+      "step": 11617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.878915442123159e-10,
+      "loss": 0.2224,
+      "step": 11618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.512819522310533e-10,
+      "loss": 0.5873,
+      "step": 11619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.154430452267381e-10,
+      "loss": 0.2794,
+      "step": 11620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.803748259604949e-10,
+      "loss": 0.2524,
+      "step": 11621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.460772971357167e-10,
+      "loss": 0.2491,
+      "step": 11622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.125504613947343e-10,
+      "loss": 0.2253,
+      "step": 11623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.797943213232572e-10,
+      "loss": 0.2627,
+      "step": 11624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.478088794448223e-10,
+      "loss": 0.3286,
+      "step": 11625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.165941382241248e-10,
+      "loss": 0.2869,
+      "step": 11626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.861501000692382e-10,
+      "loss": 0.2546,
+      "step": 11627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.564767673249538e-10,
+      "loss": 0.3231,
+      "step": 11628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.27574142278331e-10,
+      "loss": 0.2661,
+      "step": 11629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.994422271575872e-10,
+      "loss": 0.2284,
+      "step": 11630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.720810241309881e-10,
+      "loss": 0.2263,
+      "step": 11631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.454905353068473e-10,
+      "loss": 0.23,
+      "step": 11632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1967076273574657e-10,
+      "loss": 0.254,
+      "step": 11633
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.946217084072057e-10,
+      "loss": 0.2627,
+      "step": 11634
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7034337425079225e-10,
+      "loss": 0.232,
+      "step": 11635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4683576213945245e-10,
+      "loss": 0.3064,
+      "step": 11636
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.240988738839601e-10,
+      "loss": 0.24,
+      "step": 11637
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0213271123735735e-10,
+      "loss": 0.2811,
+      "step": 11638
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.80937275891624e-10,
+      "loss": 0.5745,
+      "step": 11639
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.605125694821187e-10,
+      "loss": 0.2322,
+      "step": 11640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4085859358202733e-10,
+      "loss": 0.5648,
+      "step": 11641
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2197534970569424e-10,
+      "loss": 0.2676,
+      "step": 11642
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0386283930973194e-10,
+      "loss": 0.2805,
+      "step": 11643
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8652106378969082e-10,
+      "loss": 0.2921,
+      "step": 11644
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6995002448116916e-10,
+      "loss": 0.2927,
+      "step": 11645
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5414972266314389e-10,
+      "loss": 0.2812,
+      "step": 11646
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3912015955241942e-10,
+      "loss": 0.2792,
+      "step": 11647
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.248613363069584e-10,
+      "loss": 0.2933,
+      "step": 11648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1137325402699184e-10,
+      "loss": 0.3054,
+      "step": 11649
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.865591375168848e-11,
+      "loss": 0.2521,
+      "step": 11650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.670931646026504e-11,
+      "loss": 0.2402,
+      "step": 11651
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.553346307420662e-11,
+      "loss": 0.2681,
+      "step": 11652
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.512835445615651e-11,
+      "loss": 0.2566,
+      "step": 11653
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.549399140547529e-11,
+      "loss": 0.2853,
+      "step": 11654
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.663037466712261e-11,
+      "loss": 0.2484,
+      "step": 11655
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.853750492277542e-11,
+      "loss": 0.2497,
+      "step": 11656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1215382796379036e-11,
+      "loss": 0.3162,
+      "step": 11657
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.466400885303699e-11,
+      "loss": 0.2766,
+      "step": 11658
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8883383597900763e-11,
+      "loss": 0.2817,
+      "step": 11659
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3873507475059556e-11,
+      "loss": 0.2537,
+      "step": 11660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.634380871981209e-12,
+      "loss": 0.262,
+      "step": 11661
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.166004113961066e-12,
+      "loss": 0.239,
+      "step": 11662
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.468377470783324e-12,
+      "loss": 0.2936,
+      "step": 11663
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5415011478392416e-12,
+      "loss": 0.2358,
+      "step": 11664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.853752938987043e-13,
+      "loss": 0.6515,
+      "step": 11665
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.3406,
+      "step": 11666
+    },
+    {
+      "epoch": 1.0,
+      "step": 11666,
+      "total_flos": 1.6909495139958784e+16,
+      "train_loss": 0.30232172586201356,
+      "train_runtime": 93155.4817,
+      "train_samples_per_second": 16.029,
+      "train_steps_per_second": 0.125
+    }
+  ],
+  "max_steps": 11666,
+  "num_train_epochs": 1,
+  "total_flos": 1.6909495139958784e+16,
+  "trial_name": null,
+  "trial_params": null
+}