diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31470 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9998091785135006,
+  "eval_steps": 500,
+  "global_step": 5240,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.816793893129771e-08,
+      "loss": 2.8601,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.633587786259542e-08,
+      "loss": 2.9835,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1450381679389314e-07,
+      "loss": 2.9077,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5267175572519085e-07,
+      "loss": 3.0172,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9083969465648858e-07,
+      "loss": 2.8581,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2900763358778629e-07,
+      "loss": 2.8634,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.67175572519084e-07,
+      "loss": 3.1031,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.053435114503817e-07,
+      "loss": 2.9575,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4351145038167945e-07,
+      "loss": 2.9849,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8167938931297716e-07,
+      "loss": 2.8806,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1984732824427486e-07,
+      "loss": 2.8774,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5801526717557257e-07,
+      "loss": 2.7967,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.961832061068702e-07,
+      "loss": 2.9459,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.34351145038168e-07,
+      "loss": 2.998,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.725190839694656e-07,
+      "loss": 2.9337,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.106870229007634e-07,
+      "loss": 2.8418,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.48854961832061e-07,
+      "loss": 2.8657,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.870229007633589e-07,
+      "loss": 2.7894,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.251908396946565e-07,
+      "loss": 2.8208,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.633587786259543e-07,
+      "loss": 2.8547,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.01526717557252e-07,
+      "loss": 2.8375,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.396946564885497e-07,
+      "loss": 2.6486,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.778625954198474e-07,
+      "loss": 2.7399,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.160305343511451e-07,
+      "loss": 2.7422,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.54198473282443e-07,
+      "loss": 2.6639,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.923664122137404e-07,
+      "loss": 2.5761,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0305343511450382e-06,
+      "loss": 2.4863,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.068702290076336e-06,
+      "loss": 2.4915,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1068702290076337e-06,
+      "loss": 2.4168,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1450381679389313e-06,
+      "loss": 2.3596,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1832061068702292e-06,
+      "loss": 2.2872,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2213740458015268e-06,
+      "loss": 2.2365,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2595419847328243e-06,
+      "loss": 2.2943,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.297709923664122e-06,
+      "loss": 2.2375,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.33587786259542e-06,
+      "loss": 2.2254,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3740458015267178e-06,
+      "loss": 2.1365,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4122137404580156e-06,
+      "loss": 2.1035,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.450381679389313e-06,
+      "loss": 2.1137,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4885496183206109e-06,
+      "loss": 2.1345,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5267175572519086e-06,
+      "loss": 2.0083,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5648854961832064e-06,
+      "loss": 1.9695,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.603053435114504e-06,
+      "loss": 1.9026,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6412213740458017e-06,
+      "loss": 1.8858,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6793893129770995e-06,
+      "loss": 1.9075,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7175572519083972e-06,
+      "loss": 1.8278,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7557251908396948e-06,
+      "loss": 1.7658,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7938931297709925e-06,
+      "loss": 1.7775,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8320610687022903e-06,
+      "loss": 1.7948,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.870229007633588e-06,
+      "loss": 1.6865,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.908396946564886e-06,
+      "loss": 1.6504,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.946564885496183e-06,
+      "loss": 1.6068,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.984732824427481e-06,
+      "loss": 1.6214,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0229007633587786e-06,
+      "loss": 1.5573,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0610687022900764e-06,
+      "loss": 1.5957,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.099236641221374e-06,
+      "loss": 1.5849,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.137404580152672e-06,
+      "loss": 1.5081,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1755725190839697e-06,
+      "loss": 1.5087,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2137404580152674e-06,
+      "loss": 1.4906,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2519083969465648e-06,
+      "loss": 1.5073,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2900763358778625e-06,
+      "loss": 1.5025,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3282442748091603e-06,
+      "loss": 1.4827,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3664122137404585e-06,
+      "loss": 1.4905,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4045801526717562e-06,
+      "loss": 1.3945,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4427480916030536e-06,
+      "loss": 1.4366,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4809160305343513e-06,
+      "loss": 1.2939,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5190839694656487e-06,
+      "loss": 1.4239,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5572519083969464e-06,
+      "loss": 1.3292,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.595419847328244e-06,
+      "loss": 1.3481,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.633587786259542e-06,
+      "loss": 1.4193,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.67175572519084e-06,
+      "loss": 1.4267,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.709923664122138e-06,
+      "loss": 1.3452,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7480916030534356e-06,
+      "loss": 1.3791,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7862595419847334e-06,
+      "loss": 1.3556,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.824427480916031e-06,
+      "loss": 1.4257,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.862595419847328e-06,
+      "loss": 1.3358,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.900763358778626e-06,
+      "loss": 1.2891,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.938931297709924e-06,
+      "loss": 1.3252,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9770992366412218e-06,
+      "loss": 1.317,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0152671755725195e-06,
+      "loss": 1.1875,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0534351145038173e-06,
+      "loss": 1.3842,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.091603053435115e-06,
+      "loss": 1.2908,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.129770992366413e-06,
+      "loss": 1.2179,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1679389312977097e-06,
+      "loss": 1.3105,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.206106870229008e-06,
+      "loss": 1.3245,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2442748091603056e-06,
+      "loss": 1.2454,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2824427480916034e-06,
+      "loss": 1.2095,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.320610687022901e-06,
+      "loss": 1.2077,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.358778625954199e-06,
+      "loss": 1.235,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3969465648854967e-06,
+      "loss": 1.2913,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4351145038167944e-06,
+      "loss": 1.1953,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.473282442748092e-06,
+      "loss": 1.1901,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5114503816793895e-06,
+      "loss": 1.1394,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5496183206106873e-06,
+      "loss": 1.1986,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.587786259541985e-06,
+      "loss": 1.1783,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.625954198473283e-06,
+      "loss": 1.1792,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6641221374045806e-06,
+      "loss": 1.1686,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7022900763358783e-06,
+      "loss": 1.1987,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.740458015267176e-06,
+      "loss": 1.1966,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.778625954198474e-06,
+      "loss": 1.1512,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.816793893129772e-06,
+      "loss": 1.0064,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8549618320610685e-06,
+      "loss": 1.1742,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.893129770992366e-06,
+      "loss": 1.1524,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.931297709923664e-06,
+      "loss": 1.1695,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.969465648854962e-06,
+      "loss": 1.0594,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.0076335877862595e-06,
+      "loss": 1.1005,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.045801526717557e-06,
+      "loss": 1.0862,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.083969465648855e-06,
+      "loss": 1.0728,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.122137404580153e-06,
+      "loss": 1.102,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1603053435114506e-06,
+      "loss": 1.1613,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.198473282442748e-06,
+      "loss": 1.0773,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.236641221374046e-06,
+      "loss": 1.0762,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.274809160305344e-06,
+      "loss": 1.0331,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.312977099236642e-06,
+      "loss": 1.1052,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.351145038167939e-06,
+      "loss": 1.1094,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.389312977099237e-06,
+      "loss": 1.0246,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.427480916030535e-06,
+      "loss": 1.0797,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.465648854961833e-06,
+      "loss": 1.0328,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5038167938931296e-06,
+      "loss": 0.9832,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.541984732824427e-06,
+      "loss": 0.9844,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.580152671755725e-06,
+      "loss": 0.9976,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.618320610687023e-06,
+      "loss": 1.0461,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.656488549618321e-06,
+      "loss": 0.9526,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.694656488549618e-06,
+      "loss": 0.9276,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.732824427480917e-06,
+      "loss": 1.0464,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.770992366412215e-06,
+      "loss": 1.0596,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8091603053435125e-06,
+      "loss": 1.0478,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.847328244274809e-06,
+      "loss": 1.0469,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.885496183206107e-06,
+      "loss": 1.005,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923664122137405e-06,
+      "loss": 0.9388,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.961832061068703e-06,
+      "loss": 1.0445,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5e-06,
+      "loss": 0.9611,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.038167938931297e-06,
+      "loss": 1.037,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.076335877862596e-06,
+      "loss": 1.0059,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.114503816793893e-06,
+      "loss": 0.9547,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1526717557251914e-06,
+      "loss": 1.0059,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.190839694656488e-06,
+      "loss": 0.9478,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.229007633587787e-06,
+      "loss": 0.9162,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.267175572519084e-06,
+      "loss": 0.9887,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3053435114503825e-06,
+      "loss": 1.0016,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.34351145038168e-06,
+      "loss": 0.9471,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.381679389312977e-06,
+      "loss": 1.0157,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.419847328244276e-06,
+      "loss": 0.9056,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.458015267175573e-06,
+      "loss": 0.9537,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.496183206106871e-06,
+      "loss": 0.9361,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.534351145038168e-06,
+      "loss": 0.9282,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.572519083969467e-06,
+      "loss": 0.8987,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.610687022900764e-06,
+      "loss": 0.8967,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.648854961832062e-06,
+      "loss": 0.9091,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.687022900763359e-06,
+      "loss": 0.9588,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.725190839694656e-06,
+      "loss": 0.9202,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.763358778625955e-06,
+      "loss": 0.9421,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.801526717557252e-06,
+      "loss": 0.913,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.83969465648855e-06,
+      "loss": 0.9341,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.877862595419848e-06,
+      "loss": 0.8842,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.916030534351146e-06,
+      "loss": 0.8903,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9541984732824435e-06,
+      "loss": 0.8714,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.992366412213741e-06,
+      "loss": 0.8863,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.030534351145039e-06,
+      "loss": 0.9727,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.068702290076336e-06,
+      "loss": 0.8482,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1068702290076346e-06,
+      "loss": 0.8812,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1450381679389315e-06,
+      "loss": 0.8982,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.18320610687023e-06,
+      "loss": 0.8793,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.221374045801527e-06,
+      "loss": 0.9068,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.259541984732826e-06,
+      "loss": 0.9265,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.2977099236641225e-06,
+      "loss": 0.8524,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.335877862595419e-06,
+      "loss": 0.8602,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.374045801526718e-06,
+      "loss": 0.922,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.412213740458016e-06,
+      "loss": 0.7954,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.4503816793893135e-06,
+      "loss": 0.829,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.488549618320611e-06,
+      "loss": 0.8432,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.526717557251909e-06,
+      "loss": 0.8485,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.564885496183207e-06,
+      "loss": 0.8099,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.6030534351145046e-06,
+      "loss": 0.8042,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.641221374045802e-06,
+      "loss": 0.889,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.679389312977099e-06,
+      "loss": 0.8619,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.717557251908398e-06,
+      "loss": 0.8239,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.755725190839695e-06,
+      "loss": 0.891,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.793893129770993e-06,
+      "loss": 0.8132,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.83206106870229e-06,
+      "loss": 0.8606,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.870229007633589e-06,
+      "loss": 0.8105,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.908396946564886e-06,
+      "loss": 0.8214,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.946564885496184e-06,
+      "loss": 0.845,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.984732824427481e-06,
+      "loss": 0.8349,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.022900763358779e-06,
+      "loss": 0.834,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.061068702290077e-06,
+      "loss": 0.8061,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.0992366412213746e-06,
+      "loss": 0.8584,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.137404580152672e-06,
+      "loss": 0.7525,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.17557251908397e-06,
+      "loss": 0.8322,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.213740458015268e-06,
+      "loss": 0.868,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.251908396946566e-06,
+      "loss": 0.9172,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.290076335877863e-06,
+      "loss": 0.8734,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.328244274809161e-06,
+      "loss": 0.8097,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.366412213740458e-06,
+      "loss": 0.7621,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.404580152671757e-06,
+      "loss": 0.8223,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.4427480916030536e-06,
+      "loss": 0.7873,
+      "step": 195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.480916030534352e-06,
+      "loss": 0.7954,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.519083969465649e-06,
+      "loss": 0.7821,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.557251908396948e-06,
+      "loss": 0.8133,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.595419847328245e-06,
+      "loss": 0.8013,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.633587786259543e-06,
+      "loss": 0.786,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.671755725190841e-06,
+      "loss": 0.8251,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.709923664122137e-06,
+      "loss": 0.7926,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.748091603053436e-06,
+      "loss": 0.7347,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.786259541984733e-06,
+      "loss": 0.8986,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.824427480916032e-06,
+      "loss": 0.7932,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.862595419847328e-06,
+      "loss": 0.8068,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.900763358778627e-06,
+      "loss": 0.7894,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.938931297709924e-06,
+      "loss": 0.7819,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.977099236641223e-06,
+      "loss": 0.723,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.015267175572519e-06,
+      "loss": 0.8025,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.053435114503817e-06,
+      "loss": 0.7863,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.091603053435115e-06,
+      "loss": 0.7295,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.129770992366412e-06,
+      "loss": 0.8402,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.16793893129771e-06,
+      "loss": 0.8217,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.206106870229008e-06,
+      "loss": 0.8454,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.244274809160306e-06,
+      "loss": 0.789,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.282442748091603e-06,
+      "loss": 0.8004,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.320610687022901e-06,
+      "loss": 0.7878,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.358778625954199e-06,
+      "loss": 0.7476,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.396946564885497e-06,
+      "loss": 0.8096,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.435114503816794e-06,
+      "loss": 0.8225,
+      "step": 221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.473282442748092e-06,
+      "loss": 0.7579,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.51145038167939e-06,
+      "loss": 0.7812,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.549618320610688e-06,
+      "loss": 0.7666,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.587786259541985e-06,
+      "loss": 0.7866,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.625954198473283e-06,
+      "loss": 0.7937,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.664122137404581e-06,
+      "loss": 0.7743,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.702290076335879e-06,
+      "loss": 0.8342,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.740458015267176e-06,
+      "loss": 0.7408,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.778625954198474e-06,
+      "loss": 0.74,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.816793893129772e-06,
+      "loss": 0.7816,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.85496183206107e-06,
+      "loss": 0.7403,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.893129770992368e-06,
+      "loss": 0.7397,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.931297709923665e-06,
+      "loss": 0.7948,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.969465648854963e-06,
+      "loss": 0.7901,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.007633587786259e-06,
+      "loss": 0.8117,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.045801526717559e-06,
+      "loss": 0.7121,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.083969465648855e-06,
+      "loss": 0.7426,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.122137404580154e-06,
+      "loss": 0.7892,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.16030534351145e-06,
+      "loss": 0.7721,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.19847328244275e-06,
+      "loss": 0.8216,
+      "step": 241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.236641221374046e-06,
+      "loss": 0.7793,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.274809160305345e-06,
+      "loss": 0.7659,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.312977099236641e-06,
+      "loss": 0.7702,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.351145038167939e-06,
+      "loss": 0.7671,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.389312977099237e-06,
+      "loss": 0.7592,
+      "step": 246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.427480916030534e-06,
+      "loss": 0.7701,
+      "step": 247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.465648854961834e-06,
+      "loss": 0.8196,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.50381679389313e-06,
+      "loss": 0.7281,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.54198473282443e-06,
+      "loss": 0.7348,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.580152671755725e-06,
+      "loss": 0.752,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.618320610687025e-06,
+      "loss": 0.7881,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.656488549618321e-06,
+      "loss": 0.7847,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.694656488549619e-06,
+      "loss": 0.7322,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.732824427480917e-06,
+      "loss": 0.7515,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770992366412214e-06,
+      "loss": 0.801,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.809160305343512e-06,
+      "loss": 0.7717,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.84732824427481e-06,
+      "loss": 0.7606,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885496183206108e-06,
+      "loss": 0.7916,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.923664122137405e-06,
+      "loss": 0.6814,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961832061068703e-06,
+      "loss": 0.7244,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 0.6936,
+      "step": 262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0038167938931299e-05,
+      "loss": 0.8257,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0076335877862595e-05,
+      "loss": 0.8227,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0114503816793894e-05,
+      "loss": 0.725,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0152671755725192e-05,
+      "loss": 0.7629,
+      "step": 266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.019083969465649e-05,
+      "loss": 0.799,
+      "step": 267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0229007633587786e-05,
+      "loss": 0.8142,
+      "step": 268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0267175572519085e-05,
+      "loss": 0.7742,
+      "step": 269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0305343511450383e-05,
+      "loss": 0.7825,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.034351145038168e-05,
+      "loss": 0.7907,
+      "step": 271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0381679389312977e-05,
+      "loss": 0.7807,
+      "step": 272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0419847328244274e-05,
+      "loss": 0.7538,
+      "step": 273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0458015267175574e-05,
+      "loss": 0.7366,
+      "step": 274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0496183206106872e-05,
+      "loss": 0.7907,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0534351145038168e-05,
+      "loss": 0.7767,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0572519083969465e-05,
+      "loss": 0.7592,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0610687022900765e-05,
+      "loss": 0.7503,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0648854961832063e-05,
+      "loss": 0.7779,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.068702290076336e-05,
+      "loss": 0.7378,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0725190839694657e-05,
+      "loss": 0.7174,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0763358778625954e-05,
+      "loss": 0.6916,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0801526717557254e-05,
+      "loss": 0.7806,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0839694656488552e-05,
+      "loss": 0.7564,
+      "step": 284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0877862595419848e-05,
+      "loss": 0.7766,
+      "step": 285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0916030534351145e-05,
+      "loss": 0.7244,
+      "step": 286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0954198473282445e-05,
+      "loss": 0.719,
+      "step": 287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0992366412213743e-05,
+      "loss": 0.7433,
+      "step": 288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1030534351145039e-05,
+      "loss": 0.7378,
+      "step": 289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1068702290076336e-05,
+      "loss": 0.772,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1106870229007634e-05,
+      "loss": 0.7765,
+      "step": 291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1145038167938934e-05,
+      "loss": 0.7189,
+      "step": 292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.118320610687023e-05,
+      "loss": 0.743,
+      "step": 293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1221374045801527e-05,
+      "loss": 0.7549,
+      "step": 294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1259541984732825e-05,
+      "loss": 0.7607,
+      "step": 295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1297709923664125e-05,
+      "loss": 0.7418,
+      "step": 296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.133587786259542e-05,
+      "loss": 0.7361,
+      "step": 297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1374045801526718e-05,
+      "loss": 0.7116,
+      "step": 298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1412213740458016e-05,
+      "loss": 0.6838,
+      "step": 299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1450381679389312e-05,
+      "loss": 0.6841,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1488549618320612e-05,
+      "loss": 0.7279,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.152671755725191e-05,
+      "loss": 0.7606,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1564885496183207e-05,
+      "loss": 0.774,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1603053435114503e-05,
+      "loss": 0.7571,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1641221374045803e-05,
+      "loss": 0.6712,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.16793893129771e-05,
+      "loss": 0.7246,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1717557251908398e-05,
+      "loss": 0.8182,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1755725190839696e-05,
+      "loss": 0.6826,
+      "step": 308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1793893129770992e-05,
+      "loss": 0.739,
+      "step": 309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1832061068702292e-05,
+      "loss": 0.7504,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.187022900763359e-05,
+      "loss": 0.6731,
+      "step": 311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1908396946564887e-05,
+      "loss": 0.7136,
+      "step": 312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1946564885496183e-05,
+      "loss": 0.6471,
+      "step": 313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1984732824427483e-05,
+      "loss": 0.6998,
+      "step": 314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.202290076335878e-05,
+      "loss": 0.7867,
+      "step": 315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2061068702290078e-05,
+      "loss": 0.7286,
+      "step": 316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2099236641221374e-05,
+      "loss": 0.737,
+      "step": 317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2137404580152672e-05,
+      "loss": 0.7368,
+      "step": 318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2175572519083971e-05,
+      "loss": 0.7483,
+      "step": 319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2213740458015269e-05,
+      "loss": 0.7088,
+      "step": 320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2251908396946565e-05,
+      "loss": 0.7708,
+      "step": 321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2290076335877863e-05,
+      "loss": 0.7079,
+      "step": 322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2328244274809162e-05,
+      "loss": 0.7115,
+      "step": 323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.236641221374046e-05,
+      "loss": 0.7214,
+      "step": 324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2404580152671756e-05,
+      "loss": 0.7369,
+      "step": 325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2442748091603054e-05,
+      "loss": 0.7214,
+      "step": 326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2480916030534352e-05,
+      "loss": 0.6628,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2519083969465651e-05,
+      "loss": 0.6557,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2557251908396947e-05,
+      "loss": 0.6785,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2595419847328245e-05,
+      "loss": 0.7662,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2633587786259543e-05,
+      "loss": 0.6766,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2671755725190839e-05,
+      "loss": 0.6852,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2709923664122138e-05,
+      "loss": 0.7517,
+      "step": 333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2748091603053436e-05,
+      "loss": 0.7739,
+      "step": 334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2786259541984734e-05,
+      "loss": 0.7308,
+      "step": 335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2824427480916032e-05,
+      "loss": 0.663,
+      "step": 336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2862595419847331e-05,
+      "loss": 0.7439,
+      "step": 337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2900763358778627e-05,
+      "loss": 0.7291,
+      "step": 338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2938931297709925e-05,
+      "loss": 0.6848,
+      "step": 339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2977099236641223e-05,
+      "loss": 0.6717,
+      "step": 340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3015267175572519e-05,
+      "loss": 0.7069,
+      "step": 341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3053435114503818e-05,
+      "loss": 0.6992,
+      "step": 342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3091603053435116e-05,
+      "loss": 0.6716,
+      "step": 343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3129770992366414e-05,
+      "loss": 0.6833,
+      "step": 344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.316793893129771e-05,
+      "loss": 0.702,
+      "step": 345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3206106870229009e-05,
+      "loss": 0.6599,
+      "step": 346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3244274809160307e-05,
+      "loss": 0.6983,
+      "step": 347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3282442748091605e-05,
+      "loss": 0.6947,
+      "step": 348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.33206106870229e-05,
+      "loss": 0.7185,
+      "step": 349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3358778625954198e-05,
+      "loss": 0.7439,
+      "step": 350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3396946564885498e-05,
+      "loss": 0.7242,
+      "step": 351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3435114503816796e-05,
+      "loss": 0.7134,
+      "step": 352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3473282442748092e-05,
+      "loss": 0.718,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.351145038167939e-05,
+      "loss": 0.7391,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3549618320610689e-05,
+      "loss": 0.6841,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3587786259541987e-05,
+      "loss": 0.6414,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3625954198473283e-05,
+      "loss": 0.693,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.366412213740458e-05,
+      "loss": 0.6408,
+      "step": 358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3702290076335878e-05,
+      "loss": 0.6853,
+      "step": 359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3740458015267178e-05,
+      "loss": 0.6947,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3778625954198474e-05,
+      "loss": 0.7376,
+      "step": 361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3816793893129772e-05,
+      "loss": 0.692,
+      "step": 362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.385496183206107e-05,
+      "loss": 0.6499,
+      "step": 363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3893129770992369e-05,
+      "loss": 0.7024,
+      "step": 364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3931297709923667e-05,
+      "loss": 0.6857,
+      "step": 365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3969465648854963e-05,
+      "loss": 0.6678,
+      "step": 366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.400763358778626e-05,
+      "loss": 0.7287,
+      "step": 367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4045801526717558e-05,
+      "loss": 0.7269,
+      "step": 368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4083969465648858e-05,
+      "loss": 0.6783,
+      "step": 369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4122137404580154e-05,
+      "loss": 0.6294,
+      "step": 370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4160305343511451e-05,
+      "loss": 0.7104,
+      "step": 371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4198473282442749e-05,
+      "loss": 0.7096,
+      "step": 372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4236641221374049e-05,
+      "loss": 0.6408,
+      "step": 373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4274809160305345e-05,
+      "loss": 0.7081,
+      "step": 374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4312977099236642e-05,
+      "loss": 0.6611,
+      "step": 375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.435114503816794e-05,
+      "loss": 0.6981,
+      "step": 376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4389312977099236e-05,
+      "loss": 0.6714,
+      "step": 377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4427480916030536e-05,
+      "loss": 0.7072,
+      "step": 378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4465648854961833e-05,
+      "loss": 0.6835,
+      "step": 379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4503816793893131e-05,
+      "loss": 0.6866,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4541984732824427e-05,
+      "loss": 0.6575,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4580152671755727e-05,
+      "loss": 0.7225,
+      "step": 382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4618320610687024e-05,
+      "loss": 0.7626,
+      "step": 383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4656488549618322e-05,
+      "loss": 0.673,
+      "step": 384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4694656488549618e-05,
+      "loss": 0.7332,
+      "step": 385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4732824427480916e-05,
+      "loss": 0.6801,
+      "step": 386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4770992366412216e-05,
+      "loss": 0.7073,
+      "step": 387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4809160305343513e-05,
+      "loss": 0.7401,
+      "step": 388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.484732824427481e-05,
+      "loss": 0.7283,
+      "step": 389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4885496183206107e-05,
+      "loss": 0.6599,
+      "step": 390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4923664122137407e-05,
+      "loss": 0.6432,
+      "step": 391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4961832061068704e-05,
+      "loss": 0.7017,
+      "step": 392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.6963,
+      "step": 393
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5038167938931298e-05,
+      "loss": 0.712,
+      "step": 394
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5076335877862596e-05,
+      "loss": 0.7121,
+      "step": 395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5114503816793895e-05,
+      "loss": 0.7119,
+      "step": 396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5152671755725193e-05,
+      "loss": 0.726,
+      "step": 397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.519083969465649e-05,
+      "loss": 0.6842,
+      "step": 398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5229007633587787e-05,
+      "loss": 0.7019,
+      "step": 399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5267175572519086e-05,
+      "loss": 0.6579,
+      "step": 400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5305343511450384e-05,
+      "loss": 0.7055,
+      "step": 401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5343511450381682e-05,
+      "loss": 0.7181,
+      "step": 402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5381679389312976e-05,
+      "loss": 0.637,
+      "step": 403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5419847328244274e-05,
+      "loss": 0.6354,
+      "step": 404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5458015267175575e-05,
+      "loss": 0.689,
+      "step": 405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5496183206106873e-05,
+      "loss": 0.6423,
+      "step": 406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5534351145038167e-05,
+      "loss": 0.6334,
+      "step": 407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5572519083969465e-05,
+      "loss": 0.6601,
+      "step": 408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5610687022900766e-05,
+      "loss": 0.7194,
+      "step": 409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5648854961832064e-05,
+      "loss": 0.7082,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5687022900763362e-05,
+      "loss": 0.6757,
+      "step": 411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5725190839694656e-05,
+      "loss": 0.678,
+      "step": 412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5763358778625954e-05,
+      "loss": 0.6761,
+      "step": 413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5801526717557255e-05,
+      "loss": 0.6458,
+      "step": 414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5839694656488553e-05,
+      "loss": 0.737,
+      "step": 415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5877862595419847e-05,
+      "loss": 0.6845,
+      "step": 416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5916030534351145e-05,
+      "loss": 0.7453,
+      "step": 417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5954198473282446e-05,
+      "loss": 0.6579,
+      "step": 418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5992366412213744e-05,
+      "loss": 0.6838,
+      "step": 419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6030534351145038e-05,
+      "loss": 0.6465,
+      "step": 420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6068702290076336e-05,
+      "loss": 0.614,
+      "step": 421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6106870229007634e-05,
+      "loss": 0.7095,
+      "step": 422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6145038167938935e-05,
+      "loss": 0.6086,
+      "step": 423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.618320610687023e-05,
+      "loss": 0.6546,
+      "step": 424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6221374045801527e-05,
+      "loss": 0.6783,
+      "step": 425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6259541984732825e-05,
+      "loss": 0.7063,
+      "step": 426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6297709923664126e-05,
+      "loss": 0.6845,
+      "step": 427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.633587786259542e-05,
+      "loss": 0.6598,
+      "step": 428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6374045801526718e-05,
+      "loss": 0.6324,
+      "step": 429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6412213740458016e-05,
+      "loss": 0.7012,
+      "step": 430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6450381679389314e-05,
+      "loss": 0.6635,
+      "step": 431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.648854961832061e-05,
+      "loss": 0.6558,
+      "step": 432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.652671755725191e-05,
+      "loss": 0.6348,
+      "step": 433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6564885496183207e-05,
+      "loss": 0.5867,
+      "step": 434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6603053435114505e-05,
+      "loss": 0.6771,
+      "step": 435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6641221374045802e-05,
+      "loss": 0.7384,
+      "step": 436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.66793893129771e-05,
+      "loss": 0.6217,
+      "step": 437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6717557251908398e-05,
+      "loss": 0.7207,
+      "step": 438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6755725190839696e-05,
+      "loss": 0.7034,
+      "step": 439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6793893129770993e-05,
+      "loss": 0.7212,
+      "step": 440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.683206106870229e-05,
+      "loss": 0.6623,
+      "step": 441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.687022900763359e-05,
+      "loss": 0.6735,
+      "step": 442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6908396946564887e-05,
+      "loss": 0.6737,
+      "step": 443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6946564885496184e-05,
+      "loss": 0.6752,
+      "step": 444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6984732824427482e-05,
+      "loss": 0.6515,
+      "step": 445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.702290076335878e-05,
+      "loss": 0.6803,
+      "step": 446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7061068702290078e-05,
+      "loss": 0.7071,
+      "step": 447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7099236641221375e-05,
+      "loss": 0.656,
+      "step": 448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7137404580152673e-05,
+      "loss": 0.6895,
+      "step": 449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.717557251908397e-05,
+      "loss": 0.6963,
+      "step": 450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.721374045801527e-05,
+      "loss": 0.7263,
+      "step": 451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7251908396946566e-05,
+      "loss": 0.6925,
+      "step": 452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7290076335877864e-05,
+      "loss": 0.6589,
+      "step": 453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7328244274809162e-05,
+      "loss": 0.6882,
+      "step": 454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.736641221374046e-05,
+      "loss": 0.678,
+      "step": 455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7404580152671757e-05,
+      "loss": 0.6247,
+      "step": 456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7442748091603055e-05,
+      "loss": 0.7408,
+      "step": 457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7480916030534353e-05,
+      "loss": 0.6608,
+      "step": 458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.751908396946565e-05,
+      "loss": 0.7366,
+      "step": 459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.755725190839695e-05,
+      "loss": 0.6687,
+      "step": 460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7595419847328246e-05,
+      "loss": 0.713,
+      "step": 461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7633587786259544e-05,
+      "loss": 0.6882,
+      "step": 462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.767175572519084e-05,
+      "loss": 0.6931,
+      "step": 463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.770992366412214e-05,
+      "loss": 0.6773,
+      "step": 464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7748091603053437e-05,
+      "loss": 0.6911,
+      "step": 465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7786259541984735e-05,
+      "loss": 0.6618,
+      "step": 466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7824427480916033e-05,
+      "loss": 0.662,
+      "step": 467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.786259541984733e-05,
+      "loss": 0.6838,
+      "step": 468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.790076335877863e-05,
+      "loss": 0.7071,
+      "step": 469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7938931297709926e-05,
+      "loss": 0.6631,
+      "step": 470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7977099236641224e-05,
+      "loss": 0.646,
+      "step": 471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8015267175572518e-05,
+      "loss": 0.6471,
+      "step": 472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.805343511450382e-05,
+      "loss": 0.6409,
+      "step": 473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8091603053435117e-05,
+      "loss": 0.7415,
+      "step": 474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8129770992366415e-05,
+      "loss": 0.6592,
+      "step": 475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.816793893129771e-05,
+      "loss": 0.6461,
+      "step": 476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.820610687022901e-05,
+      "loss": 0.6734,
+      "step": 477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8244274809160308e-05,
+      "loss": 0.6446,
+      "step": 478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8282442748091606e-05,
+      "loss": 0.638,
+      "step": 479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.83206106870229e-05,
+      "loss": 0.7217,
+      "step": 480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8358778625954198e-05,
+      "loss": 0.6283,
+      "step": 481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.83969465648855e-05,
+      "loss": 0.6524,
+      "step": 482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8435114503816797e-05,
+      "loss": 0.6753,
+      "step": 483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847328244274809e-05,
+      "loss": 0.6446,
+      "step": 484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.851145038167939e-05,
+      "loss": 0.6554,
+      "step": 485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.854961832061069e-05,
+      "loss": 0.697,
+      "step": 486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8587786259541988e-05,
+      "loss": 0.691,
+      "step": 487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625954198473282e-05,
+      "loss": 0.7463,
+      "step": 488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866412213740458e-05,
+      "loss": 0.6833,
+      "step": 489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702290076335878e-05,
+      "loss": 0.6978,
+      "step": 490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874045801526718e-05,
+      "loss": 0.706,
+      "step": 491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8778625954198473e-05,
+      "loss": 0.6699,
+      "step": 492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.881679389312977e-05,
+      "loss": 0.7009,
+      "step": 493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.885496183206107e-05,
+      "loss": 0.6181,
+      "step": 494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.889312977099237e-05,
+      "loss": 0.6961,
+      "step": 495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8931297709923668e-05,
+      "loss": 0.6711,
+      "step": 496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8969465648854962e-05,
+      "loss": 0.7193,
+      "step": 497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.900763358778626e-05,
+      "loss": 0.6476,
+      "step": 498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9045801526717558e-05,
+      "loss": 0.6291,
+      "step": 499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.908396946564886e-05,
+      "loss": 0.6468,
+      "step": 500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9122137404580153e-05,
+      "loss": 0.631,
+      "step": 501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.916030534351145e-05,
+      "loss": 0.6943,
+      "step": 502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.919847328244275e-05,
+      "loss": 0.724,
+      "step": 503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.923664122137405e-05,
+      "loss": 0.6244,
+      "step": 504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9274809160305344e-05,
+      "loss": 0.7207,
+      "step": 505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9312977099236642e-05,
+      "loss": 0.6746,
+      "step": 506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.935114503816794e-05,
+      "loss": 0.7018,
+      "step": 507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9389312977099238e-05,
+      "loss": 0.6392,
+      "step": 508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9427480916030535e-05,
+      "loss": 0.6812,
+      "step": 509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9465648854961833e-05,
+      "loss": 0.6629,
+      "step": 510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.950381679389313e-05,
+      "loss": 0.7177,
+      "step": 511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.954198473282443e-05,
+      "loss": 0.6646,
+      "step": 512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9580152671755726e-05,
+      "loss": 0.6768,
+      "step": 513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9618320610687024e-05,
+      "loss": 0.7424,
+      "step": 514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9656488549618322e-05,
+      "loss": 0.6169,
+      "step": 515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969465648854962e-05,
+      "loss": 0.6917,
+      "step": 516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9732824427480917e-05,
+      "loss": 0.6485,
+      "step": 517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9770992366412215e-05,
+      "loss": 0.731,
+      "step": 518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9809160305343513e-05,
+      "loss": 0.6655,
+      "step": 519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.984732824427481e-05,
+      "loss": 0.6842,
+      "step": 520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.988549618320611e-05,
+      "loss": 0.6336,
+      "step": 521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923664122137406e-05,
+      "loss": 0.6646,
+      "step": 522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9961832061068704e-05,
+      "loss": 0.6533,
+      "step": 523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.675,
+      "step": 524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999999778117965e-05,
+      "loss": 0.7152,
+      "step": 525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999991124719578e-05,
+      "loss": 0.694,
+      "step": 526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999998003062274e-05,
+      "loss": 0.6904,
+      "step": 527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999964498894057e-05,
+      "loss": 0.6623,
+      "step": 528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999944529540426e-05,
+      "loss": 0.6764,
+      "step": 529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999920122570705e-05,
+      "loss": 0.6668,
+      "step": 530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999989127799573e-05,
+      "loss": 0.6731,
+      "step": 531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999857995828295e-05,
+      "loss": 0.6028,
+      "step": 532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999820276083168e-05,
+      "loss": 0.6436,
+      "step": 533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999778118777097e-05,
+      "loss": 0.6433,
+      "step": 534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999731523928783e-05,
+      "loss": 0.6688,
+      "step": 535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99996804915589e-05,
+      "loss": 0.6163,
+      "step": 536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999625021690104e-05,
+      "loss": 0.6936,
+      "step": 537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9999565114347003e-05,
+      "loss": 0.6195,
+      "step": 538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9999500769556186e-05,
+      "loss": 0.7011,
+      "step": 539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9999431987346206e-05,
+      "loss": 0.6416,
+      "step": 540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9999358767747582e-05,
+      "loss": 0.6063,
+      "step": 541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999928111079281e-05,
+      "loss": 0.6231,
+      "step": 542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999919901651635e-05,
+      "loss": 0.631,
+      "step": 543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9999112484954635e-05,
+      "loss": 0.6703,
+      "step": 544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999902151614606e-05,
+      "loss": 0.6683,
+      "step": 545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9998926110130997e-05,
+      "loss": 0.6601,
+      "step": 546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9998826266951788e-05,
+      "loss": 0.6456,
+      "step": 547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999872198665273e-05,
+      "loss": 0.7496,
+      "step": 548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9998613269280106e-05,
+      "loss": 0.6868,
+      "step": 549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999850011488216e-05,
+      "loss": 0.6486,
+      "step": 550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99983825235091e-05,
+      "loss": 0.6271,
+      "step": 551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999826049521312e-05,
+      "loss": 0.6237,
+      "step": 552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999813403004836e-05,
+      "loss": 0.642,
+      "step": 553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9998003128070947e-05,
+      "loss": 0.6128,
+      "step": 554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999786778933897e-05,
+      "loss": 0.642,
+      "step": 555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9997728013912486e-05,
+      "loss": 0.6405,
+      "step": 556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999758380185352e-05,
+      "loss": 0.638,
+      "step": 557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999743515322608e-05,
+      "loss": 0.6327,
+      "step": 558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9997282068096116e-05,
+      "loss": 0.6324,
+      "step": 559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999712454653157e-05,
+      "loss": 0.6814,
+      "step": 560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9996962588602342e-05,
+      "loss": 0.6165,
+      "step": 561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9996796194380303e-05,
+      "loss": 0.634,
+      "step": 562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9996625363939296e-05,
+      "loss": 0.6041,
+      "step": 563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999645009735512e-05,
+      "loss": 0.5931,
+      "step": 564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9996270394705564e-05,
+      "loss": 0.7148,
+      "step": 565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999608625607037e-05,
+      "loss": 0.6413,
+      "step": 566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9995897681531245e-05,
+      "loss": 0.6689,
+      "step": 567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9995704671171882e-05,
+      "loss": 0.6903,
+      "step": 568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9995507225077926e-05,
+      "loss": 0.6742,
+      "step": 569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9995305343336997e-05,
+      "loss": 0.6847,
+      "step": 570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9995099026038686e-05,
+      "loss": 0.617,
+      "step": 571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999488827327454e-05,
+      "loss": 0.7025,
+      "step": 572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9994673085138093e-05,
+      "loss": 0.6213,
+      "step": 573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999445346172484e-05,
+      "loss": 0.6793,
+      "step": 574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999422940313223e-05,
+      "loss": 0.587,
+      "step": 575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99940009094597e-05,
+      "loss": 0.6415,
+      "step": 576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999376798080865e-05,
+      "loss": 0.6606,
+      "step": 577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9993530617282436e-05,
+      "loss": 0.6329,
+      "step": 578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9993288818986403e-05,
+      "loss": 0.6441,
+      "step": 579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9993042586027844e-05,
+      "loss": 0.6704,
+      "step": 580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999279191851603e-05,
+      "loss": 0.6695,
+      "step": 581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99925368165622e-05,
+      "loss": 0.6384,
+      "step": 582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999227728027956e-05,
+      "loss": 0.6346,
+      "step": 583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999201330978328e-05,
+      "loss": 0.6595,
+      "step": 584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99917449051905e-05,
+      "loss": 0.6424,
+      "step": 585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999147206662033e-05,
+      "loss": 0.6234,
+      "step": 586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991194794193846e-05,
+      "loss": 0.6097,
+      "step": 587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990913088034093e-05,
+      "loss": 0.7177,
+      "step": 588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999062694826608e-05,
+      "loss": 0.6711,
+      "step": 589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9990336375016786e-05,
+      "loss": 0.678,
+      "step": 590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9990041368415153e-05,
+      "loss": 0.6494,
+      "step": 591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9989741928592105e-05,
+      "loss": 0.6533,
+      "step": 592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9989438055680513e-05,
+      "loss": 0.6523,
+      "step": 593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9989129749815226e-05,
+      "loss": 0.5832,
+      "step": 594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988817011133063e-05,
+      "loss": 0.6541,
+      "step": 595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988499839772805e-05,
+      "loss": 0.6211,
+      "step": 596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99881782358752e-05,
+      "loss": 0.6482,
+      "step": 597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987852199582966e-05,
+      "loss": 0.7315,
+      "step": 598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998752173104078e-05,
+      "loss": 0.656,
+      "step": 599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99871868303953e-05,
+      "loss": 0.5615,
+      "step": 600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998684749779514e-05,
+      "loss": 0.6179,
+      "step": 601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986503733390884e-05,
+      "loss": 0.5909,
+      "step": 602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986155537335078e-05,
+      "loss": 0.6408,
+      "step": 603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985802909782246e-05,
+      "loss": 0.6249,
+      "step": 604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985445850888868e-05,
+      "loss": 0.69,
+      "step": 605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998508436081339e-05,
+      "loss": 0.6214,
+      "step": 606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9984718439716237e-05,
+      "loss": 0.5996,
+      "step": 607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9984348087759785e-05,
+      "loss": 0.6103,
+      "step": 608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9983973305108386e-05,
+      "loss": 0.6749,
+      "step": 609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998359409192835e-05,
+      "loss": 0.684,
+      "step": 610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9983210448387966e-05,
+      "loss": 0.6629,
+      "step": 611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9982822374657477e-05,
+      "loss": 0.6681,
+      "step": 612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9982429870909095e-05,
+      "loss": 0.6032,
+      "step": 613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9982032937317e-05,
+      "loss": 0.6059,
+      "step": 614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998163157405734e-05,
+      "loss": 0.6441,
+      "step": 615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998122578130822e-05,
+      "loss": 0.7159,
+      "step": 616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998081555924972e-05,
+      "loss": 0.678,
+      "step": 617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998040090806388e-05,
+      "loss": 0.678,
+      "step": 618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9979981827934712e-05,
+      "loss": 0.6327,
+      "step": 619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9979558319048183e-05,
+      "loss": 0.6926,
+      "step": 620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.997913038159223e-05,
+      "loss": 0.6721,
+      "step": 621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9978698015756765e-05,
+      "loss": 0.6277,
+      "step": 622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9978261221733644e-05,
+      "loss": 0.6651,
+      "step": 623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9977819999716713e-05,
+      "loss": 0.6678,
+      "step": 624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.997737434990176e-05,
+      "loss": 0.6474,
+      "step": 625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9976924272486556e-05,
+      "loss": 0.658,
+      "step": 626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9976469767670825e-05,
+      "loss": 0.6526,
+      "step": 627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9976010835656262e-05,
+      "loss": 0.6519,
+      "step": 628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9975547476646524e-05,
+      "loss": 0.6149,
+      "step": 629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.997507969084723e-05,
+      "loss": 0.6291,
+      "step": 630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9974607478465974e-05,
+      "loss": 0.5968,
+      "step": 631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99741308397123e-05,
+      "loss": 0.5664,
+      "step": 632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9973649774797725e-05,
+      "loss": 0.6016,
+      "step": 633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.997316428393573e-05,
+      "loss": 0.6013,
+      "step": 634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9972674367341755e-05,
+      "loss": 0.609,
+      "step": 635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.997218002523321e-05,
+      "loss": 0.657,
+      "step": 636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9971681257829465e-05,
+      "loss": 0.6576,
+      "step": 637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9971178065351857e-05,
+      "loss": 0.6376,
+      "step": 638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9970670448023684e-05,
+      "loss": 0.6731,
+      "step": 639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9970158406070206e-05,
+      "loss": 0.6157,
+      "step": 640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.996964193971865e-05,
+      "loss": 0.6686,
+      "step": 641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.996912104919821e-05,
+      "loss": 0.7007,
+      "step": 642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.996859573474003e-05,
+      "loss": 0.6288,
+      "step": 643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9968065996577235e-05,
+      "loss": 0.6229,
+      "step": 644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9967531834944896e-05,
+      "loss": 0.6784,
+      "step": 645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9966993250080055e-05,
+      "loss": 0.6514,
+      "step": 646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9966450242221722e-05,
+      "loss": 0.6809,
+      "step": 647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.996590281161086e-05,
+      "loss": 0.7369,
+      "step": 648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99653509584904e-05,
+      "loss": 0.6533,
+      "step": 649
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.996479468310524e-05,
+      "loss": 0.7027,
+      "step": 650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9964233985702227e-05,
+      "loss": 0.6463,
+      "step": 651
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.996366886653018e-05,
+      "loss": 0.612,
+      "step": 652
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9963099325839883e-05,
+      "loss": 0.6227,
+      "step": 653
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9962525363884073e-05,
+      "loss": 0.7113,
+      "step": 654
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9961946980917457e-05,
+      "loss": 0.6056,
+      "step": 655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9961364177196697e-05,
+      "loss": 0.6209,
+      "step": 656
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9960776952980426e-05,
+      "loss": 0.6708,
+      "step": 657
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9960185308529225e-05,
+      "loss": 0.6136,
+      "step": 658
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.995958924410565e-05,
+      "loss": 0.6762,
+      "step": 659
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9958988759974213e-05,
+      "loss": 0.6062,
+      "step": 660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9958383856401388e-05,
+      "loss": 0.6602,
+      "step": 661
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9957774533655608e-05,
+      "loss": 0.5912,
+      "step": 662
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9957160792007264e-05,
+      "loss": 0.6087,
+      "step": 663
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.995654263172872e-05,
+      "loss": 0.6664,
+      "step": 664
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.995592005309429e-05,
+      "loss": 0.7531,
+      "step": 665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.995529305638025e-05,
+      "loss": 0.6175,
+      "step": 666
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9954661641864844e-05,
+      "loss": 0.6575,
+      "step": 667
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9954025809828266e-05,
+      "loss": 0.5938,
+      "step": 668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9953385560552676e-05,
+      "loss": 0.6492,
+      "step": 669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99527408943222e-05,
+      "loss": 0.674,
+      "step": 670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9952091811422907e-05,
+      "loss": 0.6642,
+      "step": 671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9951438312142843e-05,
+      "loss": 0.6321,
+      "step": 672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9950780396772005e-05,
+      "loss": 0.6046,
+      "step": 673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.995011806560236e-05,
+      "loss": 0.6502,
+      "step": 674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.994945131892782e-05,
+      "loss": 0.6395,
+      "step": 675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.994878015704426e-05,
+      "loss": 0.6214,
+      "step": 676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.994810458024952e-05,
+      "loss": 0.6278,
+      "step": 677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99474245888434e-05,
+      "loss": 0.6323,
+      "step": 678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9946740183127655e-05,
+      "loss": 0.6071,
+      "step": 679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9946051363406e-05,
+      "loss": 0.6478,
+      "step": 680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9945358129984106e-05,
+      "loss": 0.6742,
+      "step": 681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9944660483169605e-05,
+      "loss": 0.6152,
+      "step": 682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.994395842327209e-05,
+      "loss": 0.6046,
+      "step": 683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9943251950603107e-05,
+      "loss": 0.6206,
+      "step": 684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9942541065476165e-05,
+      "loss": 0.6859,
+      "step": 685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.994182576820673e-05,
+      "loss": 0.6388,
+      "step": 686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9941106059112225e-05,
+      "loss": 0.6105,
+      "step": 687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.994038193851203e-05,
+      "loss": 0.6242,
+      "step": 688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9939653406727484e-05,
+      "loss": 0.6368,
+      "step": 689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9938920464081882e-05,
+      "loss": 0.6899,
+      "step": 690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993818311090048e-05,
+      "loss": 0.6065,
+      "step": 691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993744134751049e-05,
+      "loss": 0.6605,
+      "step": 692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9936695174241075e-05,
+      "loss": 0.6331,
+      "step": 693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9935944591423365e-05,
+      "loss": 0.6217,
+      "step": 694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9935189599390436e-05,
+      "loss": 0.616,
+      "step": 695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.993443019847733e-05,
+      "loss": 0.7022,
+      "step": 696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9933666389021046e-05,
+      "loss": 0.6872,
+      "step": 697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9932898171360526e-05,
+      "loss": 0.635,
+      "step": 698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.993212554583668e-05,
+      "loss": 0.6455,
+      "step": 699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.993134851279238e-05,
+      "loss": 0.6556,
+      "step": 700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9930567072572436e-05,
+      "loss": 0.6353,
+      "step": 701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9929781225523625e-05,
+      "loss": 0.5979,
+      "step": 702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9928990971994682e-05,
+      "loss": 0.6357,
+      "step": 703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9928196312336287e-05,
+      "loss": 0.6607,
+      "step": 704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9927397246901085e-05,
+      "loss": 0.6446,
+      "step": 705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.992659377604367e-05,
+      "loss": 0.6804,
+      "step": 706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99257859001206e-05,
+      "loss": 0.6634,
+      "step": 707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.992497361949037e-05,
+      "loss": 0.7019,
+      "step": 708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.992415693451345e-05,
+      "loss": 0.6107,
+      "step": 709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.992333584555225e-05,
+      "loss": 0.6078,
+      "step": 710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.992251035297115e-05,
+      "loss": 0.6014,
+      "step": 711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9921680457136463e-05,
+      "loss": 0.6087,
+      "step": 712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.992084615841647e-05,
+      "loss": 0.7108,
+      "step": 713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9920007457181403e-05,
+      "loss": 0.6311,
+      "step": 714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991916435380345e-05,
+      "loss": 0.6016,
+      "step": 715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9918316848656746e-05,
+      "loss": 0.6552,
+      "step": 716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9917464942117385e-05,
+      "loss": 0.634,
+      "step": 717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9916608634563413e-05,
+      "loss": 0.6712,
+      "step": 718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991574792637483e-05,
+      "loss": 0.6751,
+      "step": 719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9914882817933583e-05,
+      "loss": 0.6533,
+      "step": 720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9914013309623578e-05,
+      "loss": 0.6947,
+      "step": 721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9913139401830676e-05,
+      "loss": 0.5875,
+      "step": 722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.991226109494268e-05,
+      "loss": 0.6489,
+      "step": 723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.991137838934935e-05,
+      "loss": 0.6207,
+      "step": 724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.991049128544241e-05,
+      "loss": 0.6152,
+      "step": 725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9909599783615513e-05,
+      "loss": 0.6237,
+      "step": 726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9908703884264277e-05,
+      "loss": 0.6543,
+      "step": 727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907803587786273e-05,
+      "loss": 0.6362,
+      "step": 728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906898894581022e-05,
+      "loss": 0.694,
+      "step": 729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990598980504999e-05,
+      "loss": 0.6403,
+      "step": 730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.99050763195966e-05,
+      "loss": 0.5945,
+      "step": 731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9904158438626227e-05,
+      "loss": 0.62,
+      "step": 732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990323616254619e-05,
+      "loss": 0.6208,
+      "step": 733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9902309491765762e-05,
+      "loss": 0.6635,
+      "step": 734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9901378426696166e-05,
+      "loss": 0.6384,
+      "step": 735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9900442967750577e-05,
+      "loss": 0.6295,
+      "step": 736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989950311534412e-05,
+      "loss": 0.6706,
+      "step": 737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989855886989386e-05,
+      "loss": 0.6646,
+      "step": 738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9897610231818828e-05,
+      "loss": 0.6189,
+      "step": 739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989665720153999e-05,
+      "loss": 0.6002,
+      "step": 740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989569977948027e-05,
+      "loss": 0.6572,
+      "step": 741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9894737966064538e-05,
+      "loss": 0.6512,
+      "step": 742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9893771761719606e-05,
+      "loss": 0.686,
+      "step": 743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9892801166874245e-05,
+      "loss": 0.6378,
+      "step": 744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989182618195917e-05,
+      "loss": 0.6401,
+      "step": 745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9890846807407044e-05,
+      "loss": 0.6221,
+      "step": 746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9889863043652483e-05,
+      "loss": 0.6096,
+      "step": 747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9888874891132038e-05,
+      "loss": 0.6242,
+      "step": 748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9887882350284218e-05,
+      "loss": 0.5564,
+      "step": 749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.988688542154948e-05,
+      "loss": 0.6491,
+      "step": 750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9885884105370223e-05,
+      "loss": 0.6426,
+      "step": 751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9884878402190796e-05,
+      "loss": 0.7084,
+      "step": 752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.988386831245749e-05,
+      "loss": 0.6079,
+      "step": 753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9882853836618556e-05,
+      "loss": 0.5924,
+      "step": 754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.988183497512417e-05,
+      "loss": 0.7015,
+      "step": 755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9880811728426476e-05,
+      "loss": 0.6165,
+      "step": 756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9879784096979545e-05,
+      "loss": 0.6099,
+      "step": 757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9878752081239413e-05,
+      "loss": 0.6401,
+      "step": 758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9877715681664044e-05,
+      "loss": 0.6317,
+      "step": 759
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9876674898713356e-05,
+      "loss": 0.5897,
+      "step": 760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9875629732849216e-05,
+      "loss": 0.6204,
+      "step": 761
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9874580184535424e-05,
+      "loss": 0.6611,
+      "step": 762
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.987352625423774e-05,
+      "loss": 0.5813,
+      "step": 763
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9872467942423853e-05,
+      "loss": 0.59,
+      "step": 764
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9871405249563406e-05,
+      "loss": 0.6445,
+      "step": 765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9870338176127985e-05,
+      "loss": 0.5749,
+      "step": 766
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9869266722591115e-05,
+      "loss": 0.6778,
+      "step": 767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9868190889428274e-05,
+      "loss": 0.7069,
+      "step": 768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9867110677116877e-05,
+      "loss": 0.6411,
+      "step": 769
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.986602608613628e-05,
+      "loss": 0.6022,
+      "step": 770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.986493711696779e-05,
+      "loss": 0.6131,
+      "step": 771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.986384377009465e-05,
+      "loss": 0.6137,
+      "step": 772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.986274604600205e-05,
+      "loss": 0.6296,
+      "step": 773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9861643945177117e-05,
+      "loss": 0.5844,
+      "step": 774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.986053746810892e-05,
+      "loss": 0.6419,
+      "step": 775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.985942661528849e-05,
+      "loss": 0.5716,
+      "step": 776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9858311387208765e-05,
+      "loss": 0.6558,
+      "step": 777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9857191784364655e-05,
+      "loss": 0.6027,
+      "step": 778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.985606780725299e-05,
+      "loss": 0.6033,
+      "step": 779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9854939456372562e-05,
+      "loss": 0.5953,
+      "step": 780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9853806732224085e-05,
+      "loss": 0.6244,
+      "step": 781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9852669635310224e-05,
+      "loss": 0.6303,
+      "step": 782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.985152816613558e-05,
+      "loss": 0.6179,
+      "step": 783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9850382325206695e-05,
+      "loss": 0.6244,
+      "step": 784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9849232113032055e-05,
+      "loss": 0.6646,
+      "step": 785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.6677,
+      "step": 786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9846918576989137e-05,
+      "loss": 0.5989,
+      "step": 787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9845755254147526e-05,
+      "loss": 0.5956,
+      "step": 788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9844587562113484e-05,
+      "loss": 0.6314,
+      "step": 789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9843415501405193e-05,
+      "loss": 0.6887,
+      "step": 790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9842239072542776e-05,
+      "loss": 0.662,
+      "step": 791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9841058276048282e-05,
+      "loss": 0.6648,
+      "step": 792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.983987311244571e-05,
+      "loss": 0.6299,
+      "step": 793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9838683582260992e-05,
+      "loss": 0.5947,
+      "step": 794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9837489686022004e-05,
+      "loss": 0.6253,
+      "step": 795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9836291424258546e-05,
+      "loss": 0.6734,
+      "step": 796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.983508879750237e-05,
+      "loss": 0.5632,
+      "step": 797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9833881806287154e-05,
+      "loss": 0.594,
+      "step": 798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.983267045114852e-05,
+      "loss": 0.5994,
+      "step": 799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9831454732624023e-05,
+      "loss": 0.695,
+      "step": 800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9830234651253155e-05,
+      "loss": 0.5721,
+      "step": 801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9829010207577343e-05,
+      "loss": 0.5928,
+      "step": 802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.982778140213996e-05,
+      "loss": 0.6701,
+      "step": 803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.982654823548629e-05,
+      "loss": 0.6113,
+      "step": 804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.982531070816358e-05,
+      "loss": 0.5796,
+      "step": 805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9824068820720995e-05,
+      "loss": 0.6435,
+      "step": 806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9822822573709645e-05,
+      "loss": 0.6184,
+      "step": 807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9821571967682566e-05,
+      "loss": 0.6529,
+      "step": 808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9820317003194728e-05,
+      "loss": 0.6532,
+      "step": 809
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9819057680803047e-05,
+      "loss": 0.6452,
+      "step": 810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9817794001066362e-05,
+      "loss": 0.6022,
+      "step": 811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9816525964545447e-05,
+      "loss": 0.6915,
+      "step": 812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9815253571803013e-05,
+      "loss": 0.6242,
+      "step": 813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.98139768234037e-05,
+      "loss": 0.5943,
+      "step": 814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9812695719914085e-05,
+      "loss": 0.6327,
+      "step": 815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9811410261902678e-05,
+      "loss": 0.627,
+      "step": 816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.981012044993991e-05,
+      "loss": 0.6464,
+      "step": 817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9808826284598165e-05,
+      "loss": 0.5747,
+      "step": 818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.980752776645174e-05,
+      "loss": 0.5939,
+      "step": 819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9806224896076872e-05,
+      "loss": 0.6222,
+      "step": 820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.980491767405173e-05,
+      "loss": 0.5835,
+      "step": 821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9803606100956408e-05,
+      "loss": 0.5918,
+      "step": 822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.980229017737294e-05,
+      "loss": 0.636,
+      "step": 823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9800969903885285e-05,
+      "loss": 0.6625,
+      "step": 824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9799645281079324e-05,
+      "loss": 0.6287,
+      "step": 825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.979831630954289e-05,
+      "loss": 0.6455,
+      "step": 826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9796982989865722e-05,
+      "loss": 0.6668,
+      "step": 827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.979564532263951e-05,
+      "loss": 0.6012,
+      "step": 828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9794303308457855e-05,
+      "loss": 0.6523,
+      "step": 829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9792956947916294e-05,
+      "loss": 0.6467,
+      "step": 830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9791606241612298e-05,
+      "loss": 0.6595,
+      "step": 831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9790251190145258e-05,
+      "loss": 0.6461,
+      "step": 832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.97888917941165e-05,
+      "loss": 0.6511,
+      "step": 833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9787528054129278e-05,
+      "loss": 0.5986,
+      "step": 834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9786159970788762e-05,
+      "loss": 0.6799,
+      "step": 835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9784787544702065e-05,
+      "loss": 0.6052,
+      "step": 836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9783410776478218e-05,
+      "loss": 0.6322,
+      "step": 837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9782029666728185e-05,
+      "loss": 0.6824,
+      "step": 838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9780644216064846e-05,
+      "loss": 0.6166,
+      "step": 839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9779254425103017e-05,
+      "loss": 0.6551,
+      "step": 840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9777860294459444e-05,
+      "loss": 0.6136,
+      "step": 841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9776461824752782e-05,
+      "loss": 0.5657,
+      "step": 842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.977505901660363e-05,
+      "loss": 0.6589,
+      "step": 843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9773651870634496e-05,
+      "loss": 0.6617,
+      "step": 844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9772240387469824e-05,
+      "loss": 0.6173,
+      "step": 845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9770824567735982e-05,
+      "loss": 0.7058,
+      "step": 846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.976940441206126e-05,
+      "loss": 0.5862,
+      "step": 847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.976797992107587e-05,
+      "loss": 0.6785,
+      "step": 848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9766551095411947e-05,
+      "loss": 0.6248,
+      "step": 849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9765117935703556e-05,
+      "loss": 0.6456,
+      "step": 850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9763680442586682e-05,
+      "loss": 0.6649,
+      "step": 851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9762238616699233e-05,
+      "loss": 0.6346,
+      "step": 852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9760792458681037e-05,
+      "loss": 0.5732,
+      "step": 853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9759341969173853e-05,
+      "loss": 0.5976,
+      "step": 854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9757887148821348e-05,
+      "loss": 0.6572,
+      "step": 855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9756427998269125e-05,
+      "loss": 0.6683,
+      "step": 856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.97549645181647e-05,
+      "loss": 0.6143,
+      "step": 857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.975349670915751e-05,
+      "loss": 0.5752,
+      "step": 858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.975202457189892e-05,
+      "loss": 0.5927,
+      "step": 859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9750548107042215e-05,
+      "loss": 0.6626,
+      "step": 860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.974906731524259e-05,
+      "loss": 0.5798,
+      "step": 861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.974758219715717e-05,
+      "loss": 0.5811,
+      "step": 862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9746092753444997e-05,
+      "loss": 0.6402,
+      "step": 863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9744598984767035e-05,
+      "loss": 0.6115,
+      "step": 864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9743100891786157e-05,
+      "loss": 0.6226,
+      "step": 865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9741598475167173e-05,
+      "loss": 0.6421,
+      "step": 866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9740091735576798e-05,
+      "loss": 0.5813,
+      "step": 867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9738580673683666e-05,
+      "loss": 0.6112,
+      "step": 868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9737065290158335e-05,
+      "loss": 0.6189,
+      "step": 869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.973554558567327e-05,
+      "loss": 0.5811,
+      "step": 870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9734021560902875e-05,
+      "loss": 0.5913,
+      "step": 871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.973249321652345e-05,
+      "loss": 0.6324,
+      "step": 872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.973096055321322e-05,
+      "loss": 0.5925,
+      "step": 873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.972942357165232e-05,
+      "loss": 0.6393,
+      "step": 874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9727882272522817e-05,
+      "loss": 0.5978,
+      "step": 875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.972633665650868e-05,
+      "loss": 0.5587,
+      "step": 876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9724786724295796e-05,
+      "loss": 0.5714,
+      "step": 877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9723232476571973e-05,
+      "loss": 0.6057,
+      "step": 878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9721673914026925e-05,
+      "loss": 0.6501,
+      "step": 879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.972011103735229e-05,
+      "loss": 0.579,
+      "step": 880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.971854384724162e-05,
+      "loss": 0.6502,
+      "step": 881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.971697234439037e-05,
+      "loss": 0.6125,
+      "step": 882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.971539652949592e-05,
+      "loss": 0.5992,
+      "step": 883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.971381640325756e-05,
+      "loss": 0.6065,
+      "step": 884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9712231966376497e-05,
+      "loss": 0.6415,
+      "step": 885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9710643219555837e-05,
+      "loss": 0.5923,
+      "step": 886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.970905016350062e-05,
+      "loss": 0.6196,
+      "step": 887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.970745279891778e-05,
+      "loss": 0.6378,
+      "step": 888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9705851126516175e-05,
+      "loss": 0.6307,
+      "step": 889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9704245147006566e-05,
+      "loss": 0.6738,
+      "step": 890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9702634861101626e-05,
+      "loss": 0.5887,
+      "step": 891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9701020269515948e-05,
+      "loss": 0.6441,
+      "step": 892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.969940137296603e-05,
+      "loss": 0.6413,
+      "step": 893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9697778172170274e-05,
+      "loss": 0.6386,
+      "step": 894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9696150667849002e-05,
+      "loss": 0.6821,
+      "step": 895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9694518860724444e-05,
+      "loss": 0.5946,
+      "step": 896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9692882751520736e-05,
+      "loss": 0.5763,
+      "step": 897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9691242340963925e-05,
+      "loss": 0.614,
+      "step": 898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9689597629781963e-05,
+      "loss": 0.6724,
+      "step": 899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9687948618704713e-05,
+      "loss": 0.5985,
+      "step": 900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9686295308463956e-05,
+      "loss": 0.6253,
+      "step": 901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9684637699793358e-05,
+      "loss": 0.6385,
+      "step": 902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9682975793428517e-05,
+      "loss": 0.6062,
+      "step": 903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9681309590106925e-05,
+      "loss": 0.6176,
+      "step": 904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9679639090567982e-05,
+      "loss": 0.6461,
+      "step": 905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9677964295552992e-05,
+      "loss": 0.602,
+      "step": 906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9676285205805173e-05,
+      "loss": 0.6463,
+      "step": 907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.967460182206965e-05,
+      "loss": 0.6211,
+      "step": 908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9672914145093438e-05,
+      "loss": 0.6557,
+      "step": 909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.967122217562547e-05,
+      "loss": 0.5718,
+      "step": 910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9669525914416587e-05,
+      "loss": 0.6793,
+      "step": 911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9667825362219525e-05,
+      "loss": 0.5685,
+      "step": 912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9666120519788927e-05,
+      "loss": 0.5507,
+      "step": 913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9664411387881338e-05,
+      "loss": 0.5811,
+      "step": 914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.966269796725522e-05,
+      "loss": 0.6251,
+      "step": 915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.966098025867091e-05,
+      "loss": 0.6195,
+      "step": 916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.5829,
+      "step": 917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.965753198067869e-05,
+      "loss": 0.6471,
+      "step": 918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9655801412801e-05,
+      "loss": 0.6249,
+      "step": 919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9654066560025566e-05,
+      "loss": 0.5882,
+      "step": 920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.965232742312226e-05,
+      "loss": 0.5866,
+      "step": 921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9650584002862847e-05,
+      "loss": 0.6478,
+      "step": 922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9648836300020993e-05,
+      "loss": 0.6297,
+      "step": 923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.964708431537227e-05,
+      "loss": 0.6165,
+      "step": 924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9645328049694142e-05,
+      "loss": 0.6138,
+      "step": 925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.964356750376598e-05,
+      "loss": 0.5789,
+      "step": 926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9641802678369047e-05,
+      "loss": 0.6097,
+      "step": 927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.964003357428651e-05,
+      "loss": 0.6003,
+      "step": 928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9638260192303437e-05,
+      "loss": 0.6156,
+      "step": 929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9636482533206788e-05,
+      "loss": 0.6011,
+      "step": 930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9634700597785425e-05,
+      "loss": 0.6821,
+      "step": 931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.963291438683011e-05,
+      "loss": 0.6417,
+      "step": 932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9631123901133493e-05,
+      "loss": 0.6378,
+      "step": 933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.962932914149013e-05,
+      "loss": 0.6089,
+      "step": 934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9627530108696472e-05,
+      "loss": 0.6094,
+      "step": 935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9625726803550865e-05,
+      "loss": 0.5557,
+      "step": 936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9623919226853552e-05,
+      "loss": 0.6042,
+      "step": 937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9622107379406666e-05,
+      "loss": 0.5797,
+      "step": 938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9620291262014244e-05,
+      "loss": 0.5716,
+      "step": 939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9618470875482216e-05,
+      "loss": 0.607,
+      "step": 940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.96166462206184e-05,
+      "loss": 0.6259,
+      "step": 941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.961481729823251e-05,
+      "loss": 0.5872,
+      "step": 942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.961298410913616e-05,
+      "loss": 0.6063,
+      "step": 943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9611146654142856e-05,
+      "loss": 0.6492,
+      "step": 944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9609304934067984e-05,
+      "loss": 0.6724,
+      "step": 945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9607458949728845e-05,
+      "loss": 0.5899,
+      "step": 946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9605608701944615e-05,
+      "loss": 0.6374,
+      "step": 947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.960375419153637e-05,
+      "loss": 0.6152,
+      "step": 948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.960189541932707e-05,
+      "loss": 0.5607,
+      "step": 949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9600032386141578e-05,
+      "loss": 0.6251,
+      "step": 950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9598165092806634e-05,
+      "loss": 0.6585,
+      "step": 951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9596293540150883e-05,
+      "loss": 0.6581,
+      "step": 952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9594417729004845e-05,
+      "loss": 0.5726,
+      "step": 953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959253766020095e-05,
+      "loss": 0.649,
+      "step": 954
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590653334573488e-05,
+      "loss": 0.67,
+      "step": 955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958876475295867e-05,
+      "loss": 0.6012,
+      "step": 956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.958687191619457e-05,
+      "loss": 0.6291,
+      "step": 957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.958497482512117e-05,
+      "loss": 0.6294,
+      "step": 958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9583073480580323e-05,
+      "loss": 0.6411,
+      "step": 959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9581167883415785e-05,
+      "loss": 0.5974,
+      "step": 960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9579258034473183e-05,
+      "loss": 0.5691,
+      "step": 961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9577343934600048e-05,
+      "loss": 0.6929,
+      "step": 962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.957542558464578e-05,
+      "loss": 0.6309,
+      "step": 963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.957350298546168e-05,
+      "loss": 0.5652,
+      "step": 964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9571576137900926e-05,
+      "loss": 0.675,
+      "step": 965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9569645042818582e-05,
+      "loss": 0.6323,
+      "step": 966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9567709701071607e-05,
+      "loss": 0.6447,
+      "step": 967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9565770113518827e-05,
+      "loss": 0.6556,
+      "step": 968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9563826281020963e-05,
+      "loss": 0.6294,
+      "step": 969
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561878204440624e-05,
+      "loss": 0.6364,
+      "step": 970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559925884642288e-05,
+      "loss": 0.6309,
+      "step": 971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557969322492328e-05,
+      "loss": 0.6224,
+      "step": 972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556008518859e-05,
+      "loss": 0.6727,
+      "step": 973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955404347461243e-05,
+      "loss": 0.6554,
+      "step": 974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955207419062464e-05,
+      "loss": 0.642,
+      "step": 975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550100667769528e-05,
+      "loss": 0.618,
+      "step": 976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548122906922866e-05,
+      "loss": 0.6339,
+      "step": 977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954614090896232e-05,
+      "loss": 0.6326,
+      "step": 978
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544154674767428e-05,
+      "loss": 0.6513,
+      "step": 979
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542164205219605e-05,
+      "loss": 0.5742,
+      "step": 980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540169501202156e-05,
+      "loss": 0.6524,
+      "step": 981
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953817056360026e-05,
+      "loss": 0.6427,
+      "step": 982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9536167393300963e-05,
+      "loss": 0.6538,
+      "step": 983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9534159991193213e-05,
+      "loss": 0.624,
+      "step": 984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9532148358167814e-05,
+      "loss": 0.5929,
+      "step": 985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9530132495117458e-05,
+      "loss": 0.5689,
+      "step": 986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528112402936717e-05,
+      "loss": 0.6693,
+      "step": 987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952608808252203e-05,
+      "loss": 0.6218,
+      "step": 988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524059534771723e-05,
+      "loss": 0.5901,
+      "step": 989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522026760585986e-05,
+      "loss": 0.6354,
+      "step": 990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519989760866894e-05,
+      "loss": 0.6404,
+      "step": 991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95179485365184e-05,
+      "loss": 0.6196,
+      "step": 992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951590308844632e-05,
+      "loss": 0.5848,
+      "step": 993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513853417558347e-05,
+      "loss": 0.6559,
+      "step": 994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511799524764062e-05,
+      "loss": 0.6029,
+      "step": 995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95097414109749e-05,
+      "loss": 0.6159,
+      "step": 996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507679077104178e-05,
+      "loss": 0.6241,
+      "step": 997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950561252406709e-05,
+      "loss": 0.6137,
+      "step": 998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503541752780694e-05,
+      "loss": 0.6321,
+      "step": 999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950146676416393e-05,
+      "loss": 0.579,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499387559137598e-05,
+      "loss": 0.633,
+      "step": 1001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9497304138624377e-05,
+      "loss": 0.596,
+      "step": 1002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949521650354881e-05,
+      "loss": 0.6644,
+      "step": 1003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949312465483732e-05,
+      "loss": 0.6176,
+      "step": 1004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949102859341819e-05,
+      "loss": 0.6717,
+      "step": 1005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9488928320221577e-05,
+      "loss": 0.6156,
+      "step": 1006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9486823836179515e-05,
+      "loss": 0.5932,
+      "step": 1007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9484715142225886e-05,
+      "loss": 0.6316,
+      "step": 1008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948260223929646e-05,
+      "loss": 0.6322,
+      "step": 1009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480485128328866e-05,
+      "loss": 0.651,
+      "step": 1010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478363810262602e-05,
+      "loss": 0.6106,
+      "step": 1011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476238286039034e-05,
+      "loss": 0.6605,
+      "step": 1012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947410855660139e-05,
+      "loss": 0.6237,
+      "step": 1013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471974622894766e-05,
+      "loss": 0.5902,
+      "step": 1014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946983648586613e-05,
+      "loss": 0.6737,
+      "step": 1015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467694146464313e-05,
+      "loss": 0.5757,
+      "step": 1016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946554760564e-05,
+      "loss": 0.6426,
+      "step": 1017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946339686434575e-05,
+      "loss": 0.6178,
+      "step": 1018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9461241923535992e-05,
+      "loss": 0.6164,
+      "step": 1019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9459082784167004e-05,
+      "loss": 0.5456,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9456919447196935e-05,
+      "loss": 0.6164,
+      "step": 1021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94547519135858e-05,
+      "loss": 0.6236,
+      "step": 1022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.945258018429547e-05,
+      "loss": 0.6722,
+      "step": 1023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9450404260289684e-05,
+      "loss": 0.6227,
+      "step": 1024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9448224142534032e-05,
+      "loss": 0.6404,
+      "step": 1025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.944603983199598e-05,
+      "loss": 0.6105,
+      "step": 1026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.944385132964484e-05,
+      "loss": 0.5703,
+      "step": 1027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9441658636451794e-05,
+      "loss": 0.5976,
+      "step": 1028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.943946175338988e-05,
+      "loss": 0.6417,
+      "step": 1029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9437260681433993e-05,
+      "loss": 0.6846,
+      "step": 1030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9435055421560895e-05,
+      "loss": 0.5979,
+      "step": 1031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9432845974749196e-05,
+      "loss": 0.6232,
+      "step": 1032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9430632341979372e-05,
+      "loss": 0.5816,
+      "step": 1033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9428414524233753e-05,
+      "loss": 0.5965,
+      "step": 1034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942619252249653e-05,
+      "loss": 0.6382,
+      "step": 1035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942396633775374e-05,
+      "loss": 0.647,
+      "step": 1036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9421735970993288e-05,
+      "loss": 0.5703,
+      "step": 1037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9419501423204934e-05,
+      "loss": 0.6131,
+      "step": 1038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9417262695380284e-05,
+      "loss": 0.6261,
+      "step": 1039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.941501978851281e-05,
+      "loss": 0.6303,
+      "step": 1040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.941277270359783e-05,
+      "loss": 0.6282,
+      "step": 1041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.941052144163252e-05,
+      "loss": 0.5387,
+      "step": 1042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.940826600361591e-05,
+      "loss": 0.6131,
+      "step": 1043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9406006390548887e-05,
+      "loss": 0.6362,
+      "step": 1044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9403742603434176e-05,
+      "loss": 0.6175,
+      "step": 1045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.940147464327637e-05,
+      "loss": 0.6297,
+      "step": 1046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.939920251108191e-05,
+      "loss": 0.5803,
+      "step": 1047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.6096,
+      "step": 1048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9394645734618037e-05,
+      "loss": 0.5966,
+      "step": 1049
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9392361092370756e-05,
+      "loss": 0.5775,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9390072282131082e-05,
+      "loss": 0.667,
+      "step": 1051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9387779304914717e-05,
+      "loss": 0.6299,
+      "step": 1052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.938548216173919e-05,
+      "loss": 0.6139,
+      "step": 1053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9383180853623897e-05,
+      "loss": 0.6137,
+      "step": 1054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9380875381590075e-05,
+      "loss": 0.6138,
+      "step": 1055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9378565746660812e-05,
+      "loss": 0.5518,
+      "step": 1056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9376251949861035e-05,
+      "loss": 0.6533,
+      "step": 1057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9373933992217526e-05,
+      "loss": 0.6082,
+      "step": 1058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9371611874758917e-05,
+      "loss": 0.5673,
+      "step": 1059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.936928559851567e-05,
+      "loss": 0.6107,
+      "step": 1060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.936695516452011e-05,
+      "loss": 0.6171,
+      "step": 1061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.93646205738064e-05,
+      "loss": 0.6161,
+      "step": 1062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.936228182741055e-05,
+      "loss": 0.6128,
+      "step": 1063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.93599389263704e-05,
+      "loss": 0.6196,
+      "step": 1064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9357591871725656e-05,
+      "loss": 0.5769,
+      "step": 1065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9355240664517856e-05,
+      "loss": 0.6054,
+      "step": 1066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9352885305790374e-05,
+      "loss": 0.5784,
+      "step": 1067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.935052579658844e-05,
+      "loss": 0.6126,
+      "step": 1068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9348162137959117e-05,
+      "loss": 0.6263,
+      "step": 1069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9345794330951314e-05,
+      "loss": 0.6638,
+      "step": 1070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9343422376615777e-05,
+      "loss": 0.7251,
+      "step": 1071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9341046276005095e-05,
+      "loss": 0.6101,
+      "step": 1072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9338666030173694e-05,
+      "loss": 0.6234,
+      "step": 1073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.933628164017784e-05,
+      "loss": 0.5327,
+      "step": 1074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9333893107075646e-05,
+      "loss": 0.6763,
+      "step": 1075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9331500431927056e-05,
+      "loss": 0.6166,
+      "step": 1076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9329103615793846e-05,
+      "loss": 0.6436,
+      "step": 1077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9326702659739642e-05,
+      "loss": 0.6079,
+      "step": 1078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9324297564829907e-05,
+      "loss": 0.641,
+      "step": 1079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9321888332131925e-05,
+      "loss": 0.6212,
+      "step": 1080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9319474962714835e-05,
+      "loss": 0.6172,
+      "step": 1081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9317057457649598e-05,
+      "loss": 0.6027,
+      "step": 1082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9314635818009023e-05,
+      "loss": 0.6145,
+      "step": 1083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.931221004486774e-05,
+      "loss": 0.6077,
+      "step": 1084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9309780139302227e-05,
+      "loss": 0.6016,
+      "step": 1085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.930734610239078e-05,
+      "loss": 0.6146,
+      "step": 1086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.930490793521354e-05,
+      "loss": 0.6587,
+      "step": 1087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9302465638852484e-05,
+      "loss": 0.6268,
+      "step": 1088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.930001921439141e-05,
+      "loss": 0.6,
+      "step": 1089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.929756866291595e-05,
+      "loss": 0.6163,
+      "step": 1090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9295113985513578e-05,
+      "loss": 0.6088,
+      "step": 1091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.929265518327359e-05,
+      "loss": 0.5465,
+      "step": 1092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9290192257287107e-05,
+      "loss": 0.5854,
+      "step": 1093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.92877252086471e-05,
+      "loss": 0.602,
+      "step": 1094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.928525403844834e-05,
+      "loss": 0.661,
+      "step": 1095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.928277874778746e-05,
+      "loss": 0.6536,
+      "step": 1096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.92802993377629e-05,
+      "loss": 0.5943,
+      "step": 1097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9277815809474925e-05,
+      "loss": 0.6575,
+      "step": 1098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.927532816402564e-05,
+      "loss": 0.6135,
+      "step": 1099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.927283640251898e-05,
+      "loss": 0.6318,
+      "step": 1100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9270340526060688e-05,
+      "loss": 0.5945,
+      "step": 1101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9267840535758356e-05,
+      "loss": 0.6181,
+      "step": 1102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.926533643272138e-05,
+      "loss": 0.5884,
+      "step": 1103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9262828218060993e-05,
+      "loss": 0.5626,
+      "step": 1104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9260315892890254e-05,
+      "loss": 0.5869,
+      "step": 1105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.925779945832404e-05,
+      "loss": 0.6158,
+      "step": 1106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9255278915479056e-05,
+      "loss": 0.6004,
+      "step": 1107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9252754265473824e-05,
+      "loss": 0.5954,
+      "step": 1108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.92502255094287e-05,
+      "loss": 0.6368,
+      "step": 1109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.924769264846585e-05,
+      "loss": 0.6233,
+      "step": 1110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9245155683709268e-05,
+      "loss": 0.5842,
+      "step": 1111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.924261461628477e-05,
+      "loss": 0.5665,
+      "step": 1112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9240069447319984e-05,
+      "loss": 0.6319,
+      "step": 1113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.923752017794437e-05,
+      "loss": 0.6418,
+      "step": 1114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.92349668092892e-05,
+      "loss": 0.6124,
+      "step": 1115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.923240934248757e-05,
+      "loss": 0.5605,
+      "step": 1116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9229847778674392e-05,
+      "loss": 0.5778,
+      "step": 1117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9227282118986393e-05,
+      "loss": 0.5914,
+      "step": 1118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.922471236456212e-05,
+      "loss": 0.6382,
+      "step": 1119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9222138516541942e-05,
+      "loss": 0.6132,
+      "step": 1120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9219560576068034e-05,
+      "loss": 0.6241,
+      "step": 1121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.92169785442844e-05,
+      "loss": 0.6343,
+      "step": 1122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9214392422336846e-05,
+      "loss": 0.6387,
+      "step": 1123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9211802211373006e-05,
+      "loss": 0.6217,
+      "step": 1124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.920920791254232e-05,
+      "loss": 0.6556,
+      "step": 1125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.920660952699605e-05,
+      "loss": 0.6147,
+      "step": 1126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9204007055887258e-05,
+      "loss": 0.6422,
+      "step": 1127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.920140050037083e-05,
+      "loss": 0.6386,
+      "step": 1128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9198789861603462e-05,
+      "loss": 0.637,
+      "step": 1129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.919617514074366e-05,
+      "loss": 0.6491,
+      "step": 1130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9193556338951745e-05,
+      "loss": 0.627,
+      "step": 1131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.919093345738985e-05,
+      "loss": 0.6457,
+      "step": 1132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.918830649722191e-05,
+      "loss": 0.6808,
+      "step": 1133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.918567545961368e-05,
+      "loss": 0.6449,
+      "step": 1134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.918304034573272e-05,
+      "loss": 0.6661,
+      "step": 1135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9180401156748397e-05,
+      "loss": 0.5853,
+      "step": 1136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9177757893831884e-05,
+      "loss": 0.5732,
+      "step": 1137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9175110558156175e-05,
+      "loss": 0.5612,
+      "step": 1138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.917245915089605e-05,
+      "loss": 0.6299,
+      "step": 1139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9169803673228125e-05,
+      "loss": 0.5772,
+      "step": 1140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9167144126330792e-05,
+      "loss": 0.6429,
+      "step": 1141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9164480511384268e-05,
+      "loss": 0.6153,
+      "step": 1142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.916181282957057e-05,
+      "loss": 0.5456,
+      "step": 1143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9159141082073515e-05,
+      "loss": 0.5755,
+      "step": 1144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9156465270078735e-05,
+      "loss": 0.6001,
+      "step": 1145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9153785394773652e-05,
+      "loss": 0.5858,
+      "step": 1146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9151101457347505e-05,
+      "loss": 0.602,
+      "step": 1147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9148413458991325e-05,
+      "loss": 0.6188,
+      "step": 1148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9145721400897952e-05,
+      "loss": 0.6333,
+      "step": 1149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9143025284262022e-05,
+      "loss": 0.5941,
+      "step": 1150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9140325110279975e-05,
+      "loss": 0.6115,
+      "step": 1151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9137620880150054e-05,
+      "loss": 0.5829,
+      "step": 1152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9134912595072292e-05,
+      "loss": 0.6107,
+      "step": 1153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.913220025624854e-05,
+      "loss": 0.6595,
+      "step": 1154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9129483864882426e-05,
+      "loss": 0.6185,
+      "step": 1155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.912676342217939e-05,
+      "loss": 0.5508,
+      "step": 1156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.912403892934667e-05,
+      "loss": 0.5785,
+      "step": 1157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9121310387593298e-05,
+      "loss": 0.6353,
+      "step": 1158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9118577798130097e-05,
+      "loss": 0.7288,
+      "step": 1159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9115841162169695e-05,
+      "loss": 0.6529,
+      "step": 1160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9113100480926515e-05,
+      "loss": 0.6066,
+      "step": 1161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.911035575561677e-05,
+      "loss": 0.642,
+      "step": 1162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9107606987458473e-05,
+      "loss": 0.6029,
+      "step": 1163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9104854177671427e-05,
+      "loss": 0.6273,
+      "step": 1164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9102097327477228e-05,
+      "loss": 0.5883,
+      "step": 1165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9099336438099274e-05,
+      "loss": 0.5933,
+      "step": 1166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.909657151076274e-05,
+      "loss": 0.6334,
+      "step": 1167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9093802546694606e-05,
+      "loss": 0.693,
+      "step": 1168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.909102954712364e-05,
+      "loss": 0.632,
+      "step": 1169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.908825251328039e-05,
+      "loss": 0.6367,
+      "step": 1170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.908547144639722e-05,
+      "loss": 0.6513,
+      "step": 1171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9082686347708256e-05,
+      "loss": 0.6179,
+      "step": 1172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.907989721844943e-05,
+      "loss": 0.6404,
+      "step": 1173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9077104059858448e-05,
+      "loss": 0.581,
+      "step": 1174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9074306873174828e-05,
+      "loss": 0.5737,
+      "step": 1175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.907150565963985e-05,
+      "loss": 0.6194,
+      "step": 1176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9068700420496594e-05,
+      "loss": 0.5874,
+      "step": 1177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9065891156989932e-05,
+      "loss": 0.6392,
+      "step": 1178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9063077870366504e-05,
+      "loss": 0.5714,
+      "step": 1179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9060260561874747e-05,
+      "loss": 0.6379,
+      "step": 1180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9057439232764886e-05,
+      "loss": 0.6105,
+      "step": 1181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.905461388428892e-05,
+      "loss": 0.5727,
+      "step": 1182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9051784517700644e-05,
+      "loss": 0.6233,
+      "step": 1183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9048951134255626e-05,
+      "loss": 0.6064,
+      "step": 1184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9046113735211218e-05,
+      "loss": 0.6253,
+      "step": 1185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9043272321826558e-05,
+      "loss": 0.6586,
+      "step": 1186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.904042689536256e-05,
+      "loss": 0.6012,
+      "step": 1187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9037577457081925e-05,
+      "loss": 0.5965,
+      "step": 1188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9034724008249133e-05,
+      "loss": 0.6009,
+      "step": 1189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9031866550130437e-05,
+      "loss": 0.6089,
+      "step": 1190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9029005083993878e-05,
+      "loss": 0.5752,
+      "step": 1191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.902613961110927e-05,
+      "loss": 0.6255,
+      "step": 1192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.902327013274821e-05,
+      "loss": 0.5877,
+      "step": 1193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9020396650184067e-05,
+      "loss": 0.5873,
+      "step": 1194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9017519164691985e-05,
+      "loss": 0.6775,
+      "step": 1195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9014637677548895e-05,
+      "loss": 0.6193,
+      "step": 1196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9011752190033497e-05,
+      "loss": 0.5358,
+      "step": 1197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9008862703426266e-05,
+      "loss": 0.6363,
+      "step": 1198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9005969219009448e-05,
+      "loss": 0.5536,
+      "step": 1199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9003071738067073e-05,
+      "loss": 0.5446,
+      "step": 1200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9000170261884938e-05,
+      "loss": 0.5875,
+      "step": 1201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8997264791750614e-05,
+      "loss": 0.5979,
+      "step": 1202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.899435532895344e-05,
+      "loss": 0.6747,
+      "step": 1203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8991441874784536e-05,
+      "loss": 0.6639,
+      "step": 1204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8988524430536786e-05,
+      "loss": 0.5996,
+      "step": 1205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8985602997504842e-05,
+      "loss": 0.6379,
+      "step": 1206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8982677576985146e-05,
+      "loss": 0.6119,
+      "step": 1207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.897974817027588e-05,
+      "loss": 0.607,
+      "step": 1208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.897681477867701e-05,
+      "loss": 0.5931,
+      "step": 1209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8973877403490282e-05,
+      "loss": 0.5925,
+      "step": 1210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8970936046019185e-05,
+      "loss": 0.5719,
+      "step": 1211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8967990707568994e-05,
+      "loss": 0.599,
+      "step": 1212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8965041389446743e-05,
+      "loss": 0.6493,
+      "step": 1213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8962088092961232e-05,
+      "loss": 0.6146,
+      "step": 1214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.895913081942303e-05,
+      "loss": 0.6087,
+      "step": 1215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.895616957014447e-05,
+      "loss": 0.568,
+      "step": 1216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8953204346439644e-05,
+      "loss": 0.5855,
+      "step": 1217
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8950235149624412e-05,
+      "loss": 0.5776,
+      "step": 1218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.89472619810164e-05,
+      "loss": 0.6121,
+      "step": 1219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8944284841934992e-05,
+      "loss": 0.6239,
+      "step": 1220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8941303733701334e-05,
+      "loss": 0.552,
+      "step": 1221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8938318657638336e-05,
+      "loss": 0.6411,
+      "step": 1222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.893532961507067e-05,
+      "loss": 0.6204,
+      "step": 1223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8932336607324758e-05,
+      "loss": 0.6457,
+      "step": 1224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.89293396357288e-05,
+      "loss": 0.6,
+      "step": 1225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8926338701612738e-05,
+      "loss": 0.6228,
+      "step": 1226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8923333806308277e-05,
+      "loss": 0.6079,
+      "step": 1227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8920324951148887e-05,
+      "loss": 0.5925,
+      "step": 1228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8917312137469787e-05,
+      "loss": 0.6082,
+      "step": 1229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.891429536660795e-05,
+      "loss": 0.6197,
+      "step": 1230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.891127463990212e-05,
+      "loss": 0.5987,
+      "step": 1231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8908249958692785e-05,
+      "loss": 0.6368,
+      "step": 1232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8905221324322184e-05,
+      "loss": 0.6168,
+      "step": 1233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.890218873813432e-05,
+      "loss": 0.6257,
+      "step": 1234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8899152201474947e-05,
+      "loss": 0.6062,
+      "step": 1235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8896111715691568e-05,
+      "loss": 0.6214,
+      "step": 1236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8893067282133444e-05,
+      "loss": 0.54,
+      "step": 1237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8890018902151586e-05,
+      "loss": 0.6118,
+      "step": 1238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.888696657709875e-05,
+      "loss": 0.6066,
+      "step": 1239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8883910308329454e-05,
+      "loss": 0.626,
+      "step": 1240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.888085009719996e-05,
+      "loss": 0.6034,
+      "step": 1241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8877785945068276e-05,
+      "loss": 0.5729,
+      "step": 1242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.887471785329416e-05,
+      "loss": 0.5514,
+      "step": 1243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8871645823239128e-05,
+      "loss": 0.6399,
+      "step": 1244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8868569856266436e-05,
+      "loss": 0.5672,
+      "step": 1245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8865489953741085e-05,
+      "loss": 0.6055,
+      "step": 1246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8862406117029827e-05,
+      "loss": 0.6077,
+      "step": 1247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8859318347501152e-05,
+      "loss": 0.6296,
+      "step": 1248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8856226646525306e-05,
+      "loss": 0.6171,
+      "step": 1249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8853131015474278e-05,
+      "loss": 0.6183,
+      "step": 1250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.885003145572179e-05,
+      "loss": 0.561,
+      "step": 1251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8846927968643325e-05,
+      "loss": 0.6385,
+      "step": 1252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.884382055561609e-05,
+      "loss": 0.5862,
+      "step": 1253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.884070921801905e-05,
+      "loss": 0.6988,
+      "step": 1254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8837593957232897e-05,
+      "loss": 0.5932,
+      "step": 1255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.883447477464008e-05,
+      "loss": 0.5493,
+      "step": 1256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8831351671624777e-05,
+      "loss": 0.6242,
+      "step": 1257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8828224649572907e-05,
+      "loss": 0.5766,
+      "step": 1258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8825093709872134e-05,
+      "loss": 0.6044,
+      "step": 1259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.882195885391185e-05,
+      "loss": 0.5968,
+      "step": 1260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8818820083083197e-05,
+      "loss": 0.5253,
+      "step": 1261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.881567739877905e-05,
+      "loss": 0.5986,
+      "step": 1262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8812530802394015e-05,
+      "loss": 0.648,
+      "step": 1263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.880938029532444e-05,
+      "loss": 0.6119,
+      "step": 1264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8806225878968405e-05,
+      "loss": 0.6208,
+      "step": 1265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8803067554725732e-05,
+      "loss": 0.6236,
+      "step": 1266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8799905323997966e-05,
+      "loss": 0.5931,
+      "step": 1267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8796739188188396e-05,
+      "loss": 0.6226,
+      "step": 1268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8793569148702032e-05,
+      "loss": 0.593,
+      "step": 1269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8790395206945638e-05,
+      "loss": 0.6025,
+      "step": 1270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.878721736432768e-05,
+      "loss": 0.6678,
+      "step": 1271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8784035622258376e-05,
+      "loss": 0.5963,
+      "step": 1272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.878084998214967e-05,
+      "loss": 0.5511,
+      "step": 1273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8777660445415235e-05,
+      "loss": 0.5951,
+      "step": 1274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.877446701347047e-05,
+      "loss": 0.6245,
+      "step": 1275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.877126968773251e-05,
+      "loss": 0.5816,
+      "step": 1276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.876806846962021e-05,
+      "loss": 0.6127,
+      "step": 1277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8764863360554152e-05,
+      "loss": 0.6044,
+      "step": 1278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8761654361956654e-05,
+      "loss": 0.6086,
+      "step": 1279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8758441475251754e-05,
+      "loss": 0.6159,
+      "step": 1280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8755224701865214e-05,
+      "loss": 0.6315,
+      "step": 1281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8752004043224522e-05,
+      "loss": 0.6115,
+      "step": 1282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.874877950075889e-05,
+      "loss": 0.6039,
+      "step": 1283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8745551075899256e-05,
+      "loss": 0.6217,
+      "step": 1284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8742318770078277e-05,
+      "loss": 0.6225,
+      "step": 1285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.873908258473034e-05,
+      "loss": 0.65,
+      "step": 1286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.873584252129154e-05,
+      "loss": 0.5853,
+      "step": 1287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.87325985811997e-05,
+      "loss": 0.5946,
+      "step": 1288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.872935076589437e-05,
+      "loss": 0.5322,
+      "step": 1289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8726099076816816e-05,
+      "loss": 0.6376,
+      "step": 1290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8722843515410015e-05,
+      "loss": 0.5428,
+      "step": 1291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8719584083118668e-05,
+      "loss": 0.6733,
+      "step": 1292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8716320781389195e-05,
+      "loss": 0.6489,
+      "step": 1293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8713053611669732e-05,
+      "loss": 0.6033,
+      "step": 1294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8709782575410134e-05,
+      "loss": 0.6406,
+      "step": 1295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8706507674061972e-05,
+      "loss": 0.6217,
+      "step": 1296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8703228909078518e-05,
+      "loss": 0.6277,
+      "step": 1297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8699946281914784e-05,
+      "loss": 0.5838,
+      "step": 1298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.869665979402747e-05,
+      "loss": 0.6126,
+      "step": 1299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8693369446875008e-05,
+      "loss": 0.6239,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.869007524191753e-05,
+      "loss": 0.6212,
+      "step": 1301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8686777180616897e-05,
+      "loss": 0.574,
+      "step": 1302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8683475264436662e-05,
+      "loss": 0.629,
+      "step": 1303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8680169494842096e-05,
+      "loss": 0.6303,
+      "step": 1304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8676859873300183e-05,
+      "loss": 0.5652,
+      "step": 1305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.867354640127961e-05,
+      "loss": 0.6011,
+      "step": 1306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8670229080250787e-05,
+      "loss": 0.6235,
+      "step": 1307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.866690791168581e-05,
+      "loss": 0.5553,
+      "step": 1308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8663582897058506e-05,
+      "loss": 0.6225,
+      "step": 1309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.6014,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.865692133552069e-05,
+      "loss": 0.6357,
+      "step": 1311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8653584791566336e-05,
+      "loss": 0.6254,
+      "step": 1312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8650244407461974e-05,
+      "loss": 0.6047,
+      "step": 1313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8646900184689943e-05,
+      "loss": 0.5961,
+      "step": 1314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8643552124734293e-05,
+      "loss": 0.6052,
+      "step": 1315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8640200229080763e-05,
+      "loss": 0.5926,
+      "step": 1316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8636844499216816e-05,
+      "loss": 0.6271,
+      "step": 1317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8633484936631594e-05,
+      "loss": 0.6134,
+      "step": 1318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8630121542815955e-05,
+      "loss": 0.5753,
+      "step": 1319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.862675431926245e-05,
+      "loss": 0.6062,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.862338326746534e-05,
+      "loss": 0.5647,
+      "step": 1321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.862000838892056e-05,
+      "loss": 0.631,
+      "step": 1322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8616629685125777e-05,
+      "loss": 0.5911,
+      "step": 1323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.861324715758033e-05,
+      "loss": 0.622,
+      "step": 1324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8609860807785266e-05,
+      "loss": 0.5451,
+      "step": 1325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.860647063724332e-05,
+      "loss": 0.6126,
+      "step": 1326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8603076647458936e-05,
+      "loss": 0.5565,
+      "step": 1327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8599678839938237e-05,
+      "loss": 0.5948,
+      "step": 1328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8596277216189053e-05,
+      "loss": 0.5309,
+      "step": 1329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8592871777720902e-05,
+      "loss": 0.5603,
+      "step": 1330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8589462526044992e-05,
+      "loss": 0.5766,
+      "step": 1331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.858604946267423e-05,
+      "loss": 0.5971,
+      "step": 1332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8582632589123207e-05,
+      "loss": 0.6508,
+      "step": 1333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8579211906908216e-05,
+      "loss": 0.6038,
+      "step": 1334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8575787417547223e-05,
+      "loss": 0.6564,
+      "step": 1335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8572359122559897e-05,
+      "loss": 0.6168,
+      "step": 1336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8568927023467598e-05,
+      "loss": 0.5838,
+      "step": 1337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8565491121793366e-05,
+      "loss": 0.5979,
+      "step": 1338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.856205141906192e-05,
+      "loss": 0.6092,
+      "step": 1339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.855860791679969e-05,
+      "loss": 0.6246,
+      "step": 1340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8555160616534773e-05,
+      "loss": 0.5896,
+      "step": 1341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8551709519796956e-05,
+      "loss": 0.5985,
+      "step": 1342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8548254628117713e-05,
+      "loss": 0.5657,
+      "step": 1343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8544795943030202e-05,
+      "loss": 0.6085,
+      "step": 1344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8541333466069266e-05,
+      "loss": 0.6048,
+      "step": 1345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8537867198771416e-05,
+      "loss": 0.5555,
+      "step": 1346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.853439714267487e-05,
+      "loss": 0.5971,
+      "step": 1347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.853092329931951e-05,
+      "loss": 0.6454,
+      "step": 1348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.85274456702469e-05,
+      "loss": 0.6691,
+      "step": 1349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8523964257000288e-05,
+      "loss": 0.5901,
+      "step": 1350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.85204790611246e-05,
+      "loss": 0.5486,
+      "step": 1351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8516990084166444e-05,
+      "loss": 0.5985,
+      "step": 1352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.85134973276741e-05,
+      "loss": 0.6033,
+      "step": 1353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.851000079319753e-05,
+      "loss": 0.6071,
+      "step": 1354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8506500482288366e-05,
+      "loss": 0.6514,
+      "step": 1355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8502996396499922e-05,
+      "loss": 0.6237,
+      "step": 1356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8499488537387184e-05,
+      "loss": 0.5765,
+      "step": 1357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.849597690650682e-05,
+      "loss": 0.6114,
+      "step": 1358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.849246150541716e-05,
+      "loss": 0.5987,
+      "step": 1359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.848894233567822e-05,
+      "loss": 0.6442,
+      "step": 1360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8485419398851667e-05,
+      "loss": 0.5617,
+      "step": 1361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8481892696500866e-05,
+      "loss": 0.5629,
+      "step": 1362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8478362230190834e-05,
+      "loss": 0.597,
+      "step": 1363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.847482800148827e-05,
+      "loss": 0.6097,
+      "step": 1364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8471290011961535e-05,
+      "loss": 0.6576,
+      "step": 1365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.846774826318066e-05,
+      "loss": 0.6226,
+      "step": 1366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8464202756717352e-05,
+      "loss": 0.6339,
+      "step": 1367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8460653494144973e-05,
+      "loss": 0.6196,
+      "step": 1368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8457100477038564e-05,
+      "loss": 0.6071,
+      "step": 1369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8453543706974823e-05,
+      "loss": 0.5597,
+      "step": 1370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8449983185532113e-05,
+      "loss": 0.5774,
+      "step": 1371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8446418914290473e-05,
+      "loss": 0.5643,
+      "step": 1372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8442850894831588e-05,
+      "loss": 0.5351,
+      "step": 1373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.843927912873883e-05,
+      "loss": 0.5608,
+      "step": 1374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8435703617597216e-05,
+      "loss": 0.6381,
+      "step": 1375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8432124362993423e-05,
+      "loss": 0.5518,
+      "step": 1376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8428541366515805e-05,
+      "loss": 0.6238,
+      "step": 1377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8424954629754354e-05,
+      "loss": 0.5539,
+      "step": 1378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.842136415430075e-05,
+      "loss": 0.6401,
+      "step": 1379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.841776994174831e-05,
+      "loss": 0.6359,
+      "step": 1380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.841417199369201e-05,
+      "loss": 0.5436,
+      "step": 1381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8410570311728505e-05,
+      "loss": 0.599,
+      "step": 1382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8406964897456077e-05,
+      "loss": 0.6039,
+      "step": 1383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.840335575247469e-05,
+      "loss": 0.6272,
+      "step": 1384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8399742878385947e-05,
+      "loss": 0.5634,
+      "step": 1385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8396126276793118e-05,
+      "loss": 0.5791,
+      "step": 1386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.839250594930111e-05,
+      "loss": 0.5514,
+      "step": 1387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.83888818975165e-05,
+      "loss": 0.6401,
+      "step": 1388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8385254123047517e-05,
+      "loss": 0.6347,
+      "step": 1389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.838162262750403e-05,
+      "loss": 0.5845,
+      "step": 1390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.837798741249757e-05,
+      "loss": 0.6436,
+      "step": 1391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8374348479641316e-05,
+      "loss": 0.5982,
+      "step": 1392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.837070583055009e-05,
+      "loss": 0.6438,
+      "step": 1393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8367059466840373e-05,
+      "loss": 0.5828,
+      "step": 1394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.836340939013029e-05,
+      "loss": 0.5718,
+      "step": 1395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8359755602039615e-05,
+      "loss": 0.5847,
+      "step": 1396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8356098104189762e-05,
+      "loss": 0.5908,
+      "step": 1397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8352436898203803e-05,
+      "loss": 0.5935,
+      "step": 1398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8348771985706447e-05,
+      "loss": 0.6105,
+      "step": 1399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.834510336832405e-05,
+      "loss": 0.6116,
+      "step": 1400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8341431047684616e-05,
+      "loss": 0.5634,
+      "step": 1401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8337755025417788e-05,
+      "loss": 0.5719,
+      "step": 1402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.833407530315485e-05,
+      "loss": 0.6319,
+      "step": 1403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8330391882528732e-05,
+      "loss": 0.6411,
+      "step": 1404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8326704765174006e-05,
+      "loss": 0.5638,
+      "step": 1405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8323013952726877e-05,
+      "loss": 0.6572,
+      "step": 1406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.83193194468252e-05,
+      "loss": 0.6063,
+      "step": 1407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8315621249108458e-05,
+      "loss": 0.5581,
+      "step": 1408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8311919361217785e-05,
+      "loss": 0.5945,
+      "step": 1409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8308213784795942e-05,
+      "loss": 0.5768,
+      "step": 1410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8304504521487335e-05,
+      "loss": 0.597,
+      "step": 1411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8300791572937996e-05,
+      "loss": 0.6359,
+      "step": 1412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.82970749407956e-05,
+      "loss": 0.6554,
+      "step": 1413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8293354626709457e-05,
+      "loss": 0.5596,
+      "step": 1414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8289630632330502e-05,
+      "loss": 0.597,
+      "step": 1415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.828590295931132e-05,
+      "loss": 0.5594,
+      "step": 1416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8282171609306113e-05,
+      "loss": 0.5902,
+      "step": 1417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.827843658397072e-05,
+      "loss": 0.5713,
+      "step": 1418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.827469788496261e-05,
+      "loss": 0.5012,
+      "step": 1419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.827095551394089e-05,
+      "loss": 0.6353,
+      "step": 1420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.826720947256628e-05,
+      "loss": 0.5461,
+      "step": 1421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8263459762501143e-05,
+      "loss": 0.5566,
+      "step": 1422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8259706385409468e-05,
+      "loss": 0.598,
+      "step": 1423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8255949342956864e-05,
+      "loss": 0.6174,
+      "step": 1424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8252188636810576e-05,
+      "loss": 0.6299,
+      "step": 1425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.824842426863947e-05,
+      "loss": 0.5957,
+      "step": 1426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.824465624011403e-05,
+      "loss": 0.6188,
+      "step": 1427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8240884552906384e-05,
+      "loss": 0.6131,
+      "step": 1428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.823710920869026e-05,
+      "loss": 0.61,
+      "step": 1429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8233330209141026e-05,
+      "loss": 0.6306,
+      "step": 1430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8229547555935666e-05,
+      "loss": 0.595,
+      "step": 1431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8225761250752785e-05,
+      "loss": 0.6014,
+      "step": 1432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8221971295272606e-05,
+      "loss": 0.5796,
+      "step": 1433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8218177691176982e-05,
+      "loss": 0.6403,
+      "step": 1434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8214380440149373e-05,
+      "loss": 0.6006,
+      "step": 1435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.821057954387486e-05,
+      "loss": 0.4661,
+      "step": 1436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8206775004040152e-05,
+      "loss": 0.6209,
+      "step": 1437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.820296682233356e-05,
+      "loss": 0.5852,
+      "step": 1438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8199155000445025e-05,
+      "loss": 0.5899,
+      "step": 1439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8195339540066088e-05,
+      "loss": 0.571,
+      "step": 1440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.819152044288992e-05,
+      "loss": 0.6466,
+      "step": 1441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.81876977106113e-05,
+      "loss": 0.5832,
+      "step": 1442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.818387134492661e-05,
+      "loss": 0.6083,
+      "step": 1443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8180041347533862e-05,
+      "loss": 0.5716,
+      "step": 1444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.817620772013267e-05,
+      "loss": 0.5951,
+      "step": 1445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.817237046442426e-05,
+      "loss": 0.668,
+      "step": 1446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.816852958211147e-05,
+      "loss": 0.6358,
+      "step": 1447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8164685074898737e-05,
+      "loss": 0.606,
+      "step": 1448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8160836944492127e-05,
+      "loss": 0.5588,
+      "step": 1449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.815698519259929e-05,
+      "loss": 0.6186,
+      "step": 1450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8153129820929506e-05,
+      "loss": 0.5597,
+      "step": 1451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8149270831193645e-05,
+      "loss": 0.6644,
+      "step": 1452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8145408225104192e-05,
+      "loss": 0.6546,
+      "step": 1453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8141542004375224e-05,
+      "loss": 0.5464,
+      "step": 1454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.813767217072244e-05,
+      "loss": 0.5815,
+      "step": 1455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8133798725863126e-05,
+      "loss": 0.5951,
+      "step": 1456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8129921671516182e-05,
+      "loss": 0.6415,
+      "step": 1457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8126041009402104e-05,
+      "loss": 0.6051,
+      "step": 1458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8122156741242995e-05,
+      "loss": 0.5687,
+      "step": 1459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.811826886876255e-05,
+      "loss": 0.6355,
+      "step": 1460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.811437739368606e-05,
+      "loss": 0.5769,
+      "step": 1461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8110482317740432e-05,
+      "loss": 0.5443,
+      "step": 1462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8106583642654157e-05,
+      "loss": 0.5803,
+      "step": 1463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8102681370157328e-05,
+      "loss": 0.5905,
+      "step": 1464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8098775501981633e-05,
+      "loss": 0.5838,
+      "step": 1465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8094866039860356e-05,
+      "loss": 0.6328,
+      "step": 1466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8090952985528372e-05,
+      "loss": 0.5359,
+      "step": 1467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.808703634072216e-05,
+      "loss": 0.6343,
+      "step": 1468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.808311610717979e-05,
+      "loss": 0.5373,
+      "step": 1469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8079192286640906e-05,
+      "loss": 0.6396,
+      "step": 1470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8075264880846764e-05,
+      "loss": 0.607,
+      "step": 1471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8071333891540213e-05,
+      "loss": 0.5929,
+      "step": 1472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8067399320465683e-05,
+      "loss": 0.5336,
+      "step": 1473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.806346116936919e-05,
+      "loss": 0.6266,
+      "step": 1474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8059519439998346e-05,
+      "loss": 0.5798,
+      "step": 1475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.805557413410235e-05,
+      "loss": 0.5352,
+      "step": 1476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8051625253431987e-05,
+      "loss": 0.5889,
+      "step": 1477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.804767279973963e-05,
+      "loss": 0.6182,
+      "step": 1478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.804371677477923e-05,
+      "loss": 0.5648,
+      "step": 1479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8039757180306333e-05,
+      "loss": 0.5978,
+      "step": 1480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8035794018078063e-05,
+      "loss": 0.642,
+      "step": 1481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8031827289853133e-05,
+      "loss": 0.5557,
+      "step": 1482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.802785699739183e-05,
+      "loss": 0.5985,
+      "step": 1483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.802388314245603e-05,
+      "loss": 0.6296,
+      "step": 1484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8019905726809184e-05,
+      "loss": 0.6379,
+      "step": 1485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.801592475221633e-05,
+      "loss": 0.5483,
+      "step": 1486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.801194022044408e-05,
+      "loss": 0.6075,
+      "step": 1487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8007952133260624e-05,
+      "loss": 0.6093,
+      "step": 1488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.800396049243573e-05,
+      "loss": 0.6118,
+      "step": 1489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7999965299740753e-05,
+      "loss": 0.6073,
+      "step": 1490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.799596655694861e-05,
+      "loss": 0.6061,
+      "step": 1491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.79919642658338e-05,
+      "loss": 0.555,
+      "step": 1492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.79879584281724e-05,
+      "loss": 0.5703,
+      "step": 1493
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7983949045742046e-05,
+      "loss": 0.5625,
+      "step": 1494
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.797993612032197e-05,
+      "loss": 0.6492,
+      "step": 1495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7975919653692958e-05,
+      "loss": 0.5859,
+      "step": 1496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7971899647637377e-05,
+      "loss": 0.6612,
+      "step": 1497
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7967876103939158e-05,
+      "loss": 0.5868,
+      "step": 1498
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7963849024383805e-05,
+      "loss": 0.5734,
+      "step": 1499
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7959818410758395e-05,
+      "loss": 0.5966,
+      "step": 1500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7955784264851564e-05,
+      "loss": 0.5936,
+      "step": 1501
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7951746588453523e-05,
+      "loss": 0.5638,
+      "step": 1502
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7947705383356054e-05,
+      "loss": 0.5551,
+      "step": 1503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.794366065135249e-05,
+      "loss": 0.5826,
+      "step": 1504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.793961239423774e-05,
+      "loss": 0.596,
+      "step": 1505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.793556061380828e-05,
+      "loss": 0.5694,
+      "step": 1506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7931505311862135e-05,
+      "loss": 0.6174,
+      "step": 1507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.792744649019891e-05,
+      "loss": 0.6722,
+      "step": 1508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7923384150619766e-05,
+      "loss": 0.6032,
+      "step": 1509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7919318294927418e-05,
+      "loss": 0.6387,
+      "step": 1510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7915248924926147e-05,
+      "loss": 0.6242,
+      "step": 1511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7911176042421792e-05,
+      "loss": 0.5859,
+      "step": 1512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.790709964922176e-05,
+      "loss": 0.6115,
+      "step": 1513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7903019747134996e-05,
+      "loss": 0.5911,
+      "step": 1514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7898936337972022e-05,
+      "loss": 0.5932,
+      "step": 1515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7894849423544908e-05,
+      "loss": 0.576,
+      "step": 1516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.789075900566728e-05,
+      "loss": 0.5779,
+      "step": 1517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7886665086154313e-05,
+      "loss": 0.618,
+      "step": 1518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7882567666822748e-05,
+      "loss": 0.5853,
+      "step": 1519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7878466749490864e-05,
+      "loss": 0.5886,
+      "step": 1520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.787436233597851e-05,
+      "loss": 0.5993,
+      "step": 1521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7870254428107075e-05,
+      "loss": 0.6199,
+      "step": 1522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.78661430276995e-05,
+      "loss": 0.5768,
+      "step": 1523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7862028136580278e-05,
+      "loss": 0.6168,
+      "step": 1524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.785790975657544e-05,
+      "loss": 0.594,
+      "step": 1525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7853787889512588e-05,
+      "loss": 0.6042,
+      "step": 1526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7849662537220854e-05,
+      "loss": 0.5788,
+      "step": 1527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7845533701530923e-05,
+      "loss": 0.5991,
+      "step": 1528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.784140138427502e-05,
+      "loss": 0.5601,
+      "step": 1529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.783726558728692e-05,
+      "loss": 0.5995,
+      "step": 1530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.783312631240194e-05,
+      "loss": 0.5555,
+      "step": 1531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7828983561456943e-05,
+      "loss": 0.6392,
+      "step": 1532
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7824837336290336e-05,
+      "loss": 0.584,
+      "step": 1533
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.782068763874206e-05,
+      "loss": 0.5899,
+      "step": 1534
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7816534470653607e-05,
+      "loss": 0.5862,
+      "step": 1535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7812377833867998e-05,
+      "loss": 0.5826,
+      "step": 1536
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.78082177302298e-05,
+      "loss": 0.5317,
+      "step": 1537
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7804054161585123e-05,
+      "loss": 0.65,
+      "step": 1538
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7799887129781602e-05,
+      "loss": 0.57,
+      "step": 1539
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.779571663666842e-05,
+      "loss": 0.5875,
+      "step": 1540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.779154268409629e-05,
+      "loss": 0.6206,
+      "step": 1541
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7787365273917468e-05,
+      "loss": 0.6787,
+      "step": 1542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.778318440798573e-05,
+      "loss": 0.5746,
+      "step": 1543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7779000088156398e-05,
+      "loss": 0.5424,
+      "step": 1544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.777481231628632e-05,
+      "loss": 0.5691,
+      "step": 1545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.777062109423389e-05,
+      "loss": 0.5616,
+      "step": 1546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7766426423859008e-05,
+      "loss": 0.5828,
+      "step": 1547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7762228307023122e-05,
+      "loss": 0.5725,
+      "step": 1548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.775802674558921e-05,
+      "loss": 0.5955,
+      "step": 1549
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.775382174142177e-05,
+      "loss": 0.5671,
+      "step": 1550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.774961329638683e-05,
+      "loss": 0.6705,
+      "step": 1551
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.774540141235195e-05,
+      "loss": 0.6275,
+      "step": 1552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.774118609118621e-05,
+      "loss": 0.5877,
+      "step": 1553
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7736967334760224e-05,
+      "loss": 0.6123,
+      "step": 1554
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.773274514494612e-05,
+      "loss": 0.5783,
+      "step": 1555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7728519523617557e-05,
+      "loss": 0.5634,
+      "step": 1556
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7724290472649706e-05,
+      "loss": 0.5852,
+      "step": 1557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7720057993919274e-05,
+      "loss": 0.5519,
+      "step": 1558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7715822089304487e-05,
+      "loss": 0.563,
+      "step": 1559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.771158276068508e-05,
+      "loss": 0.6059,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.770734000994232e-05,
+      "loss": 0.6263,
+      "step": 1561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.770309383895898e-05,
+      "loss": 0.6034,
+      "step": 1562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7698844249619367e-05,
+      "loss": 0.6078,
+      "step": 1563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7694591243809293e-05,
+      "loss": 0.4909,
+      "step": 1564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7690334823416085e-05,
+      "loss": 0.5866,
+      "step": 1565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.768607499032859e-05,
+      "loss": 0.6287,
+      "step": 1566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7681811746437177e-05,
+      "loss": 0.5678,
+      "step": 1567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7677545093633713e-05,
+      "loss": 0.6188,
+      "step": 1568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7673275033811584e-05,
+      "loss": 0.543,
+      "step": 1569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7669001568865697e-05,
+      "loss": 0.5738,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7664724700692454e-05,
+      "loss": 0.5444,
+      "step": 1571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.6227,
+      "step": 1572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7656160762257105e-05,
+      "loss": 0.5623,
+      "step": 1573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7651873695795366e-05,
+      "loss": 0.6082,
+      "step": 1574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7647583233707012e-05,
+      "loss": 0.6082,
+      "step": 1575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7643289377895988e-05,
+      "loss": 0.6115,
+      "step": 1576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7638992130267758e-05,
+      "loss": 0.6229,
+      "step": 1577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.763469149272929e-05,
+      "loss": 0.5865,
+      "step": 1578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7630387467189046e-05,
+      "loss": 0.5824,
+      "step": 1579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7626080055557e-05,
+      "loss": 0.6001,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.762176925974463e-05,
+      "loss": 0.5774,
+      "step": 1581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7617455081664905e-05,
+      "loss": 0.6269,
+      "step": 1582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.761313752323231e-05,
+      "loss": 0.6158,
+      "step": 1583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7608816586362817e-05,
+      "loss": 0.5992,
+      "step": 1584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7604492272973903e-05,
+      "loss": 0.5916,
+      "step": 1585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7600164584984548e-05,
+      "loss": 0.6184,
+      "step": 1586
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7595833524315216e-05,
+      "loss": 0.5112,
+      "step": 1587
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7591499092887883e-05,
+      "loss": 0.6309,
+      "step": 1588
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.758716129262601e-05,
+      "loss": 0.6271,
+      "step": 1589
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7582820125454563e-05,
+      "loss": 0.5587,
+      "step": 1590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.757847559329999e-05,
+      "loss": 0.614,
+      "step": 1591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.757412769809024e-05,
+      "loss": 0.5827,
+      "step": 1592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.756977644175475e-05,
+      "loss": 0.6139,
+      "step": 1593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7565421826224455e-05,
+      "loss": 0.6221,
+      "step": 1594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7561063853431775e-05,
+      "loss": 0.5879,
+      "step": 1595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7556702525310623e-05,
+      "loss": 0.6278,
+      "step": 1596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.75523378437964e-05,
+      "loss": 0.5963,
+      "step": 1597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7547969810825996e-05,
+      "loss": 0.6078,
+      "step": 1598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.754359842833778e-05,
+      "loss": 0.5479,
+      "step": 1599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.753922369827162e-05,
+      "loss": 0.5693,
+      "step": 1600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7534845622568867e-05,
+      "loss": 0.5877,
+      "step": 1601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.753046420317235e-05,
+      "loss": 0.6012,
+      "step": 1602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7526079442026387e-05,
+      "loss": 0.5869,
+      "step": 1603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7521691341076774e-05,
+      "loss": 0.5717,
+      "step": 1604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.751729990227079e-05,
+      "loss": 0.5987,
+      "step": 1605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7512905127557212e-05,
+      "loss": 0.5829,
+      "step": 1606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7508507018886266e-05,
+      "loss": 0.6114,
+      "step": 1607
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.750410557820969e-05,
+      "loss": 0.5694,
+      "step": 1608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7499700807480674e-05,
+      "loss": 0.5095,
+      "step": 1609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7495292708653896e-05,
+      "loss": 0.602,
+      "step": 1610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.749088128368552e-05,
+      "loss": 0.5807,
+      "step": 1611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7486466534533175e-05,
+      "loss": 0.5813,
+      "step": 1612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7482048463155967e-05,
+      "loss": 0.6075,
+      "step": 1613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7477627071514477e-05,
+      "loss": 0.5465,
+      "step": 1614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.747320236157076e-05,
+      "loss": 0.5998,
+      "step": 1615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7468774335288343e-05,
+      "loss": 0.6198,
+      "step": 1616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7464342994632226e-05,
+      "loss": 0.5709,
+      "step": 1617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.745990834156888e-05,
+      "loss": 0.5915,
+      "step": 1618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7455470378066245e-05,
+      "loss": 0.6031,
+      "step": 1619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7451029106093725e-05,
+      "loss": 0.5869,
+      "step": 1620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.74465845276222e-05,
+      "loss": 0.5935,
+      "step": 1621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7442136644624015e-05,
+      "loss": 0.5908,
+      "step": 1622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.743768545907298e-05,
+      "loss": 0.6246,
+      "step": 1623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7433230972944368e-05,
+      "loss": 0.5707,
+      "step": 1624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7428773188214923e-05,
+      "loss": 0.5468,
+      "step": 1625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7424312106862854e-05,
+      "loss": 0.6268,
+      "step": 1626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.741984773086782e-05,
+      "loss": 0.6151,
+      "step": 1627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7415380062210956e-05,
+      "loss": 0.5566,
+      "step": 1628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7410909102874853e-05,
+      "loss": 0.6195,
+      "step": 1629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.740643485484356e-05,
+      "loss": 0.6281,
+      "step": 1630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7401957320102586e-05,
+      "loss": 0.587,
+      "step": 1631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7397476500638903e-05,
+      "loss": 0.5599,
+      "step": 1632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.739299239844094e-05,
+      "loss": 0.6414,
+      "step": 1633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7388505015498577e-05,
+      "loss": 0.573,
+      "step": 1634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.738401435380315e-05,
+      "loss": 0.6204,
+      "step": 1635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.737952041534746e-05,
+      "loss": 0.5958,
+      "step": 1636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7375023202125753e-05,
+      "loss": 0.5841,
+      "step": 1637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.737052271613373e-05,
+      "loss": 0.5524,
+      "step": 1638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7366018959368546e-05,
+      "loss": 0.6327,
+      "step": 1639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.73615119338288e-05,
+      "loss": 0.6221,
+      "step": 1640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7357001641514558e-05,
+      "loss": 0.6039,
+      "step": 1641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.735248808442732e-05,
+      "loss": 0.5409,
+      "step": 1642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.734797126457004e-05,
+      "loss": 0.5976,
+      "step": 1643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7343451183947127e-05,
+      "loss": 0.6028,
+      "step": 1644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.733892784456442e-05,
+      "loss": 0.6355,
+      "step": 1645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7334401248429228e-05,
+      "loss": 0.5722,
+      "step": 1646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7329871397550278e-05,
+      "loss": 0.5178,
+      "step": 1647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.732533829393776e-05,
+      "loss": 0.5519,
+      "step": 1648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7320801939603306e-05,
+      "loss": 0.6315,
+      "step": 1649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7316262336559978e-05,
+      "loss": 0.6348,
+      "step": 1650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.73117194868223e-05,
+      "loss": 0.6638,
+      "step": 1651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7307173392406223e-05,
+      "loss": 0.6635,
+      "step": 1652
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7302624055329132e-05,
+      "loss": 0.4962,
+      "step": 1653
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7298071477609868e-05,
+      "loss": 0.6376,
+      "step": 1654
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7293515661268702e-05,
+      "loss": 0.6255,
+      "step": 1655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7288956608327333e-05,
+      "loss": 0.5569,
+      "step": 1656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.728439432080891e-05,
+      "loss": 0.6022,
+      "step": 1657
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.727982880073802e-05,
+      "loss": 0.5864,
+      "step": 1658
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7275260050140662e-05,
+      "loss": 0.5809,
+      "step": 1659
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7270688071044293e-05,
+      "loss": 0.5958,
+      "step": 1660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7266112865477788e-05,
+      "loss": 0.5731,
+      "step": 1661
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7261534435471463e-05,
+      "loss": 0.5733,
+      "step": 1662
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.725695278305706e-05,
+      "loss": 0.5593,
+      "step": 1663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7252367910267752e-05,
+      "loss": 0.5289,
+      "step": 1664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7247779819138137e-05,
+      "loss": 0.5288,
+      "step": 1665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7243188511704247e-05,
+      "loss": 0.6284,
+      "step": 1666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7238593990003546e-05,
+      "loss": 0.5364,
+      "step": 1667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7233996256074906e-05,
+      "loss": 0.5137,
+      "step": 1668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7229395311958644e-05,
+      "loss": 0.5778,
+      "step": 1669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.722479115969649e-05,
+      "loss": 0.6284,
+      "step": 1670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7220183801331605e-05,
+      "loss": 0.5209,
+      "step": 1671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7215573238908564e-05,
+      "loss": 0.591,
+      "step": 1672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7210959474473372e-05,
+      "loss": 0.6175,
+      "step": 1673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7206342510073456e-05,
+      "loss": 0.5798,
+      "step": 1674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.720172234775765e-05,
+      "loss": 0.5374,
+      "step": 1675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7197098989576222e-05,
+      "loss": 0.5207,
+      "step": 1676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.719247243758085e-05,
+      "loss": 0.5948,
+      "step": 1677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7187842693824633e-05,
+      "loss": 0.607,
+      "step": 1678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7183209760362086e-05,
+      "loss": 0.6085,
+      "step": 1679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.717857363924914e-05,
+      "loss": 0.6097,
+      "step": 1680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.717393433254313e-05,
+      "loss": 0.584,
+      "step": 1681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.716929184230282e-05,
+      "loss": 0.608,
+      "step": 1682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.716464617058838e-05,
+      "loss": 0.5732,
+      "step": 1683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7159997319461393e-05,
+      "loss": 0.5954,
+      "step": 1684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.715534529098485e-05,
+      "loss": 0.6143,
+      "step": 1685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7150690087223158e-05,
+      "loss": 0.6432,
+      "step": 1686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7146031710242127e-05,
+      "loss": 0.6109,
+      "step": 1687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7141370162108973e-05,
+      "loss": 0.642,
+      "step": 1688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.713670544489233e-05,
+      "loss": 0.5928,
+      "step": 1689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7132037560662227e-05,
+      "loss": 0.5389,
+      "step": 1690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7127366511490103e-05,
+      "loss": 0.5964,
+      "step": 1691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.712269229944881e-05,
+      "loss": 0.5838,
+      "step": 1692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7118014926612586e-05,
+      "loss": 0.6001,
+      "step": 1693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7113334395057085e-05,
+      "loss": 0.5853,
+      "step": 1694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.710865070685936e-05,
+      "loss": 0.6071,
+      "step": 1695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7103963864097862e-05,
+      "loss": 0.6064,
+      "step": 1696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.709927386885244e-05,
+      "loss": 0.5941,
+      "step": 1697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7094580723204353e-05,
+      "loss": 0.5739,
+      "step": 1698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7089884429236243e-05,
+      "loss": 0.6133,
+      "step": 1699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.708518498903216e-05,
+      "loss": 0.5567,
+      "step": 1700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.708048240467755e-05,
+      "loss": 0.6206,
+      "step": 1701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.707577667825924e-05,
+      "loss": 0.6156,
+      "step": 1702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.5667,
+      "step": 1703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7066355807585874e-05,
+      "loss": 0.534,
+      "step": 1704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.706164066751146e-05,
+      "loss": 0.5522,
+      "step": 1705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7056922393734635e-05,
+      "loss": 0.5925,
+      "step": 1706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7052200988349206e-05,
+      "loss": 0.6062,
+      "step": 1707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.704747645345036e-05,
+      "loss": 0.5407,
+      "step": 1708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7042748791134675e-05,
+      "loss": 0.5908,
+      "step": 1709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.703801800350012e-05,
+      "loss": 0.5495,
+      "step": 1710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.703328409264605e-05,
+      "loss": 0.544,
+      "step": 1711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.70285470606732e-05,
+      "loss": 0.5689,
+      "step": 1712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7023806909683696e-05,
+      "loss": 0.5127,
+      "step": 1713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.701906364178105e-05,
+      "loss": 0.5686,
+      "step": 1714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.701431725907015e-05,
+      "loss": 0.5941,
+      "step": 1715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7009567763657275e-05,
+      "loss": 0.5922,
+      "step": 1716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7004815157650075e-05,
+      "loss": 0.5687,
+      "step": 1717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7000059443157587e-05,
+      "loss": 0.5928,
+      "step": 1718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6995300622290227e-05,
+      "loss": 0.5678,
+      "step": 1719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.699053869715979e-05,
+      "loss": 0.6546,
+      "step": 1720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6985773669879444e-05,
+      "loss": 0.6119,
+      "step": 1721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6981005542563738e-05,
+      "loss": 0.5613,
+      "step": 1722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6976234317328603e-05,
+      "loss": 0.6107,
+      "step": 1723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.697145999629133e-05,
+      "loss": 0.6297,
+      "step": 1724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6966682581570588e-05,
+      "loss": 0.627,
+      "step": 1725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6961902075286426e-05,
+      "loss": 0.5789,
+      "step": 1726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6957118479560263e-05,
+      "loss": 0.566,
+      "step": 1727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.695233179651488e-05,
+      "loss": 0.5833,
+      "step": 1728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6947542028274446e-05,
+      "loss": 0.5746,
+      "step": 1729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.694274917696448e-05,
+      "loss": 0.5732,
+      "step": 1730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.693795324471188e-05,
+      "loss": 0.6101,
+      "step": 1731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6933154233644904e-05,
+      "loss": 0.5659,
+      "step": 1732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.692835214589319e-05,
+      "loss": 0.5424,
+      "step": 1733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.692354698358772e-05,
+      "loss": 0.6282,
+      "step": 1734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.691873874886086e-05,
+      "loss": 0.5614,
+      "step": 1735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6913927443846333e-05,
+      "loss": 0.5616,
+      "step": 1736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.690911307067922e-05,
+      "loss": 0.6262,
+      "step": 1737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6904295631495967e-05,
+      "loss": 0.5709,
+      "step": 1738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.689947512843438e-05,
+      "loss": 0.5886,
+      "step": 1739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6894651563633628e-05,
+      "loss": 0.6057,
+      "step": 1740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6889824939234233e-05,
+      "loss": 0.6086,
+      "step": 1741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6884995257378078e-05,
+      "loss": 0.538,
+      "step": 1742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.68801625202084e-05,
+      "loss": 0.5133,
+      "step": 1743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.68753267298698e-05,
+      "loss": 0.5282,
+      "step": 1744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6870487888508225e-05,
+      "loss": 0.6264,
+      "step": 1745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6865645998270978e-05,
+      "loss": 0.5588,
+      "step": 1746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6860801061306716e-05,
+      "loss": 0.6098,
+      "step": 1747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.685595307976545e-05,
+      "loss": 0.6167,
+      "step": 1748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6851102055798533e-05,
+      "loss": 0.5685,
+      "step": 1749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6846247991558686e-05,
+      "loss": 0.6331,
+      "step": 1750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6841390889199962e-05,
+      "loss": 0.6047,
+      "step": 1751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.683653075087777e-05,
+      "loss": 0.5689,
+      "step": 1752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.683166757874886e-05,
+      "loss": 0.5032,
+      "step": 1753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.682680137497134e-05,
+      "loss": 0.5568,
+      "step": 1754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6821932141704658e-05,
+      "loss": 0.6221,
+      "step": 1755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6817059881109594e-05,
+      "loss": 0.6026,
+      "step": 1756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6812184595348292e-05,
+      "loss": 0.5854,
+      "step": 1757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6807306286584223e-05,
+      "loss": 0.5673,
+      "step": 1758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.680242495698221e-05,
+      "loss": 0.617,
+      "step": 1759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6797540608708405e-05,
+      "loss": 0.6151,
+      "step": 1760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6792653243930318e-05,
+      "loss": 0.6113,
+      "step": 1761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.678776286481677e-05,
+      "loss": 0.5934,
+      "step": 1762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6782869473537947e-05,
+      "loss": 0.6082,
+      "step": 1763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.677797307226536e-05,
+      "loss": 0.6182,
+      "step": 1764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.677307366317185e-05,
+      "loss": 0.6024,
+      "step": 1765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6768171248431603e-05,
+      "loss": 0.5407,
+      "step": 1766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6763265830220133e-05,
+      "loss": 0.6126,
+      "step": 1767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6758357410714287e-05,
+      "loss": 0.6165,
+      "step": 1768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.675344599209225e-05,
+      "loss": 0.5633,
+      "step": 1769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6748531576533527e-05,
+      "loss": 0.5962,
+      "step": 1770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6743614166218962e-05,
+      "loss": 0.5526,
+      "step": 1771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6738693763330727e-05,
+      "loss": 0.6107,
+      "step": 1772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6733770370052317e-05,
+      "loss": 0.5675,
+      "step": 1773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.672884398856856e-05,
+      "loss": 0.6405,
+      "step": 1774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.67239146210656e-05,
+      "loss": 0.4901,
+      "step": 1775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6718982269730925e-05,
+      "loss": 0.5484,
+      "step": 1776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6714046936753322e-05,
+      "loss": 0.6087,
+      "step": 1777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6709108624322924e-05,
+      "loss": 0.5923,
+      "step": 1778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.670416733463118e-05,
+      "loss": 0.524,
+      "step": 1779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6699223069870844e-05,
+      "loss": 0.6175,
+      "step": 1780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.669427583223601e-05,
+      "loss": 0.5842,
+      "step": 1781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6689325623922085e-05,
+      "loss": 0.61,
+      "step": 1782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.668437244712579e-05,
+      "loss": 0.5891,
+      "step": 1783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6679416304045172e-05,
+      "loss": 0.577,
+      "step": 1784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6674457196879585e-05,
+      "loss": 0.5935,
+      "step": 1785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6669495127829706e-05,
+      "loss": 0.6043,
+      "step": 1786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6664530099097523e-05,
+      "loss": 0.5766,
+      "step": 1787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6659562112886332e-05,
+      "loss": 0.5967,
+      "step": 1788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6654591171400752e-05,
+      "loss": 0.6632,
+      "step": 1789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6649617276846705e-05,
+      "loss": 0.5777,
+      "step": 1790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.664464043143143e-05,
+      "loss": 0.5557,
+      "step": 1791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.663966063736347e-05,
+      "loss": 0.5876,
+      "step": 1792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6634677896852676e-05,
+      "loss": 0.5592,
+      "step": 1793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.662969221211022e-05,
+      "loss": 0.5497,
+      "step": 1794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6624703585348557e-05,
+      "loss": 0.5676,
+      "step": 1795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6619712018781464e-05,
+      "loss": 0.5745,
+      "step": 1796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6614717514624024e-05,
+      "loss": 0.5874,
+      "step": 1797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.660972007509261e-05,
+      "loss": 0.6005,
+      "step": 1798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6604719702404913e-05,
+      "loss": 0.6212,
+      "step": 1799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.659971639877992e-05,
+      "loss": 0.717,
+      "step": 1800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.659471016643791e-05,
+      "loss": 0.5757,
+      "step": 1801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6589701007600475e-05,
+      "loss": 0.6004,
+      "step": 1802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.65846889244905e-05,
+      "loss": 0.5769,
+      "step": 1803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6579673919332162e-05,
+      "loss": 0.5427,
+      "step": 1804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6574655994350943e-05,
+      "loss": 0.5438,
+      "step": 1805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.656963515177362e-05,
+      "loss": 0.5649,
+      "step": 1806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.656461139382826e-05,
+      "loss": 0.6039,
+      "step": 1807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6559584722744228e-05,
+      "loss": 0.5692,
+      "step": 1808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6554555140752176e-05,
+      "loss": 0.562,
+      "step": 1809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6549522650084056e-05,
+      "loss": 0.6429,
+      "step": 1810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6544487252973107e-05,
+      "loss": 0.5848,
+      "step": 1811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6539448951653853e-05,
+      "loss": 0.5815,
+      "step": 1812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6534407748362114e-05,
+      "loss": 0.6285,
+      "step": 1813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6529363645335e-05,
+      "loss": 0.5355,
+      "step": 1814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.652431664481089e-05,
+      "loss": 0.6421,
+      "step": 1815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6519266749029473e-05,
+      "loss": 0.5718,
+      "step": 1816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6514213960231705e-05,
+      "loss": 0.5727,
+      "step": 1817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6509158280659833e-05,
+      "loss": 0.5955,
+      "step": 1818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.650409971255739e-05,
+      "loss": 0.5685,
+      "step": 1819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.649903825816918e-05,
+      "loss": 0.5436,
+      "step": 1820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6493973919741308e-05,
+      "loss": 0.6526,
+      "step": 1821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6488906699521126e-05,
+      "loss": 0.5874,
+      "step": 1822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6483836599757296e-05,
+      "loss": 0.6072,
+      "step": 1823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.647876362269975e-05,
+      "loss": 0.6134,
+      "step": 1824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.647368777059968e-05,
+      "loss": 0.5814,
+      "step": 1825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.646860904570958e-05,
+      "loss": 0.6208,
+      "step": 1826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6463527450283196e-05,
+      "loss": 0.6377,
+      "step": 1827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6458442986575564e-05,
+      "loss": 0.6299,
+      "step": 1828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.645335565684298e-05,
+      "loss": 0.6067,
+      "step": 1829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6448265463343025e-05,
+      "loss": 0.6405,
+      "step": 1830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.644317240833454e-05,
+      "loss": 0.5552,
+      "step": 1831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.643807649407764e-05,
+      "loss": 0.5824,
+      "step": 1832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6432977722833706e-05,
+      "loss": 0.576,
+      "step": 1833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.565,
+      "step": 1834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6422771618436622e-05,
+      "loss": 0.6373,
+      "step": 1835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6417664289812572e-05,
+      "loss": 0.5608,
+      "step": 1836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6412554113259696e-05,
+      "loss": 0.5441,
+      "step": 1837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6407441091045705e-05,
+      "loss": 0.5097,
+      "step": 1838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6402325225439573e-05,
+      "loss": 0.5903,
+      "step": 1839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6397206518711538e-05,
+      "loss": 0.605,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6392084973133098e-05,
+      "loss": 0.519,
+      "step": 1841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6386960590977013e-05,
+      "loss": 0.6055,
+      "step": 1842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.63818333745173e-05,
+      "loss": 0.5499,
+      "step": 1843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.637670332602923e-05,
+      "loss": 0.5634,
+      "step": 1844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6371570447789337e-05,
+      "loss": 0.676,
+      "step": 1845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6366434742075407e-05,
+      "loss": 0.5646,
+      "step": 1846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.636129621116648e-05,
+      "loss": 0.578,
+      "step": 1847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.635615485734285e-05,
+      "loss": 0.5382,
+      "step": 1848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6351010682886074e-05,
+      "loss": 0.5597,
+      "step": 1849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6345863690078942e-05,
+      "loss": 0.6044,
+      "step": 1850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.634071388120551e-05,
+      "loss": 0.5547,
+      "step": 1851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6335561258551074e-05,
+      "loss": 0.566,
+      "step": 1852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.633040582440219e-05,
+      "loss": 0.5357,
+      "step": 1853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6325247581046647e-05,
+      "loss": 0.6627,
+      "step": 1854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6320086530773493e-05,
+      "loss": 0.5988,
+      "step": 1855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6314922675873012e-05,
+      "loss": 0.6008,
+      "step": 1856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.630975601863674e-05,
+      "loss": 0.5929,
+      "step": 1857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6304586561357453e-05,
+      "loss": 0.5504,
+      "step": 1858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.629941430632917e-05,
+      "loss": 0.5474,
+      "step": 1859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6294239255847156e-05,
+      "loss": 0.5223,
+      "step": 1860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.628906141220791e-05,
+      "loss": 0.5869,
+      "step": 1861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6283880777709173e-05,
+      "loss": 0.5869,
+      "step": 1862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6278697354649916e-05,
+      "loss": 0.6088,
+      "step": 1863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6273511145330375e-05,
+      "loss": 0.5892,
+      "step": 1864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6268322152051984e-05,
+      "loss": 0.6274,
+      "step": 1865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.626313037711744e-05,
+      "loss": 0.6064,
+      "step": 1866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.625793582283067e-05,
+      "loss": 0.5912,
+      "step": 1867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6252738491496826e-05,
+      "loss": 0.5493,
+      "step": 1868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6247538385422296e-05,
+      "loss": 0.5556,
+      "step": 1869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6242335506914702e-05,
+      "loss": 0.5232,
+      "step": 1870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6237129858282897e-05,
+      "loss": 0.5366,
+      "step": 1871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6231921441836953e-05,
+      "loss": 0.605,
+      "step": 1872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6226710259888186e-05,
+      "loss": 0.5675,
+      "step": 1873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.622149631474913e-05,
+      "loss": 0.6386,
+      "step": 1874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.621627960873354e-05,
+      "loss": 0.6293,
+      "step": 1875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6211060144156414e-05,
+      "loss": 0.5744,
+      "step": 1876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6205837923333954e-05,
+      "loss": 0.5716,
+      "step": 1877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6200612948583597e-05,
+      "loss": 0.6024,
+      "step": 1878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6195385222223996e-05,
+      "loss": 0.5705,
+      "step": 1879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6190154746575033e-05,
+      "loss": 0.5413,
+      "step": 1880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6184921523957805e-05,
+      "loss": 0.5729,
+      "step": 1881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6179685556694623e-05,
+      "loss": 0.5378,
+      "step": 1882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6174446847109026e-05,
+      "loss": 0.5723,
+      "step": 1883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.616920539752576e-05,
+      "loss": 0.5755,
+      "step": 1884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.61639612102708e-05,
+      "loss": 0.6156,
+      "step": 1885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.615871428767132e-05,
+      "loss": 0.5732,
+      "step": 1886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.615346463205572e-05,
+      "loss": 0.5605,
+      "step": 1887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6148212245753608e-05,
+      "loss": 0.5832,
+      "step": 1888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.61429571310958e-05,
+      "loss": 0.5736,
+      "step": 1889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6137699290414337e-05,
+      "loss": 0.5884,
+      "step": 1890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.613243872604245e-05,
+      "loss": 0.586,
+      "step": 1891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6127175440314596e-05,
+      "loss": 0.5561,
+      "step": 1892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6121909435566422e-05,
+      "loss": 0.6349,
+      "step": 1893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6116640714134803e-05,
+      "loss": 0.6004,
+      "step": 1894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6111369278357803e-05,
+      "loss": 0.5969,
+      "step": 1895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6106095130574694e-05,
+      "loss": 0.5771,
+      "step": 1896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6100818273125956e-05,
+      "loss": 0.6398,
+      "step": 1897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6095538708353267e-05,
+      "loss": 0.5885,
+      "step": 1898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.609025643859951e-05,
+      "loss": 0.6165,
+      "step": 1899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6084971466208764e-05,
+      "loss": 0.5125,
+      "step": 1900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6079683793526314e-05,
+      "loss": 0.6003,
+      "step": 1901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6074393422898635e-05,
+      "loss": 0.6127,
+      "step": 1902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6069100356673406e-05,
+      "loss": 0.6339,
+      "step": 1903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.60638045971995e-05,
+      "loss": 0.6156,
+      "step": 1904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.605850614682698e-05,
+      "loss": 0.6449,
+      "step": 1905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6053205007907115e-05,
+      "loss": 0.6527,
+      "step": 1906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6047901182792355e-05,
+      "loss": 0.5844,
+      "step": 1907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6042594673836348e-05,
+      "loss": 0.5592,
+      "step": 1908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6037285483393933e-05,
+      "loss": 0.5613,
+      "step": 1909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.603197361382114e-05,
+      "loss": 0.5764,
+      "step": 1910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6026659067475178e-05,
+      "loss": 0.5506,
+      "step": 1911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6021341846714457e-05,
+      "loss": 0.5681,
+      "step": 1912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6016021953898572e-05,
+      "loss": 0.5993,
+      "step": 1913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6010699391388297e-05,
+      "loss": 0.607,
+      "step": 1914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.600537416154559e-05,
+      "loss": 0.6192,
+      "step": 1915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6000046266733608e-05,
+      "loss": 0.5944,
+      "step": 1916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5994715709316665e-05,
+      "loss": 0.612,
+      "step": 1917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5989382491660278e-05,
+      "loss": 0.5577,
+      "step": 1918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.598404661613114e-05,
+      "loss": 0.5623,
+      "step": 1919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5978708085097116e-05,
+      "loss": 0.5838,
+      "step": 1920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5973366900927257e-05,
+      "loss": 0.5817,
+      "step": 1921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5968023065991784e-05,
+      "loss": 0.6431,
+      "step": 1922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.596267658266211e-05,
+      "loss": 0.5884,
+      "step": 1923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.59573274533108e-05,
+      "loss": 0.6604,
+      "step": 1924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5951975680311612e-05,
+      "loss": 0.6051,
+      "step": 1925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5946621266039467e-05,
+      "loss": 0.5424,
+      "step": 1926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5941264212870463e-05,
+      "loss": 0.6183,
+      "step": 1927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5935904523181872e-05,
+      "loss": 0.5129,
+      "step": 1928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5930542199352122e-05,
+      "loss": 0.5787,
+      "step": 1929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.592517724376083e-05,
+      "loss": 0.623,
+      "step": 1930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5919809658788766e-05,
+      "loss": 0.5683,
+      "step": 1931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.591443944681787e-05,
+      "loss": 0.6542,
+      "step": 1932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5909066610231253e-05,
+      "loss": 0.5513,
+      "step": 1933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5903691151413184e-05,
+      "loss": 0.6127,
+      "step": 1934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5898313072749097e-05,
+      "loss": 0.5567,
+      "step": 1935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5892932376625593e-05,
+      "loss": 0.5392,
+      "step": 1936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.588754906543043e-05,
+      "loss": 0.6637,
+      "step": 1937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.588216314155253e-05,
+      "loss": 0.557,
+      "step": 1938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5876774607381972e-05,
+      "loss": 0.6002,
+      "step": 1939
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.587138346530999e-05,
+      "loss": 0.5648,
+      "step": 1940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5865989717728987e-05,
+      "loss": 0.5486,
+      "step": 1941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5860593367032504e-05,
+      "loss": 0.5403,
+      "step": 1942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5855194415615256e-05,
+      "loss": 0.5412,
+      "step": 1943
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.58497928658731e-05,
+      "loss": 0.632,
+      "step": 1944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5844388720203052e-05,
+      "loss": 0.6245,
+      "step": 1945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5838981981003273e-05,
+      "loss": 0.6051,
+      "step": 1946
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5833572650673085e-05,
+      "loss": 0.5825,
+      "step": 1947
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5828160731612957e-05,
+      "loss": 0.5702,
+      "step": 1948
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.582274622622449e-05,
+      "loss": 0.6525,
+      "step": 1949
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5817329136910463e-05,
+      "loss": 0.5147,
+      "step": 1950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5811909466074778e-05,
+      "loss": 0.5652,
+      "step": 1951
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5806487216122492e-05,
+      "loss": 0.6183,
+      "step": 1952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5801062389459803e-05,
+      "loss": 0.634,
+      "step": 1953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5795634988494054e-05,
+      "loss": 0.6271,
+      "step": 1954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.579020501563373e-05,
+      "loss": 0.5634,
+      "step": 1955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5784772473288463e-05,
+      "loss": 0.575,
+      "step": 1956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5779337363869018e-05,
+      "loss": 0.5555,
+      "step": 1957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5773899689787297e-05,
+      "loss": 0.5928,
+      "step": 1958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5768459453456345e-05,
+      "loss": 0.5798,
+      "step": 1959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.576301665729035e-05,
+      "loss": 0.6195,
+      "step": 1960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5757571303704622e-05,
+      "loss": 0.6165,
+      "step": 1961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5752123395115616e-05,
+      "loss": 0.6164,
+      "step": 1962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.574667293394092e-05,
+      "loss": 0.5709,
+      "step": 1963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5741219922599253e-05,
+      "loss": 0.5799,
+      "step": 1964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.573576436351046e-05,
+      "loss": 0.5935,
+      "step": 1965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.573030625909553e-05,
+      "loss": 0.6316,
+      "step": 1966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.572484561177657e-05,
+      "loss": 0.6082,
+      "step": 1967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5719382423976814e-05,
+      "loss": 0.5921,
+      "step": 1968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5713916698120634e-05,
+      "loss": 0.6115,
+      "step": 1969
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5708448436633522e-05,
+      "loss": 0.5751,
+      "step": 1970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.57029776419421e-05,
+      "loss": 0.6014,
+      "step": 1971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5697504316474102e-05,
+      "loss": 0.5849,
+      "step": 1972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.56920284626584e-05,
+      "loss": 0.5548,
+      "step": 1973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5686550082924973e-05,
+      "loss": 0.5945,
+      "step": 1974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5681069179704938e-05,
+      "loss": 0.5739,
+      "step": 1975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.567558575543052e-05,
+      "loss": 0.5574,
+      "step": 1976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5670099812535062e-05,
+      "loss": 0.4918,
+      "step": 1977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5664611353453035e-05,
+      "loss": 0.6378,
+      "step": 1978
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5659120380620013e-05,
+      "loss": 0.5641,
+      "step": 1979
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5653626896472698e-05,
+      "loss": 0.5446,
+      "step": 1980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5648130903448898e-05,
+      "loss": 0.5835,
+      "step": 1981
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5642632403987535e-05,
+      "loss": 0.6199,
+      "step": 1982
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5637131400528655e-05,
+      "loss": 0.6341,
+      "step": 1983
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.563162789551339e-05,
+      "loss": 0.6,
+      "step": 1984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5626121891384013e-05,
+      "loss": 0.6408,
+      "step": 1985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5620613390583883e-05,
+      "loss": 0.588,
+      "step": 1986
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.561510239555748e-05,
+      "loss": 0.611,
+      "step": 1987
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5609588908750378e-05,
+      "loss": 0.5797,
+      "step": 1988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5604072932609267e-05,
+      "loss": 0.5717,
+      "step": 1989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5598554469581945e-05,
+      "loss": 0.605,
+      "step": 1990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5593033522117297e-05,
+      "loss": 0.6337,
+      "step": 1991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.558751009266533e-05,
+      "loss": 0.6504,
+      "step": 1992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.558198418367714e-05,
+      "loss": 0.5528,
+      "step": 1993
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5576455797604932e-05,
+      "loss": 0.5814,
+      "step": 1994
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5570924936901994e-05,
+      "loss": 0.5852,
+      "step": 1995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5565391604022732e-05,
+      "loss": 0.5474,
+      "step": 1996
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.555985580142264e-05,
+      "loss": 0.5918,
+      "step": 1997
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5554317531558302e-05,
+      "loss": 0.5304,
+      "step": 1998
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.554877679688741e-05,
+      "loss": 0.6232,
+      "step": 1999
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5543233599868744e-05,
+      "loss": 0.5858,
+      "step": 2000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.553768794296217e-05,
+      "loss": 0.6113,
+      "step": 2001
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5532139828628653e-05,
+      "loss": 0.5974,
+      "step": 2002
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5526589259330248e-05,
+      "loss": 0.6408,
+      "step": 2003
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5521036237530096e-05,
+      "loss": 0.6726,
+      "step": 2004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.551548076569243e-05,
+      "loss": 0.5472,
+      "step": 2005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.550992284628257e-05,
+      "loss": 0.5466,
+      "step": 2006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.550436248176692e-05,
+      "loss": 0.5863,
+      "step": 2007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.549879967461297e-05,
+      "loss": 0.5667,
+      "step": 2008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5493234427289293e-05,
+      "loss": 0.5681,
+      "step": 2009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5487666742265547e-05,
+      "loss": 0.5924,
+      "step": 2010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.548209662201247e-05,
+      "loss": 0.5613,
+      "step": 2011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.547652406900188e-05,
+      "loss": 0.5467,
+      "step": 2012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.547094908570668e-05,
+      "loss": 0.566,
+      "step": 2013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.546537167460084e-05,
+      "loss": 0.5763,
+      "step": 2014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5459791838159426e-05,
+      "loss": 0.5615,
+      "step": 2015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5454209578858555e-05,
+      "loss": 0.6346,
+      "step": 2016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5448624899175447e-05,
+      "loss": 0.6132,
+      "step": 2017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.544303780158837e-05,
+      "loss": 0.5932,
+      "step": 2018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5437448288576685e-05,
+      "loss": 0.5598,
+      "step": 2019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5431856362620815e-05,
+      "loss": 0.5526,
+      "step": 2020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5426262026202254e-05,
+      "loss": 0.5982,
+      "step": 2021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.542066528180357e-05,
+      "loss": 0.5659,
+      "step": 2022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5415066131908395e-05,
+      "loss": 0.5983,
+      "step": 2023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5409464579001433e-05,
+      "loss": 0.5978,
+      "step": 2024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.540386062556845e-05,
+      "loss": 0.5819,
+      "step": 2025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.539825427409628e-05,
+      "loss": 0.5957,
+      "step": 2026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5392645527072816e-05,
+      "loss": 0.6132,
+      "step": 2027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5387034386987026e-05,
+      "loss": 0.6088,
+      "step": 2028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.538142085632893e-05,
+      "loss": 0.6395,
+      "step": 2029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5375804937589606e-05,
+      "loss": 0.5669,
+      "step": 2030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5370186633261204e-05,
+      "loss": 0.5689,
+      "step": 2031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.536456594583692e-05,
+      "loss": 0.5614,
+      "step": 2032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.535894287781102e-05,
+      "loss": 0.5254,
+      "step": 2033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5353317431678812e-05,
+      "loss": 0.6319,
+      "step": 2034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.534768960993667e-05,
+      "loss": 0.6174,
+      "step": 2035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.534205941508202e-05,
+      "loss": 0.5758,
+      "step": 2036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.533642684961334e-05,
+      "loss": 0.5959,
+      "step": 2037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5330791916030155e-05,
+      "loss": 0.5479,
+      "step": 2038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5325154616833056e-05,
+      "loss": 0.5665,
+      "step": 2039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5319514954523667e-05,
+      "loss": 0.5624,
+      "step": 2040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5313872931604667e-05,
+      "loss": 0.6197,
+      "step": 2041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5308228550579788e-05,
+      "loss": 0.5898,
+      "step": 2042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.53025818139538e-05,
+      "loss": 0.5742,
+      "step": 2043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.529693272423252e-05,
+      "loss": 0.595,
+      "step": 2044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5291281283922813e-05,
+      "loss": 0.5659,
+      "step": 2045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5285627495532592e-05,
+      "loss": 0.6741,
+      "step": 2046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5279971361570793e-05,
+      "loss": 0.5764,
+      "step": 2047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.527431288454741e-05,
+      "loss": 0.5573,
+      "step": 2048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5268652066973476e-05,
+      "loss": 0.5599,
+      "step": 2049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.526298891136105e-05,
+      "loss": 0.6288,
+      "step": 2050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5257323420223249e-05,
+      "loss": 0.5755,
+      "step": 2051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.52516555960742e-05,
+      "loss": 0.5938,
+      "step": 2052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5245985441429093e-05,
+      "loss": 0.5719,
+      "step": 2053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5240312958804132e-05,
+      "loss": 0.5814,
+      "step": 2054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5234638150716559e-05,
+      "loss": 0.5556,
+      "step": 2055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5228961019684652e-05,
+      "loss": 0.5969,
+      "step": 2056
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5223281568227726e-05,
+      "loss": 0.5711,
+      "step": 2057
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5217599798866102e-05,
+      "loss": 0.5908,
+      "step": 2058
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5211915714121157e-05,
+      "loss": 0.5202,
+      "step": 2059
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.520622931651528e-05,
+      "loss": 0.5343,
+      "step": 2060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5200540608571889e-05,
+      "loss": 0.6301,
+      "step": 2061
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5194849592815429e-05,
+      "loss": 0.5909,
+      "step": 2062
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5189156271771368e-05,
+      "loss": 0.6144,
+      "step": 2063
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.51834606479662e-05,
+      "loss": 0.5899,
+      "step": 2064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5177762723927432e-05,
+      "loss": 0.6527,
+      "step": 2065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5172062502183603e-05,
+      "loss": 0.6364,
+      "step": 2066
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5166359985264268e-05,
+      "loss": 0.654,
+      "step": 2067
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5160655175699994e-05,
+      "loss": 0.5509,
+      "step": 2068
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5154948076022372e-05,
+      "loss": 0.6082,
+      "step": 2069
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5149238688764012e-05,
+      "loss": 0.5532,
+      "step": 2070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.514352701645853e-05,
+      "loss": 0.5361,
+      "step": 2071
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5137813061640563e-05,
+      "loss": 0.5207,
+      "step": 2072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5132096826845756e-05,
+      "loss": 0.6158,
+      "step": 2073
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5126378314610774e-05,
+      "loss": 0.6017,
+      "step": 2074
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5120657527473286e-05,
+      "loss": 0.5887,
+      "step": 2075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5114934467971969e-05,
+      "loss": 0.5716,
+      "step": 2076
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5109209138646507e-05,
+      "loss": 0.5902,
+      "step": 2077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5103481542037608e-05,
+      "loss": 0.6182,
+      "step": 2078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5097751680686963e-05,
+      "loss": 0.5756,
+      "step": 2079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5092019557137282e-05,
+      "loss": 0.5617,
+      "step": 2080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5086285173932279e-05,
+      "loss": 0.5632,
+      "step": 2081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5080548533616659e-05,
+      "loss": 0.5663,
+      "step": 2082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5074809638736146e-05,
+      "loss": 0.5539,
+      "step": 2083
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5069068491837447e-05,
+      "loss": 0.5495,
+      "step": 2084
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5063325095468284e-05,
+      "loss": 0.5666,
+      "step": 2085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5057579452177362e-05,
+      "loss": 0.623,
+      "step": 2086
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5051831564514402e-05,
+      "loss": 0.5546,
+      "step": 2087
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5046081435030097e-05,
+      "loss": 0.5858,
+      "step": 2088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.504032906627616e-05,
+      "loss": 0.5846,
+      "step": 2089
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5034574460805279e-05,
+      "loss": 0.5587,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5028817621171138e-05,
+      "loss": 0.562,
+      "step": 2091
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5023058549928425e-05,
+      "loss": 0.6197,
+      "step": 2092
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5017297249632803e-05,
+      "loss": 0.6293,
+      "step": 2093
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5011533722840928e-05,
+      "loss": 0.6213,
+      "step": 2094
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.500576797211045e-05,
+      "loss": 0.6366,
+      "step": 2095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.5364,
+      "step": 2096
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.49942298090692e-05,
+      "loss": 0.5731,
+      "step": 2097
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.498845740187865e-05,
+      "loss": 0.6463,
+      "step": 2098
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4982682780989941e-05,
+      "loss": 0.5684,
+      "step": 2099
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4976905948965637e-05,
+      "loss": 0.6152,
+      "step": 2100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4971126908369289e-05,
+      "loss": 0.5246,
+      "step": 2101
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.496534566176543e-05,
+      "loss": 0.6041,
+      "step": 2102
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4959562211719572e-05,
+      "loss": 0.5834,
+      "step": 2103
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4953776560798197e-05,
+      "loss": 0.5828,
+      "step": 2104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4947988711568771e-05,
+      "loss": 0.5953,
+      "step": 2105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4942198666599734e-05,
+      "loss": 0.5752,
+      "step": 2106
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.49364064284605e-05,
+      "loss": 0.6112,
+      "step": 2107
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4930611999721457e-05,
+      "loss": 0.4933,
+      "step": 2108
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.492481538295396e-05,
+      "loss": 0.6079,
+      "step": 2109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4919016580730345e-05,
+      "loss": 0.6234,
+      "step": 2110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4913215595623908e-05,
+      "loss": 0.591,
+      "step": 2111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4907412430208923e-05,
+      "loss": 0.6007,
+      "step": 2112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4901607087060617e-05,
+      "loss": 0.5693,
+      "step": 2113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4895799568755203e-05,
+      "loss": 0.562,
+      "step": 2114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4889989877869842e-05,
+      "loss": 0.5707,
+      "step": 2115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4884178016982667e-05,
+      "loss": 0.5503,
+      "step": 2116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4878363988672779e-05,
+      "loss": 0.6314,
+      "step": 2117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4872547795520224e-05,
+      "loss": 0.5677,
+      "step": 2118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4866729440106031e-05,
+      "loss": 0.6284,
+      "step": 2119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.486090892501217e-05,
+      "loss": 0.6229,
+      "step": 2120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.485508625282158e-05,
+      "loss": 0.5528,
+      "step": 2121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4849261426118147e-05,
+      "loss": 0.6169,
+      "step": 2122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.484343444748673e-05,
+      "loss": 0.5638,
+      "step": 2123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4837605319513127e-05,
+      "loss": 0.6572,
+      "step": 2124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4831774044784095e-05,
+      "loss": 0.5867,
+      "step": 2125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4825940625887343e-05,
+      "loss": 0.5946,
+      "step": 2126
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.482010506541154e-05,
+      "loss": 0.569,
+      "step": 2127
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4814267365946288e-05,
+      "loss": 0.6616,
+      "step": 2128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4808427530082154e-05,
+      "loss": 0.5945,
+      "step": 2129
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4802585560410652e-05,
+      "loss": 0.5672,
+      "step": 2130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.479674145952423e-05,
+      "loss": 0.5771,
+      "step": 2131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4790895230016294e-05,
+      "loss": 0.5551,
+      "step": 2132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4785046874481186e-05,
+      "loss": 0.5475,
+      "step": 2133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4779196395514201e-05,
+      "loss": 0.6107,
+      "step": 2134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4773343795711572e-05,
+      "loss": 0.5702,
+      "step": 2135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4767489077670472e-05,
+      "loss": 0.5392,
+      "step": 2136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4761632243989008e-05,
+      "loss": 0.5678,
+      "step": 2137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4755773297266243e-05,
+      "loss": 0.5658,
+      "step": 2138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4749912240102158e-05,
+      "loss": 0.5252,
+      "step": 2139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4744049075097682e-05,
+      "loss": 0.5753,
+      "step": 2140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.473818380485468e-05,
+      "loss": 0.5269,
+      "step": 2141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4732316431975946e-05,
+      "loss": 0.5638,
+      "step": 2142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4726446959065209e-05,
+      "loss": 0.5621,
+      "step": 2143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4720575388727132e-05,
+      "loss": 0.5795,
+      "step": 2144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4714701723567303e-05,
+      "loss": 0.5757,
+      "step": 2145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4708825966192249e-05,
+      "loss": 0.5831,
+      "step": 2146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4702948119209415e-05,
+      "loss": 0.6257,
+      "step": 2147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.469706818522718e-05,
+      "loss": 0.5681,
+      "step": 2148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4691186166854851e-05,
+      "loss": 0.6045,
+      "step": 2149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.468530206670265e-05,
+      "loss": 0.5891,
+      "step": 2150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4679415887381736e-05,
+      "loss": 0.5865,
+      "step": 2151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.467352763150418e-05,
+      "loss": 0.6724,
+      "step": 2152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4667637301682978e-05,
+      "loss": 0.579,
+      "step": 2153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4661744900532044e-05,
+      "loss": 0.6238,
+      "step": 2154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4655850430666222e-05,
+      "loss": 0.6028,
+      "step": 2155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4649953894701261e-05,
+      "loss": 0.5582,
+      "step": 2156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4644055295253831e-05,
+      "loss": 0.5644,
+      "step": 2157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4638154634941519e-05,
+      "loss": 0.5636,
+      "step": 2158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.463225191638283e-05,
+      "loss": 0.5953,
+      "step": 2159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4626347142197173e-05,
+      "loss": 0.5814,
+      "step": 2160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4620440315004876e-05,
+      "loss": 0.5887,
+      "step": 2161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4614531437427179e-05,
+      "loss": 0.6225,
+      "step": 2162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4608620512086229e-05,
+      "loss": 0.6011,
+      "step": 2163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.460270754160508e-05,
+      "loss": 0.5094,
+      "step": 2164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4596792528607698e-05,
+      "loss": 0.6279,
+      "step": 2165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4590875475718953e-05,
+      "loss": 0.6287,
+      "step": 2166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4584956385564617e-05,
+      "loss": 0.5345,
+      "step": 2167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4579035260771374e-05,
+      "loss": 0.5646,
+      "step": 2168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4573112103966805e-05,
+      "loss": 0.5741,
+      "step": 2169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4567186917779397e-05,
+      "loss": 0.5959,
+      "step": 2170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4561259704838526e-05,
+      "loss": 0.593,
+      "step": 2171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4555330467774487e-05,
+      "loss": 0.6071,
+      "step": 2172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4549399209218455e-05,
+      "loss": 0.6189,
+      "step": 2173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4543465931802512e-05,
+      "loss": 0.5951,
+      "step": 2174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4537530638159633e-05,
+      "loss": 0.5997,
+      "step": 2175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.453159333092369e-05,
+      "loss": 0.5669,
+      "step": 2176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4525654012729443e-05,
+      "loss": 0.5707,
+      "step": 2177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.451971268621255e-05,
+      "loss": 0.5603,
+      "step": 2178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4513769354009557e-05,
+      "loss": 0.5499,
+      "step": 2179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4507824018757906e-05,
+      "loss": 0.602,
+      "step": 2180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4501876683095917e-05,
+      "loss": 0.5443,
+      "step": 2181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4495927349662805e-05,
+      "loss": 0.626,
+      "step": 2182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4489976021098675e-05,
+      "loss": 0.5879,
+      "step": 2183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4484022700044509e-05,
+      "loss": 0.5837,
+      "step": 2184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4478067389142177e-05,
+      "loss": 0.5298,
+      "step": 2185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.447211009103443e-05,
+      "loss": 0.5342,
+      "step": 2186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4466150808364911e-05,
+      "loss": 0.5798,
+      "step": 2187
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4460189543778124e-05,
+      "loss": 0.5729,
+      "step": 2188
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4454226299919477e-05,
+      "loss": 0.5639,
+      "step": 2189
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4448261079435231e-05,
+      "loss": 0.5752,
+      "step": 2190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4442293884972542e-05,
+      "loss": 0.5867,
+      "step": 2191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.443632471917944e-05,
+      "loss": 0.5417,
+      "step": 2192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4430353584704819e-05,
+      "loss": 0.5445,
+      "step": 2193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4424380484198462e-05,
+      "loss": 0.5453,
+      "step": 2194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4418405420311007e-05,
+      "loss": 0.5443,
+      "step": 2195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.441242839569398e-05,
+      "loss": 0.6014,
+      "step": 2196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4406449412999767e-05,
+      "loss": 0.5597,
+      "step": 2197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.440046847488163e-05,
+      "loss": 0.558,
+      "step": 2198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4394485583993687e-05,
+      "loss": 0.5833,
+      "step": 2199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4388500742990934e-05,
+      "loss": 0.5967,
+      "step": 2200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.438251395452923e-05,
+      "loss": 0.5118,
+      "step": 2201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4376525221265291e-05,
+      "loss": 0.5729,
+      "step": 2202
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4370534545856707e-05,
+      "loss": 0.6189,
+      "step": 2203
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4364541930961927e-05,
+      "loss": 0.5729,
+      "step": 2204
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4358547379240249e-05,
+      "loss": 0.6375,
+      "step": 2205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4352550893351845e-05,
+      "loss": 0.5717,
+      "step": 2206
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4346552475957743e-05,
+      "loss": 0.5804,
+      "step": 2207
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4340552129719818e-05,
+      "loss": 0.6206,
+      "step": 2208
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4334549857300815e-05,
+      "loss": 0.6092,
+      "step": 2209
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4328545661364321e-05,
+      "loss": 0.6092,
+      "step": 2210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4322539544574787e-05,
+      "loss": 0.6199,
+      "step": 2211
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.431653150959751e-05,
+      "loss": 0.5763,
+      "step": 2212
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.431052155909864e-05,
+      "loss": 0.652,
+      "step": 2213
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4304509695745177e-05,
+      "loss": 0.5566,
+      "step": 2214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.429849592220497e-05,
+      "loss": 0.5673,
+      "step": 2215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4292480241146715e-05,
+      "loss": 0.5428,
+      "step": 2216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4286462655239959e-05,
+      "loss": 0.6023,
+      "step": 2217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4280443167155082e-05,
+      "loss": 0.574,
+      "step": 2218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4274421779563328e-05,
+      "loss": 0.6157,
+      "step": 2219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4268398495136762e-05,
+      "loss": 0.5563,
+      "step": 2220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4262373316548305e-05,
+      "loss": 0.5719,
+      "step": 2221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4256346246471716e-05,
+      "loss": 0.5774,
+      "step": 2222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4250317287581592e-05,
+      "loss": 0.5558,
+      "step": 2223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.424428644255337e-05,
+      "loss": 0.5302,
+      "step": 2224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4238253714063315e-05,
+      "loss": 0.5463,
+      "step": 2225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4232219104788544e-05,
+      "loss": 0.5547,
+      "step": 2226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4226182617406996e-05,
+      "loss": 0.6028,
+      "step": 2227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4220144254597445e-05,
+      "loss": 0.5949,
+      "step": 2228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4214104019039503e-05,
+      "loss": 0.5454,
+      "step": 2229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4208061913413612e-05,
+      "loss": 0.6004,
+      "step": 2230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4202017940401032e-05,
+      "loss": 0.5379,
+      "step": 2231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4195972102683869e-05,
+      "loss": 0.5823,
+      "step": 2232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4189924402945045e-05,
+      "loss": 0.5939,
+      "step": 2233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4183874843868316e-05,
+      "loss": 0.6009,
+      "step": 2234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4177823428138251e-05,
+      "loss": 0.5624,
+      "step": 2235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4171770158440259e-05,
+      "loss": 0.5702,
+      "step": 2236
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4165715037460559e-05,
+      "loss": 0.5818,
+      "step": 2237
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4159658067886199e-05,
+      "loss": 0.5713,
+      "step": 2238
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4153599252405044e-05,
+      "loss": 0.5839,
+      "step": 2239
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4147538593705776e-05,
+      "loss": 0.5954,
+      "step": 2240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.41414760944779e-05,
+      "loss": 0.5916,
+      "step": 2241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4135411757411732e-05,
+      "loss": 0.5524,
+      "step": 2242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4129345585198416e-05,
+      "loss": 0.5415,
+      "step": 2243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4123277580529888e-05,
+      "loss": 0.5711,
+      "step": 2244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4117207746098922e-05,
+      "loss": 0.6182,
+      "step": 2245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4111136084599084e-05,
+      "loss": 0.6502,
+      "step": 2246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4105062598724769e-05,
+      "loss": 0.6327,
+      "step": 2247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4098987291171158e-05,
+      "loss": 0.5981,
+      "step": 2248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4092910164634266e-05,
+      "loss": 0.5848,
+      "step": 2249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4086831221810897e-05,
+      "loss": 0.5283,
+      "step": 2250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4080750465398674e-05,
+      "loss": 0.5704,
+      "step": 2251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.407466789809601e-05,
+      "loss": 0.5859,
+      "step": 2252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4068583522602137e-05,
+      "loss": 0.5968,
+      "step": 2253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4062497341617076e-05,
+      "loss": 0.6152,
+      "step": 2254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4056409357841658e-05,
+      "loss": 0.5656,
+      "step": 2255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4050319573977516e-05,
+      "loss": 0.5319,
+      "step": 2256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4044227992727069e-05,
+      "loss": 0.5859,
+      "step": 2257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4038134616793551e-05,
+      "loss": 0.5666,
+      "step": 2258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4032039448880972e-05,
+      "loss": 0.5765,
+      "step": 2259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.402594249169416e-05,
+      "loss": 0.653,
+      "step": 2260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4019843747938719e-05,
+      "loss": 0.5671,
+      "step": 2261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4013743220321054e-05,
+      "loss": 0.5448,
+      "step": 2262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.400764091154836e-05,
+      "loss": 0.5888,
+      "step": 2263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.400153682432862e-05,
+      "loss": 0.5428,
+      "step": 2264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3995430961370613e-05,
+      "loss": 0.5483,
+      "step": 2265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3989323325383901e-05,
+      "loss": 0.5841,
+      "step": 2266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3983213919078826e-05,
+      "loss": 0.6311,
+      "step": 2267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3977102745166533e-05,
+      "loss": 0.5619,
+      "step": 2268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3970989806358937e-05,
+      "loss": 0.5394,
+      "step": 2269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3964875105368741e-05,
+      "loss": 0.5723,
+      "step": 2270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3958758644909427e-05,
+      "loss": 0.6015,
+      "step": 2271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3952640427695265e-05,
+      "loss": 0.5892,
+      "step": 2272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3946520456441299e-05,
+      "loss": 0.5098,
+      "step": 2273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3940398733863348e-05,
+      "loss": 0.5651,
+      "step": 2274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3934275262678019e-05,
+      "loss": 0.5673,
+      "step": 2275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3928150045602682e-05,
+      "loss": 0.6517,
+      "step": 2276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3922023085355494e-05,
+      "loss": 0.5623,
+      "step": 2277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3915894384655373e-05,
+      "loss": 0.5734,
+      "step": 2278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3909763946222026e-05,
+      "loss": 0.594,
+      "step": 2279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3903631772775912e-05,
+      "loss": 0.5852,
+      "step": 2280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3897497867038274e-05,
+      "loss": 0.5369,
+      "step": 2281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3891362231731118e-05,
+      "loss": 0.5825,
+      "step": 2282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3885224869577218e-05,
+      "loss": 0.5766,
+      "step": 2283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3879085783300114e-05,
+      "loss": 0.5922,
+      "step": 2284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3872944975624117e-05,
+      "loss": 0.5616,
+      "step": 2285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3866802449274291e-05,
+      "loss": 0.56,
+      "step": 2286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3860658206976468e-05,
+      "loss": 0.5735,
+      "step": 2287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3854512251457248e-05,
+      "loss": 0.6018,
+      "step": 2288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.384836458544398e-05,
+      "loss": 0.5802,
+      "step": 2289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.384221521166478e-05,
+      "loss": 0.5328,
+      "step": 2290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3836064132848516e-05,
+      "loss": 0.5403,
+      "step": 2291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.382991135172482e-05,
+      "loss": 0.5315,
+      "step": 2292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3823756871024072e-05,
+      "loss": 0.5459,
+      "step": 2293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3817600693477414e-05,
+      "loss": 0.5559,
+      "step": 2294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3811442821816727e-05,
+      "loss": 0.5803,
+      "step": 2295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3805283258774662e-05,
+      "loss": 0.5912,
+      "step": 2296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3799122007084609e-05,
+      "loss": 0.5124,
+      "step": 2297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.379295906948071e-05,
+      "loss": 0.59,
+      "step": 2298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3786794448697854e-05,
+      "loss": 0.5176,
+      "step": 2299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.378062814747168e-05,
+      "loss": 0.5794,
+      "step": 2300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3774460168538573e-05,
+      "loss": 0.5477,
+      "step": 2301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3768290514635654e-05,
+      "loss": 0.539,
+      "step": 2302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3762119188500798e-05,
+      "loss": 0.5687,
+      "step": 2303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.375594619287262e-05,
+      "loss": 0.5567,
+      "step": 2304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.374977153049047e-05,
+      "loss": 0.588,
+      "step": 2305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3743595204094442e-05,
+      "loss": 0.6386,
+      "step": 2306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3737417216425368e-05,
+      "loss": 0.5502,
+      "step": 2307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3731237570224817e-05,
+      "loss": 0.6292,
+      "step": 2308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3725056268235095e-05,
+      "loss": 0.6335,
+      "step": 2309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.371887331319924e-05,
+      "loss": 0.5732,
+      "step": 2310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3712688707861027e-05,
+      "loss": 0.5853,
+      "step": 2311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.370650245496496e-05,
+      "loss": 0.5867,
+      "step": 2312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3700314557256276e-05,
+      "loss": 0.6056,
+      "step": 2313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3694125017480942e-05,
+      "loss": 0.5728,
+      "step": 2314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3687933838385654e-05,
+      "loss": 0.5769,
+      "step": 2315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3681741022717832e-05,
+      "loss": 0.4851,
+      "step": 2316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3675546573225631e-05,
+      "loss": 0.6179,
+      "step": 2317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.366935049265792e-05,
+      "loss": 0.5378,
+      "step": 2318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3663152783764298e-05,
+      "loss": 0.5548,
+      "step": 2319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3656953449295086e-05,
+      "loss": 0.521,
+      "step": 2320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3650752492001327e-05,
+      "loss": 0.5806,
+      "step": 2321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3644549914634781e-05,
+      "loss": 0.5945,
+      "step": 2322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3638345719947928e-05,
+      "loss": 0.5774,
+      "step": 2323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3632139910693971e-05,
+      "loss": 0.6661,
+      "step": 2324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3625932489626823e-05,
+      "loss": 0.6011,
+      "step": 2325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3619723459501116e-05,
+      "loss": 0.5588,
+      "step": 2326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3613512823072188e-05,
+      "loss": 0.5772,
+      "step": 2327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3607300583096106e-05,
+      "loss": 0.6221,
+      "step": 2328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.360108674232963e-05,
+      "loss": 0.5707,
+      "step": 2329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3594871303530247e-05,
+      "loss": 0.5745,
+      "step": 2330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3588654269456134e-05,
+      "loss": 0.5658,
+      "step": 2331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.35824356428662e-05,
+      "loss": 0.595,
+      "step": 2332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3576215426520046e-05,
+      "loss": 0.5701,
+      "step": 2333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3569993623177973e-05,
+      "loss": 0.5731,
+      "step": 2334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3563770235601001e-05,
+      "loss": 0.5705,
+      "step": 2335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3557545266550842e-05,
+      "loss": 0.5543,
+      "step": 2336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3551318718789914e-05,
+      "loss": 0.6036,
+      "step": 2337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3545090595081335e-05,
+      "loss": 0.626,
+      "step": 2338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3538860898188928e-05,
+      "loss": 0.5633,
+      "step": 2339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3532629630877199e-05,
+      "loss": 0.5485,
+      "step": 2340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3526396795911367e-05,
+      "loss": 0.5627,
+      "step": 2341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3520162396057342e-05,
+      "loss": 0.6154,
+      "step": 2342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.351392643408172e-05,
+      "loss": 0.5832,
+      "step": 2343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.35076889127518e-05,
+      "loss": 0.5375,
+      "step": 2344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3501449834835571e-05,
+      "loss": 0.5652,
+      "step": 2345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3495209203101711e-05,
+      "loss": 0.5475,
+      "step": 2346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.348896702031959e-05,
+      "loss": 0.5806,
+      "step": 2347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3482723289259258e-05,
+      "loss": 0.6037,
+      "step": 2348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3476478012691466e-05,
+      "loss": 0.5955,
+      "step": 2349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3470231193387639e-05,
+      "loss": 0.5411,
+      "step": 2350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3463982834119894e-05,
+      "loss": 0.5466,
+      "step": 2351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3457732937661025e-05,
+      "loss": 0.5943,
+      "step": 2352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3451481506784515e-05,
+      "loss": 0.5428,
+      "step": 2353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.344522854426452e-05,
+      "loss": 0.5744,
+      "step": 2354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3438974052875881e-05,
+      "loss": 0.5807,
+      "step": 2355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3432718035394124e-05,
+      "loss": 0.5791,
+      "step": 2356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3426460494595435e-05,
+      "loss": 0.5971,
+      "step": 2357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.6092,
+      "step": 2358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3413940854155434e-05,
+      "loss": 0.5521,
+      "step": 2359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.340767876006989e-05,
+      "loss": 0.5699,
+      "step": 2360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3401415153778948e-05,
+      "loss": 0.6049,
+      "step": 2361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3395150038062173e-05,
+      "loss": 0.591,
+      "step": 2362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3388883415699796e-05,
+      "loss": 0.6039,
+      "step": 2363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3382615289472721e-05,
+      "loss": 0.5765,
+      "step": 2364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3376345662162515e-05,
+      "loss": 0.5623,
+      "step": 2365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3370074536551414e-05,
+      "loss": 0.5764,
+      "step": 2366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3363801915422325e-05,
+      "loss": 0.5167,
+      "step": 2367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3357527801558799e-05,
+      "loss": 0.5894,
+      "step": 2368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3351252197745072e-05,
+      "loss": 0.5622,
+      "step": 2369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.334497510676603e-05,
+      "loss": 0.6188,
+      "step": 2370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3338696531407216e-05,
+      "loss": 0.5783,
+      "step": 2371
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3332416474454841e-05,
+      "loss": 0.6032,
+      "step": 2372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3326134938695767e-05,
+      "loss": 0.5857,
+      "step": 2373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3319851926917512e-05,
+      "loss": 0.5417,
+      "step": 2374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3313567441908255e-05,
+      "loss": 0.5833,
+      "step": 2375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.330728148645682e-05,
+      "loss": 0.5449,
+      "step": 2376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3300994063352693e-05,
+      "loss": 0.5354,
+      "step": 2377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3294705175386002e-05,
+      "loss": 0.5721,
+      "step": 2378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3288414825347531e-05,
+      "loss": 0.5904,
+      "step": 2379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3282123016028712e-05,
+      "loss": 0.5886,
+      "step": 2380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3275829750221625e-05,
+      "loss": 0.505,
+      "step": 2381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3269535030718995e-05,
+      "loss": 0.591,
+      "step": 2382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3263238860314184e-05,
+      "loss": 0.5319,
+      "step": 2383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3256941241801222e-05,
+      "loss": 0.5891,
+      "step": 2384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3250642177974754e-05,
+      "loss": 0.5584,
+      "step": 2385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3244341671630081e-05,
+      "loss": 0.5257,
+      "step": 2386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3238039725563142e-05,
+      "loss": 0.5288,
+      "step": 2387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3231736342570518e-05,
+      "loss": 0.5862,
+      "step": 2388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3225431525449415e-05,
+      "loss": 0.5435,
+      "step": 2389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.321912527699769e-05,
+      "loss": 0.512,
+      "step": 2390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.321281760001383e-05,
+      "loss": 0.5406,
+      "step": 2391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3206508497296957e-05,
+      "loss": 0.5452,
+      "step": 2392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3200197971646816e-05,
+      "loss": 0.5326,
+      "step": 2393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.31938860258638e-05,
+      "loss": 0.5912,
+      "step": 2394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.318757266274892e-05,
+      "loss": 0.5825,
+      "step": 2395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3181257885103818e-05,
+      "loss": 0.5098,
+      "step": 2396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.317494169573077e-05,
+      "loss": 0.6157,
+      "step": 2397
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3168624097432669e-05,
+      "loss": 0.5822,
+      "step": 2398
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3162305093013048e-05,
+      "loss": 0.5879,
+      "step": 2399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.315598468527604e-05,
+      "loss": 0.5541,
+      "step": 2400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3149662877026423e-05,
+      "loss": 0.5897,
+      "step": 2401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3143339671069591e-05,
+      "loss": 0.5655,
+      "step": 2402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3137015070211548e-05,
+      "loss": 0.5902,
+      "step": 2403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3130689077258927e-05,
+      "loss": 0.5746,
+      "step": 2404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3124361695018981e-05,
+      "loss": 0.6066,
+      "step": 2405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.311803292629957e-05,
+      "loss": 0.551,
+      "step": 2406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3111702773909176e-05,
+      "loss": 0.5927,
+      "step": 2407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3105371240656894e-05,
+      "loss": 0.5788,
+      "step": 2408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3099038329352429e-05,
+      "loss": 0.6027,
+      "step": 2409
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3092704042806101e-05,
+      "loss": 0.548,
+      "step": 2410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.308636838382884e-05,
+      "loss": 0.5486,
+      "step": 2411
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3080031355232179e-05,
+      "loss": 0.5139,
+      "step": 2412
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3073692959828267e-05,
+      "loss": 0.5338,
+      "step": 2413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3067353200429858e-05,
+      "loss": 0.5503,
+      "step": 2414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3061012079850304e-05,
+      "loss": 0.5634,
+      "step": 2415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3054669600903573e-05,
+      "loss": 0.5535,
+      "step": 2416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3048325766404223e-05,
+      "loss": 0.5418,
+      "step": 2417
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3041980579167426e-05,
+      "loss": 0.5557,
+      "step": 2418
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3035634042008942e-05,
+      "loss": 0.5591,
+      "step": 2419
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3029286157745142e-05,
+      "loss": 0.6036,
+      "step": 2420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3022936929192984e-05,
+      "loss": 0.544,
+      "step": 2421
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3016586359170032e-05,
+      "loss": 0.5284,
+      "step": 2422
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3010234450494437e-05,
+      "loss": 0.5187,
+      "step": 2423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3003881205984947e-05,
+      "loss": 0.499,
+      "step": 2424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2997526628460908e-05,
+      "loss": 0.5177,
+      "step": 2425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.299117072074225e-05,
+      "loss": 0.5783,
+      "step": 2426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2984813485649498e-05,
+      "loss": 0.5769,
+      "step": 2427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2978454926003762e-05,
+      "loss": 0.639,
+      "step": 2428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.297209504462675e-05,
+      "loss": 0.5596,
+      "step": 2429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.296573384434074e-05,
+      "loss": 0.548,
+      "step": 2430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2959371327968608e-05,
+      "loss": 0.5931,
+      "step": 2431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2953007498333807e-05,
+      "loss": 0.555,
+      "step": 2432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2946642358260382e-05,
+      "loss": 0.5698,
+      "step": 2433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2940275910572948e-05,
+      "loss": 0.5955,
+      "step": 2434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.293390815809671e-05,
+      "loss": 0.6076,
+      "step": 2435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.292753910365744e-05,
+      "loss": 0.6105,
+      "step": 2436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2921168750081507e-05,
+      "loss": 0.5956,
+      "step": 2437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2914797100195835e-05,
+      "loss": 0.5486,
+      "step": 2438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2908424156827939e-05,
+      "loss": 0.5709,
+      "step": 2439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.29020499228059e-05,
+      "loss": 0.5082,
+      "step": 2440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2895674400958373e-05,
+      "loss": 0.6173,
+      "step": 2441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.288929759411459e-05,
+      "loss": 0.5102,
+      "step": 2442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2882919505104346e-05,
+      "loss": 0.5471,
+      "step": 2443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2876540136758005e-05,
+      "loss": 0.5598,
+      "step": 2444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2870159491906504e-05,
+      "loss": 0.5329,
+      "step": 2445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2863777573381346e-05,
+      "loss": 0.506,
+      "step": 2446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2857394384014594e-05,
+      "loss": 0.551,
+      "step": 2447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2851009926638881e-05,
+      "loss": 0.5717,
+      "step": 2448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2844624204087398e-05,
+      "loss": 0.5804,
+      "step": 2449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2838237219193897e-05,
+      "loss": 0.5452,
+      "step": 2450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2831848974792696e-05,
+      "loss": 0.5776,
+      "step": 2451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2825459473718667e-05,
+      "loss": 0.5509,
+      "step": 2452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.281906871880724e-05,
+      "loss": 0.5492,
+      "step": 2453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2812676712894405e-05,
+      "loss": 0.5686,
+      "step": 2454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2806283458816704e-05,
+      "loss": 0.5506,
+      "step": 2455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.279988895941123e-05,
+      "loss": 0.5487,
+      "step": 2456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2793493217515636e-05,
+      "loss": 0.5381,
+      "step": 2457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.278709623596812e-05,
+      "loss": 0.5451,
+      "step": 2458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2780698017607434e-05,
+      "loss": 0.5386,
+      "step": 2459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2774298565272875e-05,
+      "loss": 0.5876,
+      "step": 2460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2767897881804294e-05,
+      "loss": 0.6086,
+      "step": 2461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2761495970042083e-05,
+      "loss": 0.5643,
+      "step": 2462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.275509283282718e-05,
+      "loss": 0.5698,
+      "step": 2463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2748688473001064e-05,
+      "loss": 0.5585,
+      "step": 2464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2742282893405769e-05,
+      "loss": 0.572,
+      "step": 2465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2735876096883851e-05,
+      "loss": 0.6215,
+      "step": 2466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2729468086278418e-05,
+      "loss": 0.5786,
+      "step": 2467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.272305886443312e-05,
+      "loss": 0.5237,
+      "step": 2468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2716648434192136e-05,
+      "loss": 0.5548,
+      "step": 2469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2710236798400182e-05,
+      "loss": 0.5478,
+      "step": 2470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2703823959902517e-05,
+      "loss": 0.5623,
+      "step": 2471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2697409921544926e-05,
+      "loss": 0.552,
+      "step": 2472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2690994686173728e-05,
+      "loss": 0.5517,
+      "step": 2473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2684578256635774e-05,
+      "loss": 0.6286,
+      "step": 2474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2678160635778445e-05,
+      "loss": 0.5813,
+      "step": 2475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2671741826449653e-05,
+      "loss": 0.5552,
+      "step": 2476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.266532183149783e-05,
+      "loss": 0.6074,
+      "step": 2477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2658900653771942e-05,
+      "loss": 0.5796,
+      "step": 2478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2652478296121478e-05,
+      "loss": 0.5519,
+      "step": 2479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2646054761396451e-05,
+      "loss": 0.6034,
+      "step": 2480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.263963005244739e-05,
+      "loss": 0.5743,
+      "step": 2481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2633204172125352e-05,
+      "loss": 0.5211,
+      "step": 2482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2626777123281913e-05,
+      "loss": 0.6168,
+      "step": 2483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2620348908769167e-05,
+      "loss": 0.5907,
+      "step": 2484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.261391953143972e-05,
+      "loss": 0.5835,
+      "step": 2485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2607488994146703e-05,
+      "loss": 0.5686,
+      "step": 2486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.260105729974376e-05,
+      "loss": 0.5312,
+      "step": 2487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2594624451085038e-05,
+      "loss": 0.5483,
+      "step": 2488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.5552,
+      "step": 2489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2581755302419452e-05,
+      "loss": 0.5568,
+      "step": 2490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.257531900812345e-05,
+      "loss": 0.5848,
+      "step": 2491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2568881570993403e-05,
+      "loss": 0.5902,
+      "step": 2492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2562442993886017e-05,
+      "loss": 0.5896,
+      "step": 2493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2556003279658495e-05,
+      "loss": 0.5746,
+      "step": 2494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2549562431168554e-05,
+      "loss": 0.5556,
+      "step": 2495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2543120451274409e-05,
+      "loss": 0.5627,
+      "step": 2496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2536677342834786e-05,
+      "loss": 0.5507,
+      "step": 2497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2530233108708894e-05,
+      "loss": 0.5454,
+      "step": 2498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2523787751756462e-05,
+      "loss": 0.596,
+      "step": 2499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2517341274837702e-05,
+      "loss": 0.6235,
+      "step": 2500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2510893680813334e-05,
+      "loss": 0.5281,
+      "step": 2501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2504444972544562e-05,
+      "loss": 0.6128,
+      "step": 2502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2497995152893099e-05,
+      "loss": 0.6038,
+      "step": 2503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2491544224721137e-05,
+      "loss": 0.5918,
+      "step": 2504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2485092190891367e-05,
+      "loss": 0.534,
+      "step": 2505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2478639054266973e-05,
+      "loss": 0.5519,
+      "step": 2506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2472184817711621e-05,
+      "loss": 0.532,
+      "step": 2507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2465729484089473e-05,
+      "loss": 0.5148,
+      "step": 2508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.245927305626517e-05,
+      "loss": 0.6126,
+      "step": 2509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2452815537103849e-05,
+      "loss": 0.5723,
+      "step": 2510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.244635692947112e-05,
+      "loss": 0.5694,
+      "step": 2511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.243989723623308e-05,
+      "loss": 0.5398,
+      "step": 2512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2433436460256311e-05,
+      "loss": 0.5467,
+      "step": 2513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2426974604407875e-05,
+      "loss": 0.6181,
+      "step": 2514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2420511671555311e-05,
+      "loss": 0.5446,
+      "step": 2515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2414047664566633e-05,
+      "loss": 0.5756,
+      "step": 2516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2407582586310337e-05,
+      "loss": 0.5157,
+      "step": 2517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2401116439655391e-05,
+      "loss": 0.6319,
+      "step": 2518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2394649227471244e-05,
+      "loss": 0.571,
+      "step": 2519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2388180952627806e-05,
+      "loss": 0.5239,
+      "step": 2520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.238171161799547e-05,
+      "loss": 0.6159,
+      "step": 2521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2375241226445089e-05,
+      "loss": 0.6134,
+      "step": 2522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2368769780847993e-05,
+      "loss": 0.5624,
+      "step": 2523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2362297284075977e-05,
+      "loss": 0.5968,
+      "step": 2524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2355823739001307e-05,
+      "loss": 0.5402,
+      "step": 2525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2349349148496698e-05,
+      "loss": 0.5205,
+      "step": 2526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2342873515435354e-05,
+      "loss": 0.5398,
+      "step": 2527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.233639684269092e-05,
+      "loss": 0.5625,
+      "step": 2528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2329919133137513e-05,
+      "loss": 0.5584,
+      "step": 2529
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2323440389649708e-05,
+      "loss": 0.5462,
+      "step": 2530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2316960615102538e-05,
+      "loss": 0.5538,
+      "step": 2531
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2310479812371494e-05,
+      "loss": 0.5977,
+      "step": 2532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2303997984332525e-05,
+      "loss": 0.6114,
+      "step": 2533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2297515133862027e-05,
+      "loss": 0.5195,
+      "step": 2534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2291031263836868e-05,
+      "loss": 0.5954,
+      "step": 2535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2284546377134347e-05,
+      "loss": 0.5375,
+      "step": 2536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2278060476632225e-05,
+      "loss": 0.5614,
+      "step": 2537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2271573565208717e-05,
+      "loss": 0.5797,
+      "step": 2538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2265085645742477e-05,
+      "loss": 0.5641,
+      "step": 2539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2258596721112608e-05,
+      "loss": 0.5609,
+      "step": 2540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2252106794198663e-05,
+      "loss": 0.5405,
+      "step": 2541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2245615867880645e-05,
+      "loss": 0.5637,
+      "step": 2542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2239123945038987e-05,
+      "loss": 0.5965,
+      "step": 2543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2232631028554572e-05,
+      "loss": 0.5587,
+      "step": 2544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2226137121308724e-05,
+      "loss": 0.5974,
+      "step": 2545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2219642226183209e-05,
+      "loss": 0.5785,
+      "step": 2546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.221314634606022e-05,
+      "loss": 0.5924,
+      "step": 2547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2206649483822402e-05,
+      "loss": 0.6168,
+      "step": 2548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2200151642352824e-05,
+      "loss": 0.5633,
+      "step": 2549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2193652824535e-05,
+      "loss": 0.6046,
+      "step": 2550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2187153033252869e-05,
+      "loss": 0.5741,
+      "step": 2551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2180652271390807e-05,
+      "loss": 0.5873,
+      "step": 2552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2174150541833618e-05,
+      "loss": 0.535,
+      "step": 2553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.216764784746653e-05,
+      "loss": 0.5416,
+      "step": 2554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2161144191175214e-05,
+      "loss": 0.5036,
+      "step": 2555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2154639575845751e-05,
+      "loss": 0.5448,
+      "step": 2556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2148134004364664e-05,
+      "loss": 0.5517,
+      "step": 2557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2141627479618886e-05,
+      "loss": 0.6067,
+      "step": 2558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2135120004495778e-05,
+      "loss": 0.5793,
+      "step": 2559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2128611581883128e-05,
+      "loss": 0.5668,
+      "step": 2560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2122102214669138e-05,
+      "loss": 0.5905,
+      "step": 2561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2115591905742427e-05,
+      "loss": 0.5489,
+      "step": 2562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2109080657992045e-05,
+      "loss": 0.5746,
+      "step": 2563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2102568474307446e-05,
+      "loss": 0.588,
+      "step": 2564
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.20960553575785e-05,
+      "loss": 0.5647,
+      "step": 2565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2089541310695496e-05,
+      "loss": 0.5996,
+      "step": 2566
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2083026336549137e-05,
+      "loss": 0.5465,
+      "step": 2567
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2076510438030532e-05,
+      "loss": 0.6194,
+      "step": 2568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2069993618031199e-05,
+      "loss": 0.5641,
+      "step": 2569
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2063475879443076e-05,
+      "loss": 0.5614,
+      "step": 2570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2056957225158497e-05,
+      "loss": 0.56,
+      "step": 2571
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2050437658070207e-05,
+      "loss": 0.5305,
+      "step": 2572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2043917181071354e-05,
+      "loss": 0.5182,
+      "step": 2573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2037395797055493e-05,
+      "loss": 0.568,
+      "step": 2574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2030873508916583e-05,
+      "loss": 0.6195,
+      "step": 2575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2024350319548976e-05,
+      "loss": 0.6027,
+      "step": 2576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2017826231847432e-05,
+      "loss": 0.5614,
+      "step": 2577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2011301248707106e-05,
+      "loss": 0.5234,
+      "step": 2578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.200477537302355e-05,
+      "loss": 0.5919,
+      "step": 2579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1998248607692713e-05,
+      "loss": 0.5659,
+      "step": 2580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1991720955610942e-05,
+      "loss": 0.5843,
+      "step": 2581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.198519241967497e-05,
+      "loss": 0.5403,
+      "step": 2582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1978663002781934e-05,
+      "loss": 0.5187,
+      "step": 2583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1972132707829346e-05,
+      "loss": 0.5883,
+      "step": 2584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1965601537715121e-05,
+      "loss": 0.5882,
+      "step": 2585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1959069495337557e-05,
+      "loss": 0.5507,
+      "step": 2586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.195253658359534e-05,
+      "loss": 0.5844,
+      "step": 2587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1946002805387538e-05,
+      "loss": 0.6182,
+      "step": 2588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1939468163613615e-05,
+      "loss": 0.5965,
+      "step": 2589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.19329326611734e-05,
+      "loss": 0.5839,
+      "step": 2590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1926396300967122e-05,
+      "loss": 0.6237,
+      "step": 2591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1919859085895382e-05,
+      "loss": 0.5701,
+      "step": 2592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1913321018859157e-05,
+      "loss": 0.5412,
+      "step": 2593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1906782102759808e-05,
+      "loss": 0.5768,
+      "step": 2594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1900242340499073e-05,
+      "loss": 0.623,
+      "step": 2595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1893701734979062e-05,
+      "loss": 0.6146,
+      "step": 2596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.188716028910226e-05,
+      "loss": 0.6351,
+      "step": 2597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1880618005771525e-05,
+      "loss": 0.5607,
+      "step": 2598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1874074887890092e-05,
+      "loss": 0.6117,
+      "step": 2599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1867530938361557e-05,
+      "loss": 0.5825,
+      "step": 2600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1860986160089892e-05,
+      "loss": 0.6007,
+      "step": 2601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1854440555979432e-05,
+      "loss": 0.5561,
+      "step": 2602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1847894128934885e-05,
+      "loss": 0.5989,
+      "step": 2603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1841346881861314e-05,
+      "loss": 0.5891,
+      "step": 2604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1834798817664157e-05,
+      "loss": 0.5528,
+      "step": 2605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1828249939249213e-05,
+      "loss": 0.5591,
+      "step": 2606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1821700249522627e-05,
+      "loss": 0.5892,
+      "step": 2607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1815149751390926e-05,
+      "loss": 0.5502,
+      "step": 2608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1808598447760981e-05,
+      "loss": 0.5697,
+      "step": 2609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1802046341540033e-05,
+      "loss": 0.6052,
+      "step": 2610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1795493435635658e-05,
+      "loss": 0.5666,
+      "step": 2611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.178893973295581e-05,
+      "loss": 0.5425,
+      "step": 2612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1782385236408786e-05,
+      "loss": 0.5737,
+      "step": 2613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1775829948903233e-05,
+      "loss": 0.5913,
+      "step": 2614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1769273873348156e-05,
+      "loss": 0.5391,
+      "step": 2615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.17627170126529e-05,
+      "loss": 0.5609,
+      "step": 2616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.175615936972717e-05,
+      "loss": 0.5838,
+      "step": 2617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1749600947481006e-05,
+      "loss": 0.5811,
+      "step": 2618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1743041748824806e-05,
+      "loss": 0.5309,
+      "step": 2619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.5566,
+      "step": 2620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1729921033925582e-05,
+      "loss": 0.5483,
+      "step": 2621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1723359523505059e-05,
+      "loss": 0.5775,
+      "step": 2622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.17167972483195e-05,
+      "loss": 0.5704,
+      "step": 2623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1710234211281006e-05,
+      "loss": 0.5325,
+      "step": 2624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1703670415302018e-05,
+      "loss": 0.5352,
+      "step": 2625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1697105863295308e-05,
+      "loss": 0.5527,
+      "step": 2626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1690540558173998e-05,
+      "loss": 0.5635,
+      "step": 2627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1683974502851523e-05,
+      "loss": 0.5477,
+      "step": 2628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.167740770024167e-05,
+      "loss": 0.5429,
+      "step": 2629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1670840153258546e-05,
+      "loss": 0.508,
+      "step": 2630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1664271864816597e-05,
+      "loss": 0.5544,
+      "step": 2631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.165770283783059e-05,
+      "loss": 0.5394,
+      "step": 2632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1651133075215624e-05,
+      "loss": 0.5583,
+      "step": 2633
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1644562579887129e-05,
+      "loss": 0.5631,
+      "step": 2634
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1637991354760842e-05,
+      "loss": 0.5393,
+      "step": 2635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1631419402752849e-05,
+      "loss": 0.5518,
+      "step": 2636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1624846726779536e-05,
+      "loss": 0.595,
+      "step": 2637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1618273329757632e-05,
+      "loss": 0.5589,
+      "step": 2638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1611699214604162e-05,
+      "loss": 0.6097,
+      "step": 2639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1605124384236492e-05,
+      "loss": 0.5453,
+      "step": 2640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1598548841572292e-05,
+      "loss": 0.5785,
+      "step": 2641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.159197258952955e-05,
+      "loss": 0.567,
+      "step": 2642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.158539563102657e-05,
+      "loss": 0.5632,
+      "step": 2643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1578817968981975e-05,
+      "loss": 0.58,
+      "step": 2644
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1572239606314689e-05,
+      "loss": 0.5582,
+      "step": 2645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1565660545943955e-05,
+      "loss": 0.5857,
+      "step": 2646
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1559080790789325e-05,
+      "loss": 0.5468,
+      "step": 2647
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1552500343770658e-05,
+      "loss": 0.5551,
+      "step": 2648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.154591920780812e-05,
+      "loss": 0.5806,
+      "step": 2649
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1539337385822179e-05,
+      "loss": 0.5863,
+      "step": 2650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1532754880733615e-05,
+      "loss": 0.5732,
+      "step": 2651
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1526171695463507e-05,
+      "loss": 0.5393,
+      "step": 2652
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1519587832933235e-05,
+      "loss": 0.5556,
+      "step": 2653
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.151300329606448e-05,
+      "loss": 0.6014,
+      "step": 2654
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1506418087779225e-05,
+      "loss": 0.5797,
+      "step": 2655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1499832210999746e-05,
+      "loss": 0.5798,
+      "step": 2656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.149324566864862e-05,
+      "loss": 0.5846,
+      "step": 2657
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1486658463648718e-05,
+      "loss": 0.5353,
+      "step": 2658
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1480070598923207e-05,
+      "loss": 0.5702,
+      "step": 2659
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1473482077395542e-05,
+      "loss": 0.5144,
+      "step": 2660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.146689290198947e-05,
+      "loss": 0.5471,
+      "step": 2661
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1460303075629033e-05,
+      "loss": 0.5205,
+      "step": 2662
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.145371260123856e-05,
+      "loss": 0.5715,
+      "step": 2663
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1447121481742669e-05,
+      "loss": 0.5502,
+      "step": 2664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1440529720066253e-05,
+      "loss": 0.5945,
+      "step": 2665
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1433937319134512e-05,
+      "loss": 0.5524,
+      "step": 2666
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1427344281872904e-05,
+      "loss": 0.5328,
+      "step": 2667
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1420750611207188e-05,
+      "loss": 0.6046,
+      "step": 2668
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1414156310063398e-05,
+      "loss": 0.5101,
+      "step": 2669
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1407561381367849e-05,
+      "loss": 0.5337,
+      "step": 2670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.140096582804713e-05,
+      "loss": 0.551,
+      "step": 2671
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1394369653028115e-05,
+      "loss": 0.5264,
+      "step": 2672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1387772859237946e-05,
+      "loss": 0.5039,
+      "step": 2673
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1381175449604045e-05,
+      "loss": 0.6136,
+      "step": 2674
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1374577427054102e-05,
+      "loss": 0.5004,
+      "step": 2675
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1367978794516086e-05,
+      "loss": 0.514,
+      "step": 2676
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1361379554918233e-05,
+      "loss": 0.5503,
+      "step": 2677
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1354779711189043e-05,
+      "loss": 0.5531,
+      "step": 2678
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.13481792662573e-05,
+      "loss": 0.605,
+      "step": 2679
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1341578223052035e-05,
+      "loss": 0.5775,
+      "step": 2680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1334976584502558e-05,
+      "loss": 0.5599,
+      "step": 2681
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1328374353538434e-05,
+      "loss": 0.5657,
+      "step": 2682
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1321771533089506e-05,
+      "loss": 0.5667,
+      "step": 2683
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1315168126085859e-05,
+      "loss": 0.5731,
+      "step": 2684
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1308564135457851e-05,
+      "loss": 0.5206,
+      "step": 2685
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1301959564136093e-05,
+      "loss": 0.5571,
+      "step": 2686
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1295354415051465e-05,
+      "loss": 0.5077,
+      "step": 2687
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1288748691135085e-05,
+      "loss": 0.6086,
+      "step": 2688
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1282142395318343e-05,
+      "loss": 0.548,
+      "step": 2689
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1275535530532869e-05,
+      "loss": 0.5717,
+      "step": 2690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1268928099710562e-05,
+      "loss": 0.556,
+      "step": 2691
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1262320105783548e-05,
+      "loss": 0.5377,
+      "step": 2692
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1255711551684232e-05,
+      "loss": 0.5776,
+      "step": 2693
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1249102440345246e-05,
+      "loss": 0.5853,
+      "step": 2694
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1242492774699472e-05,
+      "loss": 0.6004,
+      "step": 2695
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1235882557680051e-05,
+      "loss": 0.5717,
+      "step": 2696
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1229271792220353e-05,
+      "loss": 0.5101,
+      "step": 2697
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1222660481254004e-05,
+      "loss": 0.5323,
+      "step": 2698
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1216048627714862e-05,
+      "loss": 0.5948,
+      "step": 2699
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.120943623453703e-05,
+      "loss": 0.5132,
+      "step": 2700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1202823304654853e-05,
+      "loss": 0.6111,
+      "step": 2701
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.119620984100291e-05,
+      "loss": 0.5537,
+      "step": 2702
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1189595846516017e-05,
+      "loss": 0.5914,
+      "step": 2703
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.118298132412923e-05,
+      "loss": 0.5444,
+      "step": 2704
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1176366276777837e-05,
+      "loss": 0.545,
+      "step": 2705
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1169750707397356e-05,
+      "loss": 0.5964,
+      "step": 2706
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1163134618923537e-05,
+      "loss": 0.5401,
+      "step": 2707
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1156518014292366e-05,
+      "loss": 0.5591,
+      "step": 2708
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1149900896440055e-05,
+      "loss": 0.5225,
+      "step": 2709
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.114328326830304e-05,
+      "loss": 0.5675,
+      "step": 2710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.113666513281799e-05,
+      "loss": 0.5086,
+      "step": 2711
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1130046492921794e-05,
+      "loss": 0.5228,
+      "step": 2712
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1123427351551565e-05,
+      "loss": 0.5519,
+      "step": 2713
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1116807711644642e-05,
+      "loss": 0.556,
+      "step": 2714
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1110187576138585e-05,
+      "loss": 0.5671,
+      "step": 2715
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1103566947971167e-05,
+      "loss": 0.5237,
+      "step": 2716
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.109694583008039e-05,
+      "loss": 0.5472,
+      "step": 2717
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1090324225404463e-05,
+      "loss": 0.5309,
+      "step": 2718
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1083702136881826e-05,
+      "loss": 0.574,
+      "step": 2719
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1077079567451112e-05,
+      "loss": 0.5287,
+      "step": 2720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1070456520051187e-05,
+      "loss": 0.6042,
+      "step": 2721
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.106383299762112e-05,
+      "loss": 0.5578,
+      "step": 2722
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1057209003100192e-05,
+      "loss": 0.5821,
+      "step": 2723
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1050584539427891e-05,
+      "loss": 0.5303,
+      "step": 2724
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1043959609543922e-05,
+      "loss": 0.572,
+      "step": 2725
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1037334216388185e-05,
+      "loss": 0.5864,
+      "step": 2726
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1030708362900793e-05,
+      "loss": 0.5008,
+      "step": 2727
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1024082052022062e-05,
+      "loss": 0.563,
+      "step": 2728
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1017455286692514e-05,
+      "loss": 0.6061,
+      "step": 2729
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1010828069852865e-05,
+      "loss": 0.571,
+      "step": 2730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1004200404444036e-05,
+      "loss": 0.4859,
+      "step": 2731
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.099757229340715e-05,
+      "loss": 0.6088,
+      "step": 2732
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0990943739683521e-05,
+      "loss": 0.5253,
+      "step": 2733
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0984314746214669e-05,
+      "loss": 0.5462,
+      "step": 2734
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0977685315942292e-05,
+      "loss": 0.5447,
+      "step": 2735
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0971055451808305e-05,
+      "loss": 0.5229,
+      "step": 2736
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0964425156754795e-05,
+      "loss": 0.5591,
+      "step": 2737
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0957794433724051e-05,
+      "loss": 0.5307,
+      "step": 2738
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0951163285658551e-05,
+      "loss": 0.563,
+      "step": 2739
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0944531715500961e-05,
+      "loss": 0.5192,
+      "step": 2740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0937899726194132e-05,
+      "loss": 0.6087,
+      "step": 2741
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0931267320681103e-05,
+      "loss": 0.5506,
+      "step": 2742
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0924634501905095e-05,
+      "loss": 0.5391,
+      "step": 2743
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0918001272809518e-05,
+      "loss": 0.5434,
+      "step": 2744
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0911367636337962e-05,
+      "loss": 0.5288,
+      "step": 2745
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.090473359543419e-05,
+      "loss": 0.5957,
+      "step": 2746
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0898099153042159e-05,
+      "loss": 0.5363,
+      "step": 2747
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0891464312105988e-05,
+      "loss": 0.5249,
+      "step": 2748
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0884829075569986e-05,
+      "loss": 0.5845,
+      "step": 2749
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0878193446378633e-05,
+      "loss": 0.5894,
+      "step": 2750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0871557427476585e-05,
+      "loss": 0.5482,
+      "step": 2751
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0864921021808658e-05,
+      "loss": 0.5489,
+      "step": 2752
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0858284232319865e-05,
+      "loss": 0.5868,
+      "step": 2753
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0851647061955364e-05,
+      "loss": 0.5705,
+      "step": 2754
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0845009513660496e-05,
+      "loss": 0.5882,
+      "step": 2755
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0838371590380765e-05,
+      "loss": 0.5695,
+      "step": 2756
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0831733295061849e-05,
+      "loss": 0.5983,
+      "step": 2757
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0825094630649578e-05,
+      "loss": 0.6024,
+      "step": 2758
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0818455600089953e-05,
+      "loss": 0.6121,
+      "step": 2759
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0811816206329143e-05,
+      "loss": 0.5389,
+      "step": 2760
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0805176452313467e-05,
+      "loss": 0.5966,
+      "step": 2761
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0798536340989412e-05,
+      "loss": 0.5909,
+      "step": 2762
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0791895875303618e-05,
+      "loss": 0.5366,
+      "step": 2763
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0785255058202888e-05,
+      "loss": 0.5278,
+      "step": 2764
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0778613892634178e-05,
+      "loss": 0.5571,
+      "step": 2765
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0771972381544598e-05,
+      "loss": 0.536,
+      "step": 2766
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0765330527881411e-05,
+      "loss": 0.5465,
+      "step": 2767
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0758688334592034e-05,
+      "loss": 0.5163,
+      "step": 2768
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0752045804624031e-05,
+      "loss": 0.5005,
+      "step": 2769
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0745402940925124e-05,
+      "loss": 0.5641,
+      "step": 2770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0738759746443171e-05,
+      "loss": 0.5494,
+      "step": 2771
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0732116224126187e-05,
+      "loss": 0.5485,
+      "step": 2772
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0725472376922327e-05,
+      "loss": 0.5679,
+      "step": 2773
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0718828207779894e-05,
+      "loss": 0.5643,
+      "step": 2774
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.071218371964733e-05,
+      "loss": 0.5651,
+      "step": 2775
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0705538915473214e-05,
+      "loss": 0.5282,
+      "step": 2776
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0698893798206283e-05,
+      "loss": 0.6089,
+      "step": 2777
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.069224837079539e-05,
+      "loss": 0.5631,
+      "step": 2778
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0685602636189549e-05,
+      "loss": 0.6123,
+      "step": 2779
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0678956597337885e-05,
+      "loss": 0.5152,
+      "step": 2780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0672310257189683e-05,
+      "loss": 0.5636,
+      "step": 2781
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0665663618694341e-05,
+      "loss": 0.552,
+      "step": 2782
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0659016684801404e-05,
+      "loss": 0.5918,
+      "step": 2783
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0652369458460537e-05,
+      "loss": 0.5565,
+      "step": 2784
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0645721942621548e-05,
+      "loss": 0.5617,
+      "step": 2785
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.063907414023436e-05,
+      "loss": 0.5282,
+      "step": 2786
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.063242605424903e-05,
+      "loss": 0.5432,
+      "step": 2787
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0625777687615737e-05,
+      "loss": 0.5252,
+      "step": 2788
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0619129043284792e-05,
+      "loss": 0.5262,
+      "step": 2789
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0612480124206625e-05,
+      "loss": 0.5168,
+      "step": 2790
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0605830933331782e-05,
+      "loss": 0.562,
+      "step": 2791
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.059918147361094e-05,
+      "loss": 0.5268,
+      "step": 2792
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.059253174799489e-05,
+      "loss": 0.5791,
+      "step": 2793
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0585881759434537e-05,
+      "loss": 0.5294,
+      "step": 2794
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0579231510880907e-05,
+      "loss": 0.5496,
+      "step": 2795
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0572581005285153e-05,
+      "loss": 0.5652,
+      "step": 2796
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0565930245598513e-05,
+      "loss": 0.516,
+      "step": 2797
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0559279234772366e-05,
+      "loss": 0.559,
+      "step": 2798
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0552627975758188e-05,
+      "loss": 0.5979,
+      "step": 2799
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0545976471507573e-05,
+      "loss": 0.6333,
+      "step": 2800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0539324724972213e-05,
+      "loss": 0.5576,
+      "step": 2801
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0532672739103921e-05,
+      "loss": 0.5492,
+      "step": 2802
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0526020516854605e-05,
+      "loss": 0.5757,
+      "step": 2803
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0519368061176282e-05,
+      "loss": 0.6043,
+      "step": 2804
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0512715375021076e-05,
+      "loss": 0.5358,
+      "step": 2805
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0506062461341206e-05,
+      "loss": 0.5482,
+      "step": 2806
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0499409323089003e-05,
+      "loss": 0.5301,
+      "step": 2807
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0492755963216881e-05,
+      "loss": 0.5514,
+      "step": 2808
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0486102384677368e-05,
+      "loss": 0.5522,
+      "step": 2809
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0479448590423082e-05,
+      "loss": 0.5251,
+      "step": 2810
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.047279458340674e-05,
+      "loss": 0.5477,
+      "step": 2811
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0466140366581147e-05,
+      "loss": 0.5447,
+      "step": 2812
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0459485942899207e-05,
+      "loss": 0.5474,
+      "step": 2813
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0452831315313917e-05,
+      "loss": 0.5461,
+      "step": 2814
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.044617648677836e-05,
+      "loss": 0.5379,
+      "step": 2815
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0439521460245707e-05,
+      "loss": 0.5823,
+      "step": 2816
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.043286623866922e-05,
+      "loss": 0.5513,
+      "step": 2817
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0426210825002255e-05,
+      "loss": 0.5604,
+      "step": 2818
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0419555222198234e-05,
+      "loss": 0.5418,
+      "step": 2819
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0412899433210681e-05,
+      "loss": 0.5544,
+      "step": 2820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0406243460993195e-05,
+      "loss": 0.5983,
+      "step": 2821
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.039958730849946e-05,
+      "loss": 0.557,
+      "step": 2822
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0392930978683228e-05,
+      "loss": 0.5008,
+      "step": 2823
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0386274474498353e-05,
+      "loss": 0.5857,
+      "step": 2824
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.037961779889874e-05,
+      "loss": 0.4507,
+      "step": 2825
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0372960954838389e-05,
+      "loss": 0.5503,
+      "step": 2826
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0366303945271364e-05,
+      "loss": 0.618,
+      "step": 2827
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0359646773151815e-05,
+      "loss": 0.5403,
+      "step": 2828
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0352989441433945e-05,
+      "loss": 0.5769,
+      "step": 2829
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0346331953072044e-05,
+      "loss": 0.5885,
+      "step": 2830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0339674311020469e-05,
+      "loss": 0.5928,
+      "step": 2831
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0333016518233635e-05,
+      "loss": 0.5133,
+      "step": 2832
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0326358577666038e-05,
+      "loss": 0.5329,
+      "step": 2833
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0319700492272229e-05,
+      "loss": 0.537,
+      "step": 2834
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.031304226500683e-05,
+      "loss": 0.5686,
+      "step": 2835
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0306383898824516e-05,
+      "loss": 0.5164,
+      "step": 2836
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0299725396680043e-05,
+      "loss": 0.5376,
+      "step": 2837
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0293066761528205e-05,
+      "loss": 0.5898,
+      "step": 2838
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0286407996323869e-05,
+      "loss": 0.564,
+      "step": 2839
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.027974910402195e-05,
+      "loss": 0.5265,
+      "step": 2840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0273090087577436e-05,
+      "loss": 0.5363,
+      "step": 2841
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0266430949945349e-05,
+      "loss": 0.522,
+      "step": 2842
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0259771694080783e-05,
+      "loss": 0.5338,
+      "step": 2843
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0253112322938868e-05,
+      "loss": 0.5245,
+      "step": 2844
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.02464528394748e-05,
+      "loss": 0.4684,
+      "step": 2845
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.023979324664382e-05,
+      "loss": 0.5682,
+      "step": 2846
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0233133547401211e-05,
+      "loss": 0.532,
+      "step": 2847
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0226473744702308e-05,
+      "loss": 0.5725,
+      "step": 2848
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0219813841502495e-05,
+      "loss": 0.5605,
+      "step": 2849
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0213153840757198e-05,
+      "loss": 0.5796,
+      "step": 2850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0206493745421884e-05,
+      "loss": 0.59,
+      "step": 2851
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0199833558452063e-05,
+      "loss": 0.5422,
+      "step": 2852
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0193173282803293e-05,
+      "loss": 0.5125,
+      "step": 2853
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.018651292143116e-05,
+      "loss": 0.5592,
+      "step": 2854
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.017985247729129e-05,
+      "loss": 0.5599,
+      "step": 2855
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0173191953339358e-05,
+      "loss": 0.5747,
+      "step": 2856
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0166531352531055e-05,
+      "loss": 0.5764,
+      "step": 2857
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0159870677822126e-05,
+      "loss": 0.5445,
+      "step": 2858
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0153209932168328e-05,
+      "loss": 0.5557,
+      "step": 2859
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0146549118525472e-05,
+      "loss": 0.5539,
+      "step": 2860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.013988823984938e-05,
+      "loss": 0.5357,
+      "step": 2861
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0133227299095916e-05,
+      "loss": 0.5871,
+      "step": 2862
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.012656629922096e-05,
+      "loss": 0.5269,
+      "step": 2863
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0119905243180432e-05,
+      "loss": 0.5412,
+      "step": 2864
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0113244133930261e-05,
+      "loss": 0.5314,
+      "step": 2865
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0106582974426416e-05,
+      "loss": 0.5924,
+      "step": 2866
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0099921767624874e-05,
+      "loss": 0.6315,
+      "step": 2867
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.009326051648164e-05,
+      "loss": 0.547,
+      "step": 2868
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0086599223952741e-05,
+      "loss": 0.5173,
+      "step": 2869
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.007993789299422e-05,
+      "loss": 0.5676,
+      "step": 2870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0073276526562135e-05,
+      "loss": 0.5495,
+      "step": 2871
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0066615127612556e-05,
+      "loss": 0.5832,
+      "step": 2872
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0059953699101579e-05,
+      "loss": 0.5907,
+      "step": 2873
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0053292243985305e-05,
+      "loss": 0.4809,
+      "step": 2874
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0046630765219848e-05,
+      "loss": 0.534,
+      "step": 2875
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.003996926576133e-05,
+      "loss": 0.5393,
+      "step": 2876
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0033307748565892e-05,
+      "loss": 0.5629,
+      "step": 2877
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0026646216589668e-05,
+      "loss": 0.5244,
+      "step": 2878
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0019984672788814e-05,
+      "loss": 0.526,
+      "step": 2879
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0013323120119476e-05,
+      "loss": 0.562,
+      "step": 2880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0006661561537821e-05,
+      "loss": 0.608,
+      "step": 2881
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 0.5719,
+      "step": 2882
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.993338438462184e-06,
+      "loss": 0.553,
+      "step": 2883
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.986676879880524e-06,
+      "loss": 0.5661,
+      "step": 2884
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.98001532721119e-06,
+      "loss": 0.6047,
+      "step": 2885
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.973353783410337e-06,
+      "loss": 0.5537,
+      "step": 2886
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.966692251434111e-06,
+      "loss": 0.5576,
+      "step": 2887
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.960030734238672e-06,
+      "loss": 0.5767,
+      "step": 2888
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.953369234780154e-06,
+      "loss": 0.5332,
+      "step": 2889
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.946707756014698e-06,
+      "loss": 0.5284,
+      "step": 2890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.940046300898424e-06,
+      "loss": 0.5324,
+      "step": 2891
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.933384872387445e-06,
+      "loss": 0.5754,
+      "step": 2892
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.926723473437869e-06,
+      "loss": 0.5667,
+      "step": 2893
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.920062107005782e-06,
+      "loss": 0.5464,
+      "step": 2894
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.913400776047259e-06,
+      "loss": 0.5753,
+      "step": 2895
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.906739483518363e-06,
+      "loss": 0.5132,
+      "step": 2896
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.90007823237513e-06,
+      "loss": 0.5266,
+      "step": 2897
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.893417025573589e-06,
+      "loss": 0.5468,
+      "step": 2898
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.886755866069742e-06,
+      "loss": 0.5465,
+      "step": 2899
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.880094756819572e-06,
+      "loss": 0.5638,
+      "step": 2900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.87343370077904e-06,
+      "loss": 0.5361,
+      "step": 2901
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.866772700904087e-06,
+      "loss": 0.5426,
+      "step": 2902
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.860111760150619e-06,
+      "loss": 0.5832,
+      "step": 2903
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.85345088147453e-06,
+      "loss": 0.5583,
+      "step": 2904
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.846790067831674e-06,
+      "loss": 0.5836,
+      "step": 2905
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.840129322177876e-06,
+      "loss": 0.5642,
+      "step": 2906
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.833468647468946e-06,
+      "loss": 0.5961,
+      "step": 2907
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.826808046660647e-06,
+      "loss": 0.525,
+      "step": 2908
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.820147522708709e-06,
+      "loss": 0.5389,
+      "step": 2909
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.813487078568843e-06,
+      "loss": 0.5497,
+      "step": 2910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.806826717196708e-06,
+      "loss": 0.5238,
+      "step": 2911
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.800166441547937e-06,
+      "loss": 0.5654,
+      "step": 2912
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.793506254578117e-06,
+      "loss": 0.5791,
+      "step": 2913
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.786846159242807e-06,
+      "loss": 0.5722,
+      "step": 2914
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.780186158497508e-06,
+      "loss": 0.6052,
+      "step": 2915
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.773526255297695e-06,
+      "loss": 0.5282,
+      "step": 2916
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.766866452598792e-06,
+      "loss": 0.5138,
+      "step": 2917
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.760206753356184e-06,
+      "loss": 0.5491,
+      "step": 2918
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.753547160525203e-06,
+      "loss": 0.5559,
+      "step": 2919
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.746887677061134e-06,
+      "loss": 0.5566,
+      "step": 2920
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.74022830591922e-06,
+      "loss": 0.5207,
+      "step": 2921
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.733569050054656e-06,
+      "loss": 0.5405,
+      "step": 2922
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.726909912422566e-06,
+      "loss": 0.5772,
+      "step": 2923
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.720250895978052e-06,
+      "loss": 0.5861,
+      "step": 2924
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.713592003676136e-06,
+      "loss": 0.5708,
+      "step": 2925
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.706933238471798e-06,
+      "loss": 0.5346,
+      "step": 2926
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.70027460331996e-06,
+      "loss": 0.5613,
+      "step": 2927
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.693616101175485e-06,
+      "loss": 0.558,
+      "step": 2928
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.686957734993173e-06,
+      "loss": 0.5675,
+      "step": 2929
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.680299507727774e-06,
+      "loss": 0.5437,
+      "step": 2930
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.673641422333964e-06,
+      "loss": 0.5819,
+      "step": 2931
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.666983481766366e-06,
+      "loss": 0.5225,
+      "step": 2932
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.660325688979537e-06,
+      "loss": 0.6053,
+      "step": 2933
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.653668046927958e-06,
+      "loss": 0.5641,
+      "step": 2934
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.647010558566058e-06,
+      "loss": 0.5808,
+      "step": 2935
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.64035322684819e-06,
+      "loss": 0.5432,
+      "step": 2936
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.633696054728636e-06,
+      "loss": 0.5247,
+      "step": 2937
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.627039045161614e-06,
+      "loss": 0.5357,
+      "step": 2938
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.620382201101266e-06,
+      "loss": 0.508,
+      "step": 2939
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.61372552550165e-06,
+      "loss": 0.5058,
+      "step": 2940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.607069021316773e-06,
+      "loss": 0.576,
+      "step": 2941
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.600412691500547e-06,
+      "loss": 0.5541,
+      "step": 2942
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.593756539006808e-06,
+      "loss": 0.6048,
+      "step": 2943
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.587100566789322e-06,
+      "loss": 0.5576,
+      "step": 2944
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.580444777801767e-06,
+      "loss": 0.5329,
+      "step": 2945
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.57378917499775e-06,
+      "loss": 0.511,
+      "step": 2946
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.567133761330781e-06,
+      "loss": 0.5207,
+      "step": 2947
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.560478539754294e-06,
+      "loss": 0.5753,
+      "step": 2948
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.553823513221643e-06,
+      "loss": 0.583,
+      "step": 2949
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.547168684686088e-06,
+      "loss": 0.5679,
+      "step": 2950
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.540514057100791e-06,
+      "loss": 0.5488,
+      "step": 2951
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.533859633418856e-06,
+      "loss": 0.483,
+      "step": 2952
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.527205416593265e-06,
+      "loss": 0.5923,
+      "step": 2953
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.52055140957692e-06,
+      "loss": 0.5482,
+      "step": 2954
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.513897615322635e-06,
+      "loss": 0.5839,
+      "step": 2955
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.507244036783124e-06,
+      "loss": 0.5048,
+      "step": 2956
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.500590676911e-06,
+      "loss": 0.5467,
+      "step": 2957
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.493937538658795e-06,
+      "loss": 0.5615,
+      "step": 2958
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.487284624978926e-06,
+      "loss": 0.5501,
+      "step": 2959
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.48063193882372e-06,
+      "loss": 0.6013,
+      "step": 2960
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.4739794831454e-06,
+      "loss": 0.4957,
+      "step": 2961
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.467327260896082e-06,
+      "loss": 0.4981,
+      "step": 2962
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.46067527502779e-06,
+      "loss": 0.5588,
+      "step": 2963
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.454023528492432e-06,
+      "loss": 0.5464,
+      "step": 2964
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.447372024241814e-06,
+      "loss": 0.5525,
+      "step": 2965
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.440720765227637e-06,
+      "loss": 0.5812,
+      "step": 2966
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.434069754401494e-06,
+      "loss": 0.5297,
+      "step": 2967
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.427418994714852e-06,
+      "loss": 0.5726,
+      "step": 2968
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.420768489119095e-06,
+      "loss": 0.522,
+      "step": 2969
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.41411824056547e-06,
+      "loss": 0.5711,
+      "step": 2970
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.407468252005114e-06,
+      "loss": 0.5395,
+      "step": 2971
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.400818526389063e-06,
+      "loss": 0.5534,
+      "step": 2972
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.39416906666822e-06,
+      "loss": 0.512,
+      "step": 2973
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.387519875793376e-06,
+      "loss": 0.5751,
+      "step": 2974
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.38087095671521e-06,
+      "loss": 0.5931,
+      "step": 2975
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.374222312384263e-06,
+      "loss": 0.5934,
+      "step": 2976
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.367573945750974e-06,
+      "loss": 0.5629,
+      "step": 2977
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.360925859765645e-06,
+      "loss": 0.5595,
+      "step": 2978
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.354278057378457e-06,
+      "loss": 0.5601,
+      "step": 2979
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.347630541539465e-06,
+      "loss": 0.5342,
+      "step": 2980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.340983315198601e-06,
+      "loss": 0.6229,
+      "step": 2981
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.334336381305662e-06,
+      "loss": 0.5738,
+      "step": 2982
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.327689742810322e-06,
+      "loss": 0.5895,
+      "step": 2983
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.321043402662116e-06,
+      "loss": 0.548,
+      "step": 2984
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.314397363810455e-06,
+      "loss": 0.5725,
+      "step": 2985
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.307751629204612e-06,
+      "loss": 0.5758,
+      "step": 2986
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.301106201793719e-06,
+      "loss": 0.5293,
+      "step": 2987
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.294461084526787e-06,
+      "loss": 0.5831,
+      "step": 2988
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.287816280352676e-06,
+      "loss": 0.5835,
+      "step": 2989
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.281171792220108e-06,
+      "loss": 0.5471,
+      "step": 2990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.274527623077674e-06,
+      "loss": 0.5314,
+      "step": 2991
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.267883775873816e-06,
+      "loss": 0.5726,
+      "step": 2992
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.26124025355683e-06,
+      "loss": 0.566,
+      "step": 2993
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.25459705907488e-06,
+      "loss": 0.539,
+      "step": 2994
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.247954195375974e-06,
+      "loss": 0.5514,
+      "step": 2995
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.241311665407968e-06,
+      "loss": 0.5712,
+      "step": 2996
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.234669472118594e-06,
+      "loss": 0.5189,
+      "step": 2997
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.228027618455405e-06,
+      "loss": 0.4999,
+      "step": 2998
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.221386107365823e-06,
+      "loss": 0.5681,
+      "step": 2999
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.214744941797115e-06,
+      "loss": 0.573,
+      "step": 3000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.208104124696382e-06,
+      "loss": 0.5048,
+      "step": 3001
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.201463659010592e-06,
+      "loss": 0.5746,
+      "step": 3002
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.194823547686538e-06,
+      "loss": 0.545,
+      "step": 3003
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.188183793670859e-06,
+      "loss": 0.569,
+      "step": 3004
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.181544399910049e-06,
+      "loss": 0.5681,
+      "step": 3005
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.174905369350427e-06,
+      "loss": 0.5059,
+      "step": 3006
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.168266704938155e-06,
+      "loss": 0.5092,
+      "step": 3007
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.161628409619237e-06,
+      "loss": 0.5975,
+      "step": 3008
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.154990486339509e-06,
+      "loss": 0.5526,
+      "step": 3009
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.14835293804464e-06,
+      "loss": 0.5329,
+      "step": 3010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.14171576768014e-06,
+      "loss": 0.5421,
+      "step": 3011
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.13507897819134e-06,
+      "loss": 0.5216,
+      "step": 3012
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.128442572523418e-06,
+      "loss": 0.6285,
+      "step": 3013
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.121806553621369e-06,
+      "loss": 0.5661,
+      "step": 3014
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.115170924430012e-06,
+      "loss": 0.5746,
+      "step": 3015
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.108535687894015e-06,
+      "loss": 0.538,
+      "step": 3016
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.101900846957846e-06,
+      "loss": 0.4999,
+      "step": 3017
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.095266404565812e-06,
+      "loss": 0.5482,
+      "step": 3018
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.088632363662043e-06,
+      "loss": 0.5799,
+      "step": 3019
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.081998727190484e-06,
+      "loss": 0.5482,
+      "step": 3020
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.075365498094906e-06,
+      "loss": 0.5831,
+      "step": 3021
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.0687326793189e-06,
+      "loss": 0.547,
+      "step": 3022
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.062100273805874e-06,
+      "loss": 0.5387,
+      "step": 3023
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.055468284499044e-06,
+      "loss": 0.5614,
+      "step": 3024
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.048836714341452e-06,
+      "loss": 0.5384,
+      "step": 3025
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.042205566275952e-06,
+      "loss": 0.5565,
+      "step": 3026
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.03557484324521e-06,
+      "loss": 0.5625,
+      "step": 3027
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.0289445481917e-06,
+      "loss": 0.5108,
+      "step": 3028
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.02231468405771e-06,
+      "loss": 0.5353,
+      "step": 3029
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.015685253785336e-06,
+      "loss": 0.5485,
+      "step": 3030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.009056260316482e-06,
+      "loss": 0.6264,
+      "step": 3031
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.00242770659285e-06,
+      "loss": 0.5731,
+      "step": 3032
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.995799595555967e-06,
+      "loss": 0.5684,
+      "step": 3033
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.98917193014714e-06,
+      "loss": 0.5486,
+      "step": 3034
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.982544713307488e-06,
+      "loss": 0.5351,
+      "step": 3035
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.97591794797794e-06,
+      "loss": 0.5608,
+      "step": 3036
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.96929163709921e-06,
+      "loss": 0.5353,
+      "step": 3037
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.962665783611819e-06,
+      "loss": 0.5278,
+      "step": 3038
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.956040390456083e-06,
+      "loss": 0.5566,
+      "step": 3039
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.949415460572109e-06,
+      "loss": 0.5395,
+      "step": 3040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.94279099689981e-06,
+      "loss": 0.5717,
+      "step": 3041
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.936167002378884e-06,
+      "loss": 0.5366,
+      "step": 3042
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.929543479948815e-06,
+      "loss": 0.6078,
+      "step": 3043
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.922920432548891e-06,
+      "loss": 0.533,
+      "step": 3044
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.916297863118179e-06,
+      "loss": 0.5878,
+      "step": 3045
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.909675774595537e-06,
+      "loss": 0.5525,
+      "step": 3046
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.903054169919614e-06,
+      "loss": 0.5544,
+      "step": 3047
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.896433052028838e-06,
+      "loss": 0.5421,
+      "step": 3048
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.889812423861418e-06,
+      "loss": 0.5179,
+      "step": 3049
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.883192288355362e-06,
+      "loss": 0.5632,
+      "step": 3050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.87657264844844e-06,
+      "loss": 0.572,
+      "step": 3051
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.86995350707821e-06,
+      "loss": 0.6054,
+      "step": 3052
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.863334867182013e-06,
+      "loss": 0.5226,
+      "step": 3053
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.85671673169696e-06,
+      "loss": 0.5407,
+      "step": 3054
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.850099103559946e-06,
+      "loss": 0.5347,
+      "step": 3055
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.843481985707635e-06,
+      "loss": 0.523,
+      "step": 3056
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.836865381076465e-06,
+      "loss": 0.5575,
+      "step": 3057
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.830249292602649e-06,
+      "loss": 0.5318,
+      "step": 3058
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.823633723222168e-06,
+      "loss": 0.523,
+      "step": 3059
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.817018675870772e-06,
+      "loss": 0.5869,
+      "step": 3060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.810404153483987e-06,
+      "loss": 0.5688,
+      "step": 3061
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.803790158997095e-06,
+      "loss": 0.5457,
+      "step": 3062
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.797176695345149e-06,
+      "loss": 0.5391,
+      "step": 3063
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.790563765462972e-06,
+      "loss": 0.6083,
+      "step": 3064
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.78395137228514e-06,
+      "loss": 0.5341,
+      "step": 3065
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.777339518745997e-06,
+      "loss": 0.5587,
+      "step": 3066
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.770728207779649e-06,
+      "loss": 0.5314,
+      "step": 3067
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.764117442319949e-06,
+      "loss": 0.5195,
+      "step": 3068
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.75750722530053e-06,
+      "loss": 0.5232,
+      "step": 3069
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.750897559654761e-06,
+      "loss": 0.5643,
+      "step": 3070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.744288448315771e-06,
+      "loss": 0.5282,
+      "step": 3071
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.737679894216454e-06,
+      "loss": 0.5633,
+      "step": 3072
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.731071900289445e-06,
+      "loss": 0.6044,
+      "step": 3073
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.724464469467133e-06,
+      "loss": 0.5482,
+      "step": 3074
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.717857604681662e-06,
+      "loss": 0.5566,
+      "step": 3075
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.71125130886492e-06,
+      "loss": 0.5347,
+      "step": 3076
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.704645584948537e-06,
+      "loss": 0.5345,
+      "step": 3077
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.69804043586391e-06,
+      "loss": 0.5177,
+      "step": 3078
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.69143586454215e-06,
+      "loss": 0.5354,
+      "step": 3079
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.684831873914146e-06,
+      "loss": 0.5292,
+      "step": 3080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.6782284669105e-06,
+      "loss": 0.5327,
+      "step": 3081
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.671625646461566e-06,
+      "loss": 0.5625,
+      "step": 3082
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.665023415497445e-06,
+      "loss": 0.593,
+      "step": 3083
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.658421776947968e-06,
+      "loss": 0.5057,
+      "step": 3084
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.651820733742701e-06,
+      "loss": 0.5905,
+      "step": 3085
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.645220288810958e-06,
+      "loss": 0.5331,
+      "step": 3086
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.638620445081772e-06,
+      "loss": 0.5372,
+      "step": 3087
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.632021205483917e-06,
+      "loss": 0.5267,
+      "step": 3088
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.625422572945902e-06,
+      "loss": 0.5943,
+      "step": 3089
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.618824550395962e-06,
+      "loss": 0.5535,
+      "step": 3090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.612227140762057e-06,
+      "loss": 0.5475,
+      "step": 3091
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.605630346971888e-06,
+      "loss": 0.5873,
+      "step": 3092
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.59903417195287e-06,
+      "loss": 0.517,
+      "step": 3093
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.592438618632153e-06,
+      "loss": 0.5177,
+      "step": 3094
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.585843689936605e-06,
+      "loss": 0.5619,
+      "step": 3095
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.579249388792813e-06,
+      "loss": 0.5744,
+      "step": 3096
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.572655718127101e-06,
+      "loss": 0.5029,
+      "step": 3097
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.566062680865495e-06,
+      "loss": 0.5508,
+      "step": 3098
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.559470279933748e-06,
+      "loss": 0.4779,
+      "step": 3099
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.552878518257335e-06,
+      "loss": 0.5095,
+      "step": 3100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.546287398761442e-06,
+      "loss": 0.5043,
+      "step": 3101
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.539696924370968e-06,
+      "loss": 0.5254,
+      "step": 3102
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.533107098010532e-06,
+      "loss": 0.5716,
+      "step": 3103
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.526517922604465e-06,
+      "loss": 0.5335,
+      "step": 3104
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.519929401076796e-06,
+      "loss": 0.5447,
+      "step": 3105
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.513341536351285e-06,
+      "loss": 0.5051,
+      "step": 3106
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.506754331351382e-06,
+      "loss": 0.552,
+      "step": 3107
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.500167789000257e-06,
+      "loss": 0.5326,
+      "step": 3108
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.49358191222078e-06,
+      "loss": 0.5505,
+      "step": 3109
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.486996703935522e-06,
+      "loss": 0.547,
+      "step": 3110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.480412167066766e-06,
+      "loss": 0.5526,
+      "step": 3111
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.473828304536498e-06,
+      "loss": 0.6276,
+      "step": 3112
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.467245119266385e-06,
+      "loss": 0.5302,
+      "step": 3113
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.460662614177825e-06,
+      "loss": 0.5937,
+      "step": 3114
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.454080792191885e-06,
+      "loss": 0.5463,
+      "step": 3115
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.447499656229344e-06,
+      "loss": 0.561,
+      "step": 3116
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.440919209210678e-06,
+      "loss": 0.5887,
+      "step": 3117
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.434339454056048e-06,
+      "loss": 0.5223,
+      "step": 3118
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.427760393685313e-06,
+      "loss": 0.613,
+      "step": 3119
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.421182031018029e-06,
+      "loss": 0.4813,
+      "step": 3120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.41460436897343e-06,
+      "loss": 0.5659,
+      "step": 3121
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.408027410470452e-06,
+      "loss": 0.5772,
+      "step": 3122
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.401451158427714e-06,
+      "loss": 0.5363,
+      "step": 3123
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.394875615763508e-06,
+      "loss": 0.5744,
+      "step": 3124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.38830078539584e-06,
+      "loss": 0.5722,
+      "step": 3125
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.381726670242374e-06,
+      "loss": 0.4922,
+      "step": 3126
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.375153273220464e-06,
+      "loss": 0.5389,
+      "step": 3127
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.368580597247156e-06,
+      "loss": 0.5397,
+      "step": 3128
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.362008645239161e-06,
+      "loss": 0.5156,
+      "step": 3129
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.355437420112876e-06,
+      "loss": 0.5684,
+      "step": 3130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.348866924784378e-06,
+      "loss": 0.4869,
+      "step": 3131
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.342297162169414e-06,
+      "loss": 0.5091,
+      "step": 3132
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.335728135183404e-06,
+      "loss": 0.5754,
+      "step": 3133
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.329159846741457e-06,
+      "loss": 0.5424,
+      "step": 3134
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.322592299758333e-06,
+      "loss": 0.5242,
+      "step": 3135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.316025497148482e-06,
+      "loss": 0.5541,
+      "step": 3136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.309459441826009e-06,
+      "loss": 0.528,
+      "step": 3137
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.302894136704694e-06,
+      "loss": 0.5279,
+      "step": 3138
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.296329584697987e-06,
+      "loss": 0.5042,
+      "step": 3139
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.289765788719e-06,
+      "loss": 0.5822,
+      "step": 3140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.283202751680501e-06,
+      "loss": 0.5227,
+      "step": 3141
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.276640476494943e-06,
+      "loss": 0.5564,
+      "step": 3142
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.270078966074421e-06,
+      "loss": 0.5501,
+      "step": 3143
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.5381,
+      "step": 3144
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.256958251175197e-06,
+      "loss": 0.5176,
+      "step": 3145
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.250399052518998e-06,
+      "loss": 0.5775,
+      "step": 3146
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.243840630272834e-06,
+      "loss": 0.5676,
+      "step": 3147
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.237282987347103e-06,
+      "loss": 0.5469,
+      "step": 3148
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.230726126651846e-06,
+      "loss": 0.5437,
+      "step": 3149
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.224170051096769e-06,
+      "loss": 0.474,
+      "step": 3150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.217614763591217e-06,
+      "loss": 0.5307,
+      "step": 3151
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.211060267044191e-06,
+      "loss": 0.5981,
+      "step": 3152
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.204506564364343e-06,
+      "loss": 0.5706,
+      "step": 3153
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.197953658459972e-06,
+      "loss": 0.5636,
+      "step": 3154
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.191401552239019e-06,
+      "loss": 0.549,
+      "step": 3155
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.184850248609076e-06,
+      "loss": 0.5837,
+      "step": 3156
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.178299750477378e-06,
+      "loss": 0.5556,
+      "step": 3157
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.171750060750792e-06,
+      "loss": 0.5295,
+      "step": 3158
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.165201182335846e-06,
+      "loss": 0.5778,
+      "step": 3159
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.158653118138685e-06,
+      "loss": 0.5442,
+      "step": 3160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.152105871065118e-06,
+      "loss": 0.6025,
+      "step": 3161
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.145559444020571e-06,
+      "loss": 0.5361,
+      "step": 3162
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.13901383991011e-06,
+      "loss": 0.556,
+      "step": 3163
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.132469061638444e-06,
+      "loss": 0.5149,
+      "step": 3164
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.125925112109911e-06,
+      "loss": 0.5268,
+      "step": 3165
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.119381994228475e-06,
+      "loss": 0.5674,
+      "step": 3166
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.112839710897742e-06,
+      "loss": 0.5203,
+      "step": 3167
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.106298265020943e-06,
+      "loss": 0.5028,
+      "step": 3168
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.099757659500929e-06,
+      "loss": 0.5476,
+      "step": 3169
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.093217897240196e-06,
+      "loss": 0.5608,
+      "step": 3170
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.086678981140848e-06,
+      "loss": 0.5487,
+      "step": 3171
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.080140914104621e-06,
+      "loss": 0.5257,
+      "step": 3172
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.073603699032881e-06,
+      "loss": 0.5629,
+      "step": 3173
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.0670673388266e-06,
+      "loss": 0.5095,
+      "step": 3174
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.060531836386389e-06,
+      "loss": 0.5327,
+      "step": 3175
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.053997194612465e-06,
+      "loss": 0.5519,
+      "step": 3176
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.047463416404662e-06,
+      "loss": 0.5386,
+      "step": 3177
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.040930504662447e-06,
+      "loss": 0.5471,
+      "step": 3178
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.034398462284882e-06,
+      "loss": 0.5832,
+      "step": 3179
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.027867292170657e-06,
+      "loss": 0.5632,
+      "step": 3180
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.02133699721807e-06,
+      "loss": 0.5499,
+      "step": 3181
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.014807580325031e-06,
+      "loss": 0.5141,
+      "step": 3182
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.00827904438906e-06,
+      "loss": 0.5351,
+      "step": 3183
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.001751392307288e-06,
+      "loss": 0.5129,
+      "step": 3184
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.995224626976457e-06,
+      "loss": 0.5528,
+      "step": 3185
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.988698751292896e-06,
+      "loss": 0.5445,
+      "step": 3186
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.982173768152573e-06,
+      "loss": 0.5615,
+      "step": 3187
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.975649680451024e-06,
+      "loss": 0.574,
+      "step": 3188
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.96912649108342e-06,
+      "loss": 0.5048,
+      "step": 3189
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.962604202944508e-06,
+      "loss": 0.5738,
+      "step": 3190
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.956082818928647e-06,
+      "loss": 0.5353,
+      "step": 3191
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.949562341929795e-06,
+      "loss": 0.6193,
+      "step": 3192
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.943042774841505e-06,
+      "loss": 0.5876,
+      "step": 3193
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.936524120556924e-06,
+      "loss": 0.5234,
+      "step": 3194
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.930006381968803e-06,
+      "loss": 0.5603,
+      "step": 3195
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.923489561969473e-06,
+      "loss": 0.5684,
+      "step": 3196
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.916973663450865e-06,
+      "loss": 0.5281,
+      "step": 3197
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.910458689304506e-06,
+      "loss": 0.5256,
+      "step": 3198
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.903944642421505e-06,
+      "loss": 0.5284,
+      "step": 3199
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.897431525692557e-06,
+      "loss": 0.5312,
+      "step": 3200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.890919342007956e-06,
+      "loss": 0.5314,
+      "step": 3201
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.884408094257573e-06,
+      "loss": 0.5652,
+      "step": 3202
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.877897785330866e-06,
+      "loss": 0.506,
+      "step": 3203
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.871388418116877e-06,
+      "loss": 0.5507,
+      "step": 3204
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.864879995504224e-06,
+      "loss": 0.5239,
+      "step": 3205
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.85837252038112e-06,
+      "loss": 0.5485,
+      "step": 3206
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.85186599563534e-06,
+      "loss": 0.586,
+      "step": 3207
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.84536042415425e-06,
+      "loss": 0.5474,
+      "step": 3208
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.838855808824791e-06,
+      "loss": 0.5763,
+      "step": 3209
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.832352152533473e-06,
+      "loss": 0.6288,
+      "step": 3210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.825849458166385e-06,
+      "loss": 0.5864,
+      "step": 3211
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.819347728609195e-06,
+      "loss": 0.5804,
+      "step": 3212
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.812846966747134e-06,
+      "loss": 0.5491,
+      "step": 3213
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.806347175465001e-06,
+      "loss": 0.5152,
+      "step": 3214
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.799848357647179e-06,
+      "loss": 0.5984,
+      "step": 3215
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.793350516177603e-06,
+      "loss": 0.5746,
+      "step": 3216
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.786853653939784e-06,
+      "loss": 0.5261,
+      "step": 3217
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.780357773816798e-06,
+      "loss": 0.5777,
+      "step": 3218
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.773862878691278e-06,
+      "loss": 0.5638,
+      "step": 3219
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.76736897144543e-06,
+      "loss": 0.5614,
+      "step": 3220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.760876054961018e-06,
+      "loss": 0.5422,
+      "step": 3221
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.754384132119357e-06,
+      "loss": 0.5269,
+      "step": 3222
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.747893205801339e-06,
+      "loss": 0.5286,
+      "step": 3223
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.741403278887397e-06,
+      "loss": 0.55,
+      "step": 3224
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.734914354257528e-06,
+      "loss": 0.5407,
+      "step": 3225
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.728426434791287e-06,
+      "loss": 0.5669,
+      "step": 3226
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.721939523367777e-06,
+      "loss": 0.5464,
+      "step": 3227
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.715453622865655e-06,
+      "loss": 0.5323,
+      "step": 3228
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.708968736163135e-06,
+      "loss": 0.5247,
+      "step": 3229
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.702484866137971e-06,
+      "loss": 0.5343,
+      "step": 3230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.696002015667478e-06,
+      "loss": 0.5498,
+      "step": 3231
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.68952018762851e-06,
+      "loss": 0.5409,
+      "step": 3232
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.683039384897464e-06,
+      "loss": 0.5538,
+      "step": 3233
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.676559610350297e-06,
+      "loss": 0.5247,
+      "step": 3234
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.67008086686249e-06,
+      "loss": 0.4978,
+      "step": 3235
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.663603157309082e-06,
+      "loss": 0.5358,
+      "step": 3236
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.657126484564648e-06,
+      "loss": 0.5296,
+      "step": 3237
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.650650851503304e-06,
+      "loss": 0.559,
+      "step": 3238
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.644176260998696e-06,
+      "loss": 0.5411,
+      "step": 3239
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.637702715924026e-06,
+      "loss": 0.5488,
+      "step": 3240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.631230219152012e-06,
+      "loss": 0.5076,
+      "step": 3241
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.624758773554915e-06,
+      "loss": 0.5467,
+      "step": 3242
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.618288382004535e-06,
+      "loss": 0.5503,
+      "step": 3243
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.6118190473721955e-06,
+      "loss": 0.5331,
+      "step": 3244
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.605350772528759e-06,
+      "loss": 0.5298,
+      "step": 3245
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.59888356034461e-06,
+      "loss": 0.5361,
+      "step": 3246
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.592417413689664e-06,
+      "loss": 0.571,
+      "step": 3247
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.5859523354333695e-06,
+      "loss": 0.5506,
+      "step": 3248
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.579488328444694e-06,
+      "loss": 0.5776,
+      "step": 3249
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.573025395592125e-06,
+      "loss": 0.5966,
+      "step": 3250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.566563539743691e-06,
+      "loss": 0.5579,
+      "step": 3251
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.560102763766925e-06,
+      "loss": 0.5207,
+      "step": 3252
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.5536430705288846e-06,
+      "loss": 0.512,
+      "step": 3253
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.547184462896156e-06,
+      "loss": 0.5387,
+      "step": 3254
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.54072694373483e-06,
+      "loss": 0.5213,
+      "step": 3255
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.53427051591053e-06,
+      "loss": 0.5508,
+      "step": 3256
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.527815182288381e-06,
+      "loss": 0.542,
+      "step": 3257
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.52136094573303e-06,
+      "loss": 0.5728,
+      "step": 3258
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.514907809108637e-06,
+      "loss": 0.5416,
+      "step": 3259
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.508455775278868e-06,
+      "loss": 0.5513,
+      "step": 3260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.502004847106904e-06,
+      "loss": 0.5776,
+      "step": 3261
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.495555027455439e-06,
+      "loss": 0.5436,
+      "step": 3262
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.48910631918667e-06,
+      "loss": 0.5617,
+      "step": 3263
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.4826587251623e-06,
+      "loss": 0.5112,
+      "step": 3264
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.4762122482435415e-06,
+      "loss": 0.5571,
+      "step": 3265
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.46976689129111e-06,
+      "loss": 0.5668,
+      "step": 3266
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.463322657165218e-06,
+      "loss": 0.5225,
+      "step": 3267
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.456879548725593e-06,
+      "loss": 0.5743,
+      "step": 3268
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.450437568831446e-06,
+      "loss": 0.5267,
+      "step": 3269
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.443996720341509e-06,
+      "loss": 0.5912,
+      "step": 3270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.437557006113987e-06,
+      "loss": 0.5315,
+      "step": 3271
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.431118429006596e-06,
+      "loss": 0.5356,
+      "step": 3272
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.424680991876552e-06,
+      "loss": 0.5776,
+      "step": 3273
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.418244697580552e-06,
+      "loss": 0.5195,
+      "step": 3274
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 0.5492,
+      "step": 3275
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.405375548914963e-06,
+      "loss": 0.5588,
+      "step": 3276
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.398942700256246e-06,
+      "loss": 0.5022,
+      "step": 3277
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.3925110058532976e-06,
+      "loss": 0.5454,
+      "step": 3278
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.386080468560283e-06,
+      "loss": 0.5489,
+      "step": 3279
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.379651091230838e-06,
+      "loss": 0.5605,
+      "step": 3280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.3732228767180895e-06,
+      "loss": 0.5515,
+      "step": 3281
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.366795827874651e-06,
+      "loss": 0.5408,
+      "step": 3282
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.360369947552612e-06,
+      "loss": 0.5172,
+      "step": 3283
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.353945238603552e-06,
+      "loss": 0.4916,
+      "step": 3284
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.347521703878525e-06,
+      "loss": 0.5265,
+      "step": 3285
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.3410993462280575e-06,
+      "loss": 0.5348,
+      "step": 3286
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.334678168502174e-06,
+      "loss": 0.6026,
+      "step": 3287
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.328258173550353e-06,
+      "loss": 0.5579,
+      "step": 3288
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.321839364221557e-06,
+      "loss": 0.5731,
+      "step": 3289
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.31542174336423e-06,
+      "loss": 0.5616,
+      "step": 3290
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.309005313826276e-06,
+      "loss": 0.5501,
+      "step": 3291
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.302590078455076e-06,
+      "loss": 0.5384,
+      "step": 3292
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.296176040097484e-06,
+      "loss": 0.5159,
+      "step": 3293
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.289763201599822e-06,
+      "loss": 0.6154,
+      "step": 3294
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.283351565807866e-06,
+      "loss": 0.5781,
+      "step": 3295
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.276941135566884e-06,
+      "loss": 0.5069,
+      "step": 3296
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.270531913721582e-06,
+      "loss": 0.5359,
+      "step": 3297
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.264123903116154e-06,
+      "loss": 0.6208,
+      "step": 3298
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.257717106594235e-06,
+      "loss": 0.5147,
+      "step": 3299
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.251311526998934e-06,
+      "loss": 0.5749,
+      "step": 3300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.244907167172821e-06,
+      "loss": 0.5576,
+      "step": 3301
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.2385040299579175e-06,
+      "loss": 0.6322,
+      "step": 3302
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.232102118195705e-06,
+      "loss": 0.5366,
+      "step": 3303
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.225701434727127e-06,
+      "loss": 0.5259,
+      "step": 3304
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.21930198239257e-06,
+      "loss": 0.5353,
+      "step": 3305
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.212903764031884e-06,
+      "loss": 0.5609,
+      "step": 3306
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.206506782484368e-06,
+      "loss": 0.5581,
+      "step": 3307
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.2001110405887744e-06,
+      "loss": 0.5563,
+      "step": 3308
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.1937165411833e-06,
+      "loss": 0.601,
+      "step": 3309
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.187323287105598e-06,
+      "loss": 0.4961,
+      "step": 3310
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.18093128119276e-06,
+      "loss": 0.5599,
+      "step": 3311
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.174540526281335e-06,
+      "loss": 0.5578,
+      "step": 3312
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.168151025207309e-06,
+      "loss": 0.5243,
+      "step": 3313
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.161762780806103e-06,
+      "loss": 0.5613,
+      "step": 3314
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.1553757959126065e-06,
+      "loss": 0.5469,
+      "step": 3315
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.148990073361123e-06,
+      "loss": 0.5272,
+      "step": 3316
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.142605615985407e-06,
+      "loss": 0.5881,
+      "step": 3317
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.136222426618656e-06,
+      "loss": 0.5538,
+      "step": 3318
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.129840508093499e-06,
+      "loss": 0.535,
+      "step": 3319
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.123459863241997e-06,
+      "loss": 0.6048,
+      "step": 3320
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.117080494895658e-06,
+      "loss": 0.5619,
+      "step": 3321
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.110702405885414e-06,
+      "loss": 0.5256,
+      "step": 3322
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.104325599041629e-06,
+      "loss": 0.5349,
+      "step": 3323
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.097950077194105e-06,
+      "loss": 0.5479,
+      "step": 3324
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.091575843172065e-06,
+      "loss": 0.572,
+      "step": 3325
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.085202899804169e-06,
+      "loss": 0.5634,
+      "step": 3326
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.078831249918499e-06,
+      "loss": 0.5456,
+      "step": 3327
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.072460896342561e-06,
+      "loss": 0.5166,
+      "step": 3328
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.066091841903295e-06,
+      "loss": 0.5226,
+      "step": 3329
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.059724089427058e-06,
+      "loss": 0.5682,
+      "step": 3330
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.05335764173962e-06,
+      "loss": 0.5558,
+      "step": 3331
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.046992501666196e-06,
+      "loss": 0.6049,
+      "step": 3332
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.040628672031398e-06,
+      "loss": 0.5694,
+      "step": 3333
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.0342661556592626e-06,
+      "loss": 0.5289,
+      "step": 3334
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.027904955373252e-06,
+      "loss": 0.5371,
+      "step": 3335
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.021545073996237e-06,
+      "loss": 0.5658,
+      "step": 3336
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.015186514350502e-06,
+      "loss": 0.5144,
+      "step": 3337
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.008829279257752e-06,
+      "loss": 0.5874,
+      "step": 3338
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.002473371539094e-06,
+      "loss": 0.5507,
+      "step": 3339
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.996118794015054e-06,
+      "loss": 0.5912,
+      "step": 3340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.989765549505569e-06,
+      "loss": 0.5842,
+      "step": 3341
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.983413640829972e-06,
+      "loss": 0.4581,
+      "step": 3342
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.977063070807019e-06,
+      "loss": 0.6109,
+      "step": 3343
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.970713842254861e-06,
+      "loss": 0.4832,
+      "step": 3344
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.96436595799106e-06,
+      "loss": 0.5648,
+      "step": 3345
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.9580194208325765e-06,
+      "loss": 0.5632,
+      "step": 3346
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.951674233595781e-06,
+      "loss": 0.5551,
+      "step": 3347
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.945330399096428e-06,
+      "loss": 0.5273,
+      "step": 3348
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.938987920149699e-06,
+      "loss": 0.5376,
+      "step": 3349
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.932646799570144e-06,
+      "loss": 0.5436,
+      "step": 3350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.926307040171736e-06,
+      "loss": 0.5333,
+      "step": 3351
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.919968644767825e-06,
+      "loss": 0.4761,
+      "step": 3352
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.913631616171164e-06,
+      "loss": 0.5282,
+      "step": 3353
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.9072959571939015e-06,
+      "loss": 0.5793,
+      "step": 3354
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.900961670647573e-06,
+      "loss": 0.5394,
+      "step": 3355
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.8946287593431075e-06,
+      "loss": 0.5394,
+      "step": 3356
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.888297226090825e-06,
+      "loss": 0.5558,
+      "step": 3357
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.881967073700434e-06,
+      "loss": 0.5666,
+      "step": 3358
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.875638304981021e-06,
+      "loss": 0.5438,
+      "step": 3359
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.869310922741076e-06,
+      "loss": 0.5553,
+      "step": 3360
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.8629849297884564e-06,
+      "loss": 0.5641,
+      "step": 3361
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.856660328930413e-06,
+      "loss": 0.4977,
+      "step": 3362
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.85033712297358e-06,
+      "loss": 0.5505,
+      "step": 3363
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.844015314723962e-06,
+      "loss": 0.5549,
+      "step": 3364
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.837694906986956e-06,
+      "loss": 0.5171,
+      "step": 3365
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.831375902567332e-06,
+      "loss": 0.4993,
+      "step": 3366
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.82505830426923e-06,
+      "loss": 0.538,
+      "step": 3367
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.818742114896185e-06,
+      "loss": 0.5435,
+      "step": 3368
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.8124273372510866e-06,
+      "loss": 0.585,
+      "step": 3369
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.806113974136204e-06,
+      "loss": 0.581,
+      "step": 3370
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.799802028353187e-06,
+      "loss": 0.5455,
+      "step": 3371
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.793491502703049e-06,
+      "loss": 0.522,
+      "step": 3372
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.787182399986172e-06,
+      "loss": 0.5576,
+      "step": 3373
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.780874723002311e-06,
+      "loss": 0.5442,
+      "step": 3374
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.77456847455059e-06,
+      "loss": 0.5262,
+      "step": 3375
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.768263657429485e-06,
+      "loss": 0.5145,
+      "step": 3376
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.761960274436861e-06,
+      "loss": 0.5268,
+      "step": 3377
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.755658328369919e-06,
+      "loss": 0.5569,
+      "step": 3378
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.749357822025249e-06,
+      "loss": 0.5556,
+      "step": 3379
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.743058758198781e-06,
+      "loss": 0.5413,
+      "step": 3380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.736761139685815e-06,
+      "loss": 0.5543,
+      "step": 3381
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.73046496928101e-06,
+      "loss": 0.5831,
+      "step": 3382
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.724170249778379e-06,
+      "loss": 0.5453,
+      "step": 3383
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.71787698397129e-06,
+      "loss": 0.5612,
+      "step": 3384
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.71158517465247e-06,
+      "loss": 0.5595,
+      "step": 3385
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.705294824614004e-06,
+      "loss": 0.4944,
+      "step": 3386
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.699005936647311e-06,
+      "loss": 0.5787,
+      "step": 3387
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.692718513543184e-06,
+      "loss": 0.5181,
+      "step": 3388
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.6864325580917495e-06,
+      "loss": 0.5498,
+      "step": 3389
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.680148073082491e-06,
+      "loss": 0.5528,
+      "step": 3390
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.673865061304236e-06,
+      "loss": 0.5838,
+      "step": 3391
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.66758352554516e-06,
+      "loss": 0.5635,
+      "step": 3392
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.661303468592785e-06,
+      "loss": 0.5064,
+      "step": 3393
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.655024893233975e-06,
+      "loss": 0.5672,
+      "step": 3394
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.648747802254929e-06,
+      "loss": 0.5407,
+      "step": 3395
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.642472198441203e-06,
+      "loss": 0.5513,
+      "step": 3396
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.636198084577681e-06,
+      "loss": 0.5437,
+      "step": 3397
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.629925463448586e-06,
+      "loss": 0.5402,
+      "step": 3398
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.623654337837487e-06,
+      "loss": 0.5692,
+      "step": 3399
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.617384710527282e-06,
+      "loss": 0.517,
+      "step": 3400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.6111165843002055e-06,
+      "loss": 0.5459,
+      "step": 3401
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.60484996193783e-06,
+      "loss": 0.5627,
+      "step": 3402
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.598584846221057e-06,
+      "loss": 0.5512,
+      "step": 3403
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.592321239930112e-06,
+      "loss": 0.5359,
+      "step": 3404
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.58605914584457e-06,
+      "loss": 0.5075,
+      "step": 3405
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.5513,
+      "step": 3406
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.5735395054045695e-06,
+      "loss": 0.5186,
+      "step": 3407
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.56728196460588e-06,
+      "loss": 0.5908,
+      "step": 3408
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.561025947124119e-06,
+      "loss": 0.5373,
+      "step": 3409
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.554771455735483e-06,
+      "loss": 0.5873,
+      "step": 3410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.548518493215492e-06,
+      "loss": 0.503,
+      "step": 3411
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.542267062338976e-06,
+      "loss": 0.5259,
+      "step": 3412
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.5360171658801105e-06,
+      "loss": 0.6032,
+      "step": 3413
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.5297688066123644e-06,
+      "loss": 0.5594,
+      "step": 3414
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.5235219873085365e-06,
+      "loss": 0.5113,
+      "step": 3415
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.5172767107407446e-06,
+      "loss": 0.5254,
+      "step": 3416
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.511032979680416e-06,
+      "loss": 0.5522,
+      "step": 3417
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.50479079689829e-06,
+      "loss": 0.5066,
+      "step": 3418
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.498550165164432e-06,
+      "loss": 0.5055,
+      "step": 3419
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.492311087248202e-06,
+      "loss": 0.5792,
+      "step": 3420
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.486073565918283e-06,
+      "loss": 0.5081,
+      "step": 3421
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.479837603942665e-06,
+      "loss": 0.5678,
+      "step": 3422
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.473603204088633e-06,
+      "loss": 0.5673,
+      "step": 3423
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.467370369122804e-06,
+      "loss": 0.5194,
+      "step": 3424
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.461139101811078e-06,
+      "loss": 0.5278,
+      "step": 3425
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.454909404918665e-06,
+      "loss": 0.555,
+      "step": 3426
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.448681281210089e-06,
+      "loss": 0.5256,
+      "step": 3427
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.442454733449162e-06,
+      "loss": 0.5275,
+      "step": 3428
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.436229764399001e-06,
+      "loss": 0.5253,
+      "step": 3429
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.430006376822031e-06,
+      "loss": 0.577,
+      "step": 3430
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.423784573479955e-06,
+      "loss": 0.5855,
+      "step": 3431
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.417564357133801e-06,
+      "loss": 0.562,
+      "step": 3432
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.4113457305438676e-06,
+      "loss": 0.5465,
+      "step": 3433
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.4051286964697575e-06,
+      "loss": 0.5971,
+      "step": 3434
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.398913257670374e-06,
+      "loss": 0.5043,
+      "step": 3435
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.3926994169038996e-06,
+      "loss": 0.5249,
+      "step": 3436
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.3864871769278135e-06,
+      "loss": 0.5285,
+      "step": 3437
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.380276540498888e-06,
+      "loss": 0.5461,
+      "step": 3438
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.3740675103731805e-06,
+      "loss": 0.5186,
+      "step": 3439
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.367860089306028e-06,
+      "loss": 0.5509,
+      "step": 3440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.3616542800520745e-06,
+      "loss": 0.557,
+      "step": 3441
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.355450085365224e-06,
+      "loss": 0.5746,
+      "step": 3442
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.349247507998677e-06,
+      "loss": 0.5701,
+      "step": 3443
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.343046550704917e-06,
+      "loss": 0.5377,
+      "step": 3444
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.336847216235704e-06,
+      "loss": 0.5533,
+      "step": 3445
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.330649507342084e-06,
+      "loss": 0.504,
+      "step": 3446
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.324453426774373e-06,
+      "loss": 0.5794,
+      "step": 3447
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.318258977282168e-06,
+      "loss": 0.5491,
+      "step": 3448
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.31206616161435e-06,
+      "loss": 0.5331,
+      "step": 3449
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.305874982519064e-06,
+      "loss": 0.5692,
+      "step": 3450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.299685442743728e-06,
+      "loss": 0.491,
+      "step": 3451
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.293497545035046e-06,
+      "loss": 0.5753,
+      "step": 3452
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.287311292138978e-06,
+      "loss": 0.4834,
+      "step": 3453
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.281126686800762e-06,
+      "loss": 0.5386,
+      "step": 3454
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.274943731764909e-06,
+      "loss": 0.5624,
+      "step": 3455
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.2687624297751885e-06,
+      "loss": 0.5518,
+      "step": 3456
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.262582783574633e-06,
+      "loss": 0.5188,
+      "step": 3457
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.256404795905562e-06,
+      "loss": 0.5286,
+      "step": 3458
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.250228469509531e-06,
+      "loss": 0.6031,
+      "step": 3459
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.244053807127383e-06,
+      "loss": 0.5586,
+      "step": 3460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.237880811499204e-06,
+      "loss": 0.5363,
+      "step": 3461
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.231709485364348e-06,
+      "loss": 0.6148,
+      "step": 3462
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.2255398314614315e-06,
+      "loss": 0.5694,
+      "step": 3463
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.219371852528322e-06,
+      "loss": 0.5545,
+      "step": 3464
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.213205551302147e-06,
+      "loss": 0.5615,
+      "step": 3465
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.207040930519292e-06,
+      "loss": 0.5742,
+      "step": 3466
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.200877992915395e-06,
+      "loss": 0.4962,
+      "step": 3467
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.194716741225339e-06,
+      "loss": 0.5753,
+      "step": 3468
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.1885571781832766e-06,
+      "loss": 0.5709,
+      "step": 3469
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.182399306522593e-06,
+      "loss": 0.5105,
+      "step": 3470
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.176243128975929e-06,
+      "loss": 0.5492,
+      "step": 3471
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.170088648275183e-06,
+      "loss": 0.5597,
+      "step": 3472
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.163935867151485e-06,
+      "loss": 0.5331,
+      "step": 3473
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.157784788335223e-06,
+      "loss": 0.5595,
+      "step": 3474
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.151635414556024e-06,
+      "loss": 0.509,
+      "step": 3475
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.145487748542753e-06,
+      "loss": 0.5203,
+      "step": 3476
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.139341793023534e-06,
+      "loss": 0.5891,
+      "step": 3477
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.133197550725714e-06,
+      "loss": 0.5403,
+      "step": 3478
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.127055024375887e-06,
+      "loss": 0.489,
+      "step": 3479
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.120914216699887e-06,
+      "loss": 0.5803,
+      "step": 3480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.114775130422785e-06,
+      "loss": 0.5229,
+      "step": 3481
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.108637768268884e-06,
+      "loss": 0.5732,
+      "step": 3482
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.102502132961728e-06,
+      "loss": 0.5709,
+      "step": 3483
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.096368227224093e-06,
+      "loss": 0.5471,
+      "step": 3484
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.090236053777976e-06,
+      "loss": 0.5187,
+      "step": 3485
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.0841056153446285e-06,
+      "loss": 0.5792,
+      "step": 3486
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.077976914644508e-06,
+      "loss": 0.5227,
+      "step": 3487
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.071849954397321e-06,
+      "loss": 0.5583,
+      "step": 3488
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.065724737321985e-06,
+      "loss": 0.5633,
+      "step": 3489
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.059601266136653e-06,
+      "loss": 0.5366,
+      "step": 3490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.053479543558704e-06,
+      "loss": 0.5079,
+      "step": 3491
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.047359572304736e-06,
+      "loss": 0.5713,
+      "step": 3492
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.0412413550905724e-06,
+      "loss": 0.5655,
+      "step": 3493
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.035124894631263e-06,
+      "loss": 0.5061,
+      "step": 3494
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.029010193641067e-06,
+      "loss": 0.5258,
+      "step": 3495
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.0228972548334685e-06,
+      "loss": 0.5467,
+      "step": 3496
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.0167860809211755e-06,
+      "loss": 0.5517,
+      "step": 3497
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.010676674616105e-06,
+      "loss": 0.5647,
+      "step": 3498
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.0045690386293885e-06,
+      "loss": 0.5384,
+      "step": 3499
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.998463175671382e-06,
+      "loss": 0.5474,
+      "step": 3500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.9923590884516425e-06,
+      "loss": 0.5468,
+      "step": 3501
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.986256779678948e-06,
+      "loss": 0.5716,
+      "step": 3502
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.9801562520612854e-06,
+      "loss": 0.539,
+      "step": 3503
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.9740575083058404e-06,
+      "loss": 0.5727,
+      "step": 3504
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.967960551119029e-06,
+      "loss": 0.5509,
+      "step": 3505
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.961865383206454e-06,
+      "loss": 0.5318,
+      "step": 3506
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.955772007272932e-06,
+      "loss": 0.5381,
+      "step": 3507
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.949680426022487e-06,
+      "loss": 0.5563,
+      "step": 3508
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.943590642158343e-06,
+      "loss": 0.5936,
+      "step": 3509
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.937502658382925e-06,
+      "loss": 0.5322,
+      "step": 3510
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.931416477397867e-06,
+      "loss": 0.4984,
+      "step": 3511
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.925332101903994e-06,
+      "loss": 0.4646,
+      "step": 3512
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.919249534601328e-06,
+      "loss": 0.5386,
+      "step": 3513
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.913168778189105e-06,
+      "loss": 0.5625,
+      "step": 3514
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.9070898353657355e-06,
+      "loss": 0.5476,
+      "step": 3515
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.901012708828844e-06,
+      "loss": 0.548,
+      "step": 3516
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.894937401275238e-06,
+      "loss": 0.565,
+      "step": 3517
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.888863915400916e-06,
+      "loss": 0.5353,
+      "step": 3518
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.882792253901081e-06,
+      "loss": 0.5024,
+      "step": 3519
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.876722419470115e-06,
+      "loss": 0.5724,
+      "step": 3520
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.870654414801587e-06,
+      "loss": 0.5355,
+      "step": 3521
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.8645882425882695e-06,
+      "loss": 0.5738,
+      "step": 3522
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.858523905522104e-06,
+      "loss": 0.5608,
+      "step": 3523
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.8524614062942265e-06,
+      "loss": 0.5611,
+      "step": 3524
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.84640074759496e-06,
+      "loss": 0.5188,
+      "step": 3525
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.840341932113802e-06,
+      "loss": 0.5416,
+      "step": 3526
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.834284962539443e-06,
+      "loss": 0.4912,
+      "step": 3527
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.828229841559744e-06,
+      "loss": 0.5335,
+      "step": 3528
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.822176571861753e-06,
+      "loss": 0.5644,
+      "step": 3529
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.816125156131691e-06,
+      "loss": 0.5505,
+      "step": 3530
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.8100755970549605e-06,
+      "loss": 0.6215,
+      "step": 3531
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.8040278973161336e-06,
+      "loss": 0.5081,
+      "step": 3532
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.79798205959897e-06,
+      "loss": 0.5421,
+      "step": 3533
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.7919380865863965e-06,
+      "loss": 0.5228,
+      "step": 3534
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.785895980960498e-06,
+      "loss": 0.5763,
+      "step": 3535
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.779855745402556e-06,
+      "loss": 0.5279,
+      "step": 3536
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.773817382593008e-06,
+      "loss": 0.5585,
+      "step": 3537
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.767780895211459e-06,
+      "loss": 0.527,
+      "step": 3538
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.761746285936687e-06,
+      "loss": 0.5654,
+      "step": 3539
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.755713557446632e-06,
+      "loss": 0.5177,
+      "step": 3540
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.749682712418408e-06,
+      "loss": 0.5849,
+      "step": 3541
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.743653753528284e-06,
+      "loss": 0.5543,
+      "step": 3542
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.737626683451697e-06,
+      "loss": 0.5552,
+      "step": 3543
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.7316015048632425e-06,
+      "loss": 0.5288,
+      "step": 3544
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.725578220436677e-06,
+      "loss": 0.5693,
+      "step": 3545
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.71955683284492e-06,
+      "loss": 0.5524,
+      "step": 3546
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.713537344760047e-06,
+      "loss": 0.5564,
+      "step": 3547
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.707519758853288e-06,
+      "loss": 0.5316,
+      "step": 3548
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.70150407779503e-06,
+      "loss": 0.5327,
+      "step": 3549
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.695490304254825e-06,
+      "loss": 0.5918,
+      "step": 3550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.689478440901366e-06,
+      "loss": 0.5345,
+      "step": 3551
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.683468490402493e-06,
+      "loss": 0.5963,
+      "step": 3552
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.677460455425215e-06,
+      "loss": 0.5771,
+      "step": 3553
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.671454338635678e-06,
+      "loss": 0.5403,
+      "step": 3554
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.665450142699189e-06,
+      "loss": 0.5,
+      "step": 3555
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.6594478702801856e-06,
+      "loss": 0.5554,
+      "step": 3556
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.653447524042259e-06,
+      "loss": 0.5128,
+      "step": 3557
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.647449106648155e-06,
+      "loss": 0.6135,
+      "step": 3558
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.641452620759753e-06,
+      "loss": 0.5277,
+      "step": 3559
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.6354580690380774e-06,
+      "loss": 0.5884,
+      "step": 3560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.6294654541432946e-06,
+      "loss": 0.5152,
+      "step": 3561
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.623474778734712e-06,
+      "loss": 0.5476,
+      "step": 3562
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.617486045470772e-06,
+      "loss": 0.5515,
+      "step": 3563
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.61149925700907e-06,
+      "loss": 0.5536,
+      "step": 3564
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.6055144160063194e-06,
+      "loss": 0.5793,
+      "step": 3565
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.599531525118373e-06,
+      "loss": 0.5224,
+      "step": 3566
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.5935505870002335e-06,
+      "loss": 0.5349,
+      "step": 3567
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.5875716043060216e-06,
+      "loss": 0.5577,
+      "step": 3568
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.581594579688996e-06,
+      "loss": 0.547,
+      "step": 3569
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.575619515801543e-06,
+      "loss": 0.5457,
+      "step": 3570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.56964641529518e-06,
+      "loss": 0.5552,
+      "step": 3571
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.5636752808205664e-06,
+      "loss": 0.5715,
+      "step": 3572
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.557706115027461e-06,
+      "loss": 0.4998,
+      "step": 3573
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.5517389205647705e-06,
+      "loss": 0.5621,
+      "step": 3574
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.545773700080528e-06,
+      "loss": 0.6026,
+      "step": 3575
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.539810456221877e-06,
+      "loss": 0.5376,
+      "step": 3576
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.533849191635093e-06,
+      "loss": 0.5316,
+      "step": 3577
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.527889908965571e-06,
+      "loss": 0.5561,
+      "step": 3578
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.5219326108578275e-06,
+      "loss": 0.5572,
+      "step": 3579
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.515977299955493e-06,
+      "loss": 0.5617,
+      "step": 3580
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.51002397890133e-06,
+      "loss": 0.5011,
+      "step": 3581
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.504072650337195e-06,
+      "loss": 0.5691,
+      "step": 3582
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.498123316904085e-06,
+      "loss": 0.5173,
+      "step": 3583
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.492175981242097e-06,
+      "loss": 0.5526,
+      "step": 3584
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.486230645990444e-06,
+      "loss": 0.5745,
+      "step": 3585
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.480287313787453e-06,
+      "loss": 0.5214,
+      "step": 3586
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.474345987270562e-06,
+      "loss": 0.5657,
+      "step": 3587
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.468406669076311e-06,
+      "loss": 0.5125,
+      "step": 3588
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.4624693618403676e-06,
+      "loss": 0.5122,
+      "step": 3589
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.456534068197493e-06,
+      "loss": 0.5262,
+      "step": 3590
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.450600790781548e-06,
+      "loss": 0.5403,
+      "step": 3591
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.444669532225516e-06,
+      "loss": 0.5147,
+      "step": 3592
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.438740295161475e-06,
+      "loss": 0.5151,
+      "step": 3593
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.4328130822206094e-06,
+      "loss": 0.5555,
+      "step": 3594
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.426887896033198e-06,
+      "loss": 0.5684,
+      "step": 3595
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.420964739228627e-06,
+      "loss": 0.5158,
+      "step": 3596
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.415043614435385e-06,
+      "loss": 0.5512,
+      "step": 3597
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.409124524281054e-06,
+      "loss": 0.5312,
+      "step": 3598
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.403207471392304e-06,
+      "loss": 0.5636,
+      "step": 3599
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.397292458394923e-06,
+      "loss": 0.5185,
+      "step": 3600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.3913794879137745e-06,
+      "loss": 0.5022,
+      "step": 3601
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.3854685625728234e-06,
+      "loss": 0.5087,
+      "step": 3602
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.379559684995126e-06,
+      "loss": 0.5693,
+      "step": 3603
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.373652857802832e-06,
+      "loss": 0.5502,
+      "step": 3604
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.3677480836171706e-06,
+      "loss": 0.5462,
+      "step": 3605
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.36184536505848e-06,
+      "loss": 0.5201,
+      "step": 3606
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.355944704746174e-06,
+      "loss": 0.562,
+      "step": 3607
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.35004610529874e-06,
+      "loss": 0.5312,
+      "step": 3608
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.344149569333781e-06,
+      "loss": 0.5414,
+      "step": 3609
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.338255099467957e-06,
+      "loss": 0.5585,
+      "step": 3610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.332362698317027e-06,
+      "loss": 0.6203,
+      "step": 3611
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.3264723684958254e-06,
+      "loss": 0.4409,
+      "step": 3612
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.320584112618264e-06,
+      "loss": 0.5527,
+      "step": 3613
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.314697933297349e-06,
+      "loss": 0.558,
+      "step": 3614
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.308813833145155e-06,
+      "loss": 0.5282,
+      "step": 3615
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.30293181477282e-06,
+      "loss": 0.5671,
+      "step": 3616
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.297051880790587e-06,
+      "loss": 0.5737,
+      "step": 3617
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.291174033807755e-06,
+      "loss": 0.5169,
+      "step": 3618
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.2852982764327e-06,
+      "loss": 0.5013,
+      "step": 3619
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.279424611272873e-06,
+      "loss": 0.5293,
+      "step": 3620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.273553040934791e-06,
+      "loss": 0.5411,
+      "step": 3621
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.267683568024056e-06,
+      "loss": 0.5713,
+      "step": 3622
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.261816195145322e-06,
+      "loss": 0.5343,
+      "step": 3623
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.25595092490232e-06,
+      "loss": 0.492,
+      "step": 3624
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.2500877598978466e-06,
+      "loss": 0.561,
+      "step": 3625
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.244226702733761e-06,
+      "loss": 0.5189,
+      "step": 3626
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.238367756010991e-06,
+      "loss": 0.5488,
+      "step": 3627
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.232510922329535e-06,
+      "loss": 0.537,
+      "step": 3628
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.226656204288432e-06,
+      "loss": 0.5583,
+      "step": 3629
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.220803604485798e-06,
+      "loss": 0.5897,
+      "step": 3630
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.2149531255188155e-06,
+      "loss": 0.5342,
+      "step": 3631
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.209104769983711e-06,
+      "loss": 0.5729,
+      "step": 3632
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.203258540475773e-06,
+      "loss": 0.5243,
+      "step": 3633
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.197414439589352e-06,
+      "loss": 0.5158,
+      "step": 3634
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.191572469917843e-06,
+      "loss": 0.5171,
+      "step": 3635
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.185732634053715e-06,
+      "loss": 0.557,
+      "step": 3636
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.179894934588467e-06,
+      "loss": 0.5338,
+      "step": 3637
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.1740593741126574e-06,
+      "loss": 0.5421,
+      "step": 3638
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.168225955215909e-06,
+      "loss": 0.5295,
+      "step": 3639
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.162394680486877e-06,
+      "loss": 0.5509,
+      "step": 3640
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.156565552513273e-06,
+      "loss": 0.5394,
+      "step": 3641
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.150738573881855e-06,
+      "loss": 0.5417,
+      "step": 3642
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.144913747178426e-06,
+      "loss": 0.5592,
+      "step": 3643
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.139091074987829e-06,
+      "loss": 0.5698,
+      "step": 3644
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.133270559893974e-06,
+      "loss": 0.5499,
+      "step": 3645
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.127452204479779e-06,
+      "loss": 0.4849,
+      "step": 3646
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.121636011327225e-06,
+      "loss": 0.5911,
+      "step": 3647
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.115821983017334e-06,
+      "loss": 0.5362,
+      "step": 3648
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.1100101221301605e-06,
+      "loss": 0.5244,
+      "step": 3649
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.104200431244802e-06,
+      "loss": 0.5556,
+      "step": 3650
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.098392912939385e-06,
+      "loss": 0.5483,
+      "step": 3651
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.092587569791079e-06,
+      "loss": 0.5734,
+      "step": 3652
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.086784404376092e-06,
+      "loss": 0.52,
+      "step": 3653
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.080983419269661e-06,
+      "loss": 0.5382,
+      "step": 3654
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.07518461704604e-06,
+      "loss": 0.51,
+      "step": 3655
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.069388000278546e-06,
+      "loss": 0.4593,
+      "step": 3656
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.063593571539502e-06,
+      "loss": 0.5553,
+      "step": 3657
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.057801333400269e-06,
+      "loss": 0.5555,
+      "step": 3658
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.052011288431233e-06,
+      "loss": 0.5216,
+      "step": 3659
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.046223439201807e-06,
+      "loss": 0.5217,
+      "step": 3660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.04043778828043e-06,
+      "loss": 0.5137,
+      "step": 3661
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.034654338234574e-06,
+      "loss": 0.5442,
+      "step": 3662
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.028873091630713e-06,
+      "loss": 0.5258,
+      "step": 3663
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.0230940510343675e-06,
+      "loss": 0.5775,
+      "step": 3664
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.017317219010063e-06,
+      "loss": 0.5655,
+      "step": 3665
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.011542598121352e-06,
+      "loss": 0.5265,
+      "step": 3666
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.005770190930803e-06,
+      "loss": 0.5374,
+      "step": 3667
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.5479,
+      "step": 3668
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.99423202788955e-06,
+      "loss": 0.5268,
+      "step": 3669
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.988466277159073e-06,
+      "loss": 0.5373,
+      "step": 3670
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.982702750367205e-06,
+      "loss": 0.5107,
+      "step": 3671
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.976941450071578e-06,
+      "loss": 0.618,
+      "step": 3672
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.971182378828863e-06,
+      "loss": 0.5255,
+      "step": 3673
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.965425539194726e-06,
+      "loss": 0.5498,
+      "step": 3674
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.959670933723844e-06,
+      "loss": 0.5462,
+      "step": 3675
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.953918564969905e-06,
+      "loss": 0.5704,
+      "step": 3676
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.948168435485601e-06,
+      "loss": 0.5191,
+      "step": 3677
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.942420547822637e-06,
+      "loss": 0.4967,
+      "step": 3678
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.936674904531723e-06,
+      "loss": 0.5169,
+      "step": 3679
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.930931508162554e-06,
+      "loss": 0.5296,
+      "step": 3680
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.9251903612638575e-06,
+      "loss": 0.5558,
+      "step": 3681
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.9194514663833425e-06,
+      "loss": 0.521,
+      "step": 3682
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.913714826067725e-06,
+      "loss": 0.5503,
+      "step": 3683
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.907980442862719e-06,
+      "loss": 0.5836,
+      "step": 3684
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.9022483193130395e-06,
+      "loss": 0.5485,
+      "step": 3685
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.896518457962393e-06,
+      "loss": 0.5407,
+      "step": 3686
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.890790861353492e-06,
+      "loss": 0.5454,
+      "step": 3687
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.8850655320280384e-06,
+      "loss": 0.5639,
+      "step": 3688
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.879342472526717e-06,
+      "loss": 0.5694,
+      "step": 3689
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.873621685389229e-06,
+      "loss": 0.5298,
+      "step": 3690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.867903173154241e-06,
+      "loss": 0.5066,
+      "step": 3691
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.8621869383594415e-06,
+      "loss": 0.5176,
+      "step": 3692
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.856472983541475e-06,
+      "loss": 0.5007,
+      "step": 3693
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.8507613112359895e-06,
+      "loss": 0.5818,
+      "step": 3694
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.845051923977628e-06,
+      "loss": 0.5446,
+      "step": 3695
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.8393448243000084e-06,
+      "loss": 0.6199,
+      "step": 3696
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.833640014735736e-06,
+      "loss": 0.5778,
+      "step": 3697
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.8279374978164e-06,
+      "loss": 0.5297,
+      "step": 3698
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.822237276072571e-06,
+      "loss": 0.4898,
+      "step": 3699
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.816539352033806e-06,
+      "loss": 0.5649,
+      "step": 3700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.810843728228636e-06,
+      "loss": 0.5371,
+      "step": 3701
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.805150407184572e-06,
+      "loss": 0.5959,
+      "step": 3702
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.799459391428113e-06,
+      "loss": 0.5701,
+      "step": 3703
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.793770683484722e-06,
+      "loss": 0.574,
+      "step": 3704
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.788084285878845e-06,
+      "loss": 0.5501,
+      "step": 3705
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7824002011339e-06,
+      "loss": 0.499,
+      "step": 3706
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.776718431772281e-06,
+      "loss": 0.5131,
+      "step": 3707
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.771038980315345e-06,
+      "loss": 0.5333,
+      "step": 3708
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.765361849283445e-06,
+      "loss": 0.4896,
+      "step": 3709
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.759687041195874e-06,
+      "loss": 0.5946,
+      "step": 3710
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.754014558570909e-06,
+      "loss": 0.556,
+      "step": 3711
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7483444039257996e-06,
+      "loss": 0.5594,
+      "step": 3712
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.742676579776755e-06,
+      "loss": 0.6024,
+      "step": 3713
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.737011088638951e-06,
+      "loss": 0.5514,
+      "step": 3714
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.731347933026527e-06,
+      "loss": 0.5387,
+      "step": 3715
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7256871154525895e-06,
+      "loss": 0.5797,
+      "step": 3716
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.720028638429208e-06,
+      "loss": 0.5438,
+      "step": 3717
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.714372504467415e-06,
+      "loss": 0.5195,
+      "step": 3718
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.708718716077185e-06,
+      "loss": 0.5094,
+      "step": 3719
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.703067275767481e-06,
+      "loss": 0.501,
+      "step": 3720
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.697418186046203e-06,
+      "loss": 0.596,
+      "step": 3721
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6917714494202145e-06,
+      "loss": 0.5294,
+      "step": 3722
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.686127068395334e-06,
+      "loss": 0.5432,
+      "step": 3723
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.680485045476336e-06,
+      "loss": 0.5555,
+      "step": 3724
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.674845383166944e-06,
+      "loss": 0.571,
+      "step": 3725
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6692080839698475e-06,
+      "loss": 0.5458,
+      "step": 3726
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.663573150386666e-06,
+      "loss": 0.5638,
+      "step": 3727
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.657940584917983e-06,
+      "loss": 0.5726,
+      "step": 3728
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.652310390063333e-06,
+      "loss": 0.5441,
+      "step": 3729
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.646682568321192e-06,
+      "loss": 0.5737,
+      "step": 3730
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6410571221889845e-06,
+      "loss": 0.5004,
+      "step": 3731
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.635434054163082e-06,
+      "loss": 0.5577,
+      "step": 3732
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.629813366738797e-06,
+      "loss": 0.5508,
+      "step": 3733
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.624195062410394e-06,
+      "loss": 0.5539,
+      "step": 3734
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.618579143671077e-06,
+      "loss": 0.5545,
+      "step": 3735
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.612965613012975e-06,
+      "loss": 0.5193,
+      "step": 3736
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.607354472927186e-06,
+      "loss": 0.5704,
+      "step": 3737
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.601745725903725e-06,
+      "loss": 0.4864,
+      "step": 3738
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.596139374431554e-06,
+      "loss": 0.521,
+      "step": 3739
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.590535420998571e-06,
+      "loss": 0.5346,
+      "step": 3740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.584933868091609e-06,
+      "loss": 0.5334,
+      "step": 3741
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.579334718196431e-06,
+      "loss": 0.5317,
+      "step": 3742
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.57373797379775e-06,
+      "loss": 0.5767,
+      "step": 3743
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.568143637379186e-06,
+      "loss": 0.5634,
+      "step": 3744
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.562551711423317e-06,
+      "loss": 0.5738,
+      "step": 3745
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.556962198411631e-06,
+      "loss": 0.5526,
+      "step": 3746
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.551375100824557e-06,
+      "loss": 0.6053,
+      "step": 3747
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5457904211414465e-06,
+      "loss": 0.5782,
+      "step": 3748
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.540208161840578e-06,
+      "loss": 0.5535,
+      "step": 3749
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.534628325399157e-06,
+      "loss": 0.5591,
+      "step": 3750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.529050914293322e-06,
+      "loss": 0.6124,
+      "step": 3751
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.523475930998125e-06,
+      "loss": 0.5292,
+      "step": 3752
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.517903377987532e-06,
+      "loss": 0.5429,
+      "step": 3753
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.512333257734456e-06,
+      "loss": 0.5471,
+      "step": 3754
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5067655727107104e-06,
+      "loss": 0.5678,
+      "step": 3755
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.501200325387034e-06,
+      "loss": 0.5436,
+      "step": 3756
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.4956375182330846e-06,
+      "loss": 0.5655,
+      "step": 3757
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.490077153717432e-06,
+      "loss": 0.5396,
+      "step": 3758
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.484519234307571e-06,
+      "loss": 0.5275,
+      "step": 3759
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.478963762469907e-06,
+      "loss": 0.5995,
+      "step": 3760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.4734107406697555e-06,
+      "loss": 0.5876,
+      "step": 3761
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.467860171371351e-06,
+      "loss": 0.5616,
+      "step": 3762
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.462312057037834e-06,
+      "loss": 0.5536,
+      "step": 3763
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.456766400131261e-06,
+      "loss": 0.528,
+      "step": 3764
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.451223203112592e-06,
+      "loss": 0.5692,
+      "step": 3765
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.445682468441701e-06,
+      "loss": 0.5129,
+      "step": 3766
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.440144198577363e-06,
+      "loss": 0.5357,
+      "step": 3767
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.43460839597727e-06,
+      "loss": 0.5202,
+      "step": 3768
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.429075063098013e-06,
+      "loss": 0.4979,
+      "step": 3769
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.423544202395073e-06,
+      "loss": 0.5361,
+      "step": 3770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.418015816322862e-06,
+      "loss": 0.5409,
+      "step": 3771
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.412489907334669e-06,
+      "loss": 0.5146,
+      "step": 3772
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.406966477882707e-06,
+      "loss": 0.5431,
+      "step": 3773
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.401445530418063e-06,
+      "loss": 0.5025,
+      "step": 3774
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.395927067390734e-06,
+      "loss": 0.5949,
+      "step": 3775
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.390411091249626e-06,
+      "loss": 0.5913,
+      "step": 3776
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.384897604442525e-06,
+      "loss": 0.5262,
+      "step": 3777
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.3793866094161185e-06,
+      "loss": 0.4927,
+      "step": 3778
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.37387810861599e-06,
+      "loss": 0.5314,
+      "step": 3779
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.368372104486611e-06,
+      "loss": 0.5094,
+      "step": 3780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.362868599471351e-06,
+      "loss": 0.5678,
+      "step": 3781
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.357367596012468e-06,
+      "loss": 0.5714,
+      "step": 3782
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.351869096551108e-06,
+      "loss": 0.5199,
+      "step": 3783
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.346373103527303e-06,
+      "loss": 0.5484,
+      "step": 3784
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.340879619379988e-06,
+      "loss": 0.5548,
+      "step": 3785
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.335388646546969e-06,
+      "loss": 0.5473,
+      "step": 3786
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.3299001874649395e-06,
+      "loss": 0.5833,
+      "step": 3787
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.324414244569484e-06,
+      "loss": 0.5316,
+      "step": 3788
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.318930820295062e-06,
+      "loss": 0.6065,
+      "step": 3789
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.31344991707503e-06,
+      "loss": 0.5233,
+      "step": 3790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.307971537341606e-06,
+      "loss": 0.5468,
+      "step": 3791
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.3024956835258995e-06,
+      "loss": 0.5636,
+      "step": 3792
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.297022358057903e-06,
+      "loss": 0.543,
+      "step": 3793
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2915515633664776e-06,
+      "loss": 0.5334,
+      "step": 3794
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2860833018793665e-06,
+      "loss": 0.5687,
+      "step": 3795
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.28061757602319e-06,
+      "loss": 0.4931,
+      "step": 3796
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2751543882234315e-06,
+      "loss": 0.5431,
+      "step": 3797
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.26969374090447e-06,
+      "loss": 0.5729,
+      "step": 3798
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.264235636489542e-06,
+      "loss": 0.5775,
+      "step": 3799
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2587800774007485e-06,
+      "loss": 0.5321,
+      "step": 3800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.25332706605908e-06,
+      "loss": 0.5395,
+      "step": 3801
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.247876604884384e-06,
+      "loss": 0.538,
+      "step": 3802
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.242428696295381e-06,
+      "loss": 0.5349,
+      "step": 3803
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.236983342709654e-06,
+      "loss": 0.518,
+      "step": 3804
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2315405465436575e-06,
+      "loss": 0.5277,
+      "step": 3805
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.226100310212705e-06,
+      "loss": 0.5447,
+      "step": 3806
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.220662636130989e-06,
+      "loss": 0.549,
+      "step": 3807
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2152275267115416e-06,
+      "loss": 0.5602,
+      "step": 3808
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.20979498436627e-06,
+      "loss": 0.5156,
+      "step": 3809
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.204365011505949e-06,
+      "loss": 0.5311,
+      "step": 3810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.198937610540202e-06,
+      "loss": 0.5234,
+      "step": 3811
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.193512783877513e-06,
+      "loss": 0.5406,
+      "step": 3812
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.188090533925225e-06,
+      "loss": 0.5759,
+      "step": 3813
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.182670863089536e-06,
+      "loss": 0.5477,
+      "step": 3814
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.177253773775508e-06,
+      "loss": 0.5272,
+      "step": 3815
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.17183926838705e-06,
+      "loss": 0.5277,
+      "step": 3816
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1664273493269145e-06,
+      "loss": 0.586,
+      "step": 3817
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.161018018996728e-06,
+      "loss": 0.5725,
+      "step": 3818
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.155611279796951e-06,
+      "loss": 0.5647,
+      "step": 3819
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.150207134126902e-06,
+      "loss": 0.5389,
+      "step": 3820
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.144805584384748e-06,
+      "loss": 0.5124,
+      "step": 3821
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1394066329675e-06,
+      "loss": 0.5322,
+      "step": 3822
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.134010282271016e-06,
+      "loss": 0.5947,
+      "step": 3823
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.128616534690011e-06,
+      "loss": 0.5857,
+      "step": 3824
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.123225392618031e-06,
+      "loss": 0.5614,
+      "step": 3825
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.117836858447472e-06,
+      "loss": 0.5604,
+      "step": 3826
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.112450934569572e-06,
+      "loss": 0.5375,
+      "step": 3827
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.10706762337441e-06,
+      "loss": 0.4751,
+      "step": 3828
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.101686927250906e-06,
+      "loss": 0.527,
+      "step": 3829
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.096308848586821e-06,
+      "loss": 0.5378,
+      "step": 3830
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.090933389768748e-06,
+      "loss": 0.5464,
+      "step": 3831
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.08556055318213e-06,
+      "loss": 0.5478,
+      "step": 3832
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.080190341211239e-06,
+      "loss": 0.5514,
+      "step": 3833
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.074822756239171e-06,
+      "loss": 0.5214,
+      "step": 3834
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.069457800647879e-06,
+      "loss": 0.5421,
+      "step": 3835
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.064095476818133e-06,
+      "loss": 0.5049,
+      "step": 3836
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.058735787129539e-06,
+      "loss": 0.5499,
+      "step": 3837
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.053378733960538e-06,
+      "loss": 0.5648,
+      "step": 3838
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.04802431968839e-06,
+      "loss": 0.5267,
+      "step": 3839
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.042672546689202e-06,
+      "loss": 0.531,
+      "step": 3840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.037323417337894e-06,
+      "loss": 0.5729,
+      "step": 3841
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.031976934008216e-06,
+      "loss": 0.5626,
+      "step": 3842
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.026633099072746e-06,
+      "loss": 0.5367,
+      "step": 3843
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.021291914902887e-06,
+      "loss": 0.5287,
+      "step": 3844
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.015953383868864e-06,
+      "loss": 0.5348,
+      "step": 3845
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.010617508339724e-06,
+      "loss": 0.56,
+      "step": 3846
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.00528429068334e-06,
+      "loss": 0.5343,
+      "step": 3847
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.999953733266395e-06,
+      "loss": 0.5332,
+      "step": 3848
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.994625838454409e-06,
+      "loss": 0.5049,
+      "step": 3849
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.989300608611709e-06,
+      "loss": 0.497,
+      "step": 3850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.98397804610143e-06,
+      "loss": 0.5096,
+      "step": 3851
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.978658153285544e-06,
+      "loss": 0.6125,
+      "step": 3852
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.973340932524823e-06,
+      "loss": 0.6045,
+      "step": 3853
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.968026386178867e-06,
+      "loss": 0.5425,
+      "step": 3854
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.962714516606072e-06,
+      "loss": 0.5693,
+      "step": 3855
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.957405326163653e-06,
+      "loss": 0.5231,
+      "step": 3856
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.952098817207647e-06,
+      "loss": 0.5388,
+      "step": 3857
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.946794992092887e-06,
+      "loss": 0.5205,
+      "step": 3858
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.941493853173022e-06,
+      "loss": 0.5421,
+      "step": 3859
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.936195402800504e-06,
+      "loss": 0.5597,
+      "step": 3860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.930899643326598e-06,
+      "loss": 0.5359,
+      "step": 3861
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.925606577101365e-06,
+      "loss": 0.5046,
+      "step": 3862
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.920316206473691e-06,
+      "loss": 0.5099,
+      "step": 3863
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.9150285337912404e-06,
+      "loss": 0.5707,
+      "step": 3864
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.909743561400493e-06,
+      "loss": 0.5233,
+      "step": 3865
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.904461291646736e-06,
+      "loss": 0.5103,
+      "step": 3866
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.899181726874048e-06,
+      "loss": 0.5648,
+      "step": 3867
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.89390486942531e-06,
+      "loss": 0.491,
+      "step": 3868
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.888630721642203e-06,
+      "loss": 0.5353,
+      "step": 3869
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.883359285865197e-06,
+      "loss": 0.5017,
+      "step": 3870
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.878090564433581e-06,
+      "loss": 0.544,
+      "step": 3871
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.87282455968541e-06,
+      "loss": 0.5278,
+      "step": 3872
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.86756127395755e-06,
+      "loss": 0.5221,
+      "step": 3873
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.862300709585664e-06,
+      "loss": 0.5278,
+      "step": 3874
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8570428689042e-06,
+      "loss": 0.495,
+      "step": 3875
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.851787754246395e-06,
+      "loss": 0.5176,
+      "step": 3876
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.846535367944283e-06,
+      "loss": 0.5247,
+      "step": 3877
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8412857123286825e-06,
+      "loss": 0.4974,
+      "step": 3878
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.836038789729201e-06,
+      "loss": 0.5426,
+      "step": 3879
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.830794602474242e-06,
+      "loss": 0.5853,
+      "step": 3880
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.825553152890976e-06,
+      "loss": 0.5497,
+      "step": 3881
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.820314443305379e-06,
+      "loss": 0.5763,
+      "step": 3882
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.815078476042198e-06,
+      "loss": 0.5311,
+      "step": 3883
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.809845253424969e-06,
+      "loss": 0.5324,
+      "step": 3884
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8046147777760055e-06,
+      "loss": 0.5682,
+      "step": 3885
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.799387051416408e-06,
+      "loss": 0.5706,
+      "step": 3886
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.794162076666047e-06,
+      "loss": 0.4897,
+      "step": 3887
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.7889398558435873e-06,
+      "loss": 0.5162,
+      "step": 3888
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.783720391266462e-06,
+      "loss": 0.5537,
+      "step": 3889
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.7785036852508726e-06,
+      "loss": 0.5869,
+      "step": 3890
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.7732897401118164e-06,
+      "loss": 0.543,
+      "step": 3891
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.7680785581630495e-06,
+      "loss": 0.5456,
+      "step": 3892
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.7628701417171087e-06,
+      "loss": 0.5005,
+      "step": 3893
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.7576644930853025e-06,
+      "loss": 0.5763,
+      "step": 3894
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.752461614577705e-06,
+      "loss": 0.5662,
+      "step": 3895
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.747261508503176e-06,
+      "loss": 0.4918,
+      "step": 3896
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.7420641771693355e-06,
+      "loss": 0.5491,
+      "step": 3897
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.7368696228825608e-06,
+      "loss": 0.528,
+      "step": 3898
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.7316778479480187e-06,
+      "loss": 0.4886,
+      "step": 3899
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.726488854669631e-06,
+      "loss": 0.5697,
+      "step": 3900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.7213026453500847e-06,
+      "loss": 0.5332,
+      "step": 3901
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.7161192222908338e-06,
+      "loss": 0.542,
+      "step": 3902
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.7109385877920946e-06,
+      "loss": 0.5502,
+      "step": 3903
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.705760744152844e-06,
+      "loss": 0.581,
+      "step": 3904
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.7005856936708296e-06,
+      "loss": 0.5522,
+      "step": 3905
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.6954134386425487e-06,
+      "loss": 0.4677,
+      "step": 3906
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.690243981363263e-06,
+      "loss": 0.5412,
+      "step": 3907
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.6850773241269923e-06,
+      "loss": 0.5401,
+      "step": 3908
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.6799134692265116e-06,
+      "loss": 0.5302,
+      "step": 3909
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.6747524189533567e-06,
+      "loss": 0.5375,
+      "step": 3910
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.6695941755978135e-06,
+      "loss": 0.5656,
+      "step": 3911
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.6644387414489248e-06,
+      "loss": 0.5538,
+      "step": 3912
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.6592861187944916e-06,
+      "loss": 0.5504,
+      "step": 3913
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.6541363099210625e-06,
+      "loss": 0.5679,
+      "step": 3914
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.648989317113929e-06,
+      "loss": 0.5858,
+      "step": 3915
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.643845142657151e-06,
+      "loss": 0.5287,
+      "step": 3916
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.638703788833525e-06,
+      "loss": 0.5406,
+      "step": 3917
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.6335652579245995e-06,
+      "loss": 0.5827,
+      "step": 3918
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.6284295522106684e-06,
+      "loss": 0.5446,
+      "step": 3919
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.6232966739707707e-06,
+      "loss": 0.5191,
+      "step": 3920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.618166625482702e-06,
+      "loss": 0.571,
+      "step": 3921
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.613039409022987e-06,
+      "loss": 0.5643,
+      "step": 3922
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.607915026866903e-06,
+      "loss": 0.472,
+      "step": 3923
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.6027934812884647e-06,
+      "loss": 0.5061,
+      "step": 3924
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.59767477456043e-06,
+      "loss": 0.5753,
+      "step": 3925
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5925589089542954e-06,
+      "loss": 0.5713,
+      "step": 3926
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.587445886740307e-06,
+      "loss": 0.5521,
+      "step": 3927
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.58233571018743e-06,
+      "loss": 0.55,
+      "step": 3928
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.577228381563379e-06,
+      "loss": 0.5352,
+      "step": 3929
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.5165,
+      "step": 3930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5670222771662964e-06,
+      "loss": 0.559,
+      "step": 3931
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5619235059223657e-06,
+      "loss": 0.5356,
+      "step": 3932
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.556827591665465e-06,
+      "loss": 0.5698,
+      "step": 3933
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5517345366569767e-06,
+      "loss": 0.5571,
+      "step": 3934
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.546644343157024e-06,
+      "loss": 0.5778,
+      "step": 3935
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5415570134244414e-06,
+      "loss": 0.5399,
+      "step": 3936
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5364725497168052e-06,
+      "loss": 0.5869,
+      "step": 3937
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5313909542904224e-06,
+      "loss": 0.5763,
+      "step": 3938
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5263122294003203e-06,
+      "loss": 0.5633,
+      "step": 3939
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5212363773002545e-06,
+      "loss": 0.5746,
+      "step": 3940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5161634002427047e-06,
+      "loss": 0.5166,
+      "step": 3941
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5110933004788773e-06,
+      "loss": 0.4887,
+      "step": 3942
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5060260802586954e-06,
+      "loss": 0.6018,
+      "step": 3943
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5009617418308208e-06,
+      "loss": 0.5639,
+      "step": 3944
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.495900287442614e-06,
+      "loss": 0.5214,
+      "step": 3945
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4908417193401667e-06,
+      "loss": 0.5391,
+      "step": 3946
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4857860397682974e-06,
+      "loss": 0.5367,
+      "step": 3947
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4807332509705305e-06,
+      "loss": 0.6216,
+      "step": 3948
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4756833551891123e-06,
+      "loss": 0.5046,
+      "step": 3949
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.470636354665006e-06,
+      "loss": 0.5698,
+      "step": 3950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4655922516378847e-06,
+      "loss": 0.5335,
+      "step": 3951
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.460551048346147e-06,
+      "loss": 0.5538,
+      "step": 3952
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.455512747026898e-06,
+      "loss": 0.568,
+      "step": 3953
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.450477349915945e-06,
+      "loss": 0.5518,
+      "step": 3954
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.445444859247825e-06,
+      "loss": 0.5626,
+      "step": 3955
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.440415277255775e-06,
+      "loss": 0.498,
+      "step": 3956
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4353886061717433e-06,
+      "loss": 0.4739,
+      "step": 3957
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.430364848226383e-06,
+      "loss": 0.5423,
+      "step": 3958
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4253440056490592e-06,
+      "loss": 0.4818,
+      "step": 3959
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4203260806678383e-06,
+      "loss": 0.5785,
+      "step": 3960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4153110755095053e-06,
+      "loss": 0.5718,
+      "step": 3961
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4102989923995246e-06,
+      "loss": 0.4876,
+      "step": 3962
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.40528983356209e-06,
+      "loss": 0.547,
+      "step": 3963
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4002836012200824e-06,
+      "loss": 0.498,
+      "step": 3964
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.395280297595087e-06,
+      "loss": 0.5807,
+      "step": 3965
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.3902799249073912e-06,
+      "loss": 0.5334,
+      "step": 3966
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.3852824853759813e-06,
+      "loss": 0.5669,
+      "step": 3967
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.380287981218536e-06,
+      "loss": 0.5194,
+      "step": 3968
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.3752964146514454e-06,
+      "loss": 0.5155,
+      "step": 3969
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.370307787889787e-06,
+      "loss": 0.5257,
+      "step": 3970
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.3653221031473236e-06,
+      "loss": 0.5759,
+      "step": 3971
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.3603393626365322e-06,
+      "loss": 0.5293,
+      "step": 3972
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.3553595685685725e-06,
+      "loss": 0.5143,
+      "step": 3973
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.350382723153298e-06,
+      "loss": 0.5113,
+      "step": 3974
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.3454088285992525e-06,
+      "loss": 0.6393,
+      "step": 3975
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.340437887113669e-06,
+      "loss": 0.5254,
+      "step": 3976
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.3354699009024795e-06,
+      "loss": 0.5452,
+      "step": 3977
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.330504872170298e-06,
+      "loss": 0.5566,
+      "step": 3978
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.325542803120416e-06,
+      "loss": 0.5587,
+      "step": 3979
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.32058369595483e-06,
+      "loss": 0.5481,
+      "step": 3980
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.315627552874212e-06,
+      "loss": 0.545,
+      "step": 3981
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.3106743760779204e-06,
+      "loss": 0.5343,
+      "step": 3982
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.305724167763995e-06,
+      "loss": 0.5424,
+      "step": 3983
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.3007769301291613e-06,
+      "loss": 0.5411,
+      "step": 3984
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2958326653688245e-06,
+      "loss": 0.4837,
+      "step": 3985
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2908913756770744e-06,
+      "loss": 0.5677,
+      "step": 3986
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2859530632466772e-06,
+      "loss": 0.5596,
+      "step": 3987
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2810177302690792e-06,
+      "loss": 0.5516,
+      "step": 3988
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2760853789344005e-06,
+      "loss": 0.4762,
+      "step": 3989
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2711560114314412e-06,
+      "loss": 0.4958,
+      "step": 3990
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.266229629947687e-06,
+      "loss": 0.5671,
+      "step": 3991
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2613062366692773e-06,
+      "loss": 0.5388,
+      "step": 3992
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2563858337810385e-06,
+      "loss": 0.5388,
+      "step": 3993
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.251468423466475e-06,
+      "loss": 0.5705,
+      "step": 3994
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.246554007907753e-06,
+      "loss": 0.5436,
+      "step": 3995
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.241642589285715e-06,
+      "loss": 0.5402,
+      "step": 3996
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2367341697798705e-06,
+      "loss": 0.5348,
+      "step": 3997
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.231828751568401e-06,
+      "loss": 0.5459,
+      "step": 3998
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2269263368281532e-06,
+      "loss": 0.5125,
+      "step": 3999
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2220269277346437e-06,
+      "loss": 0.5385,
+      "step": 4000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2171305264620523e-06,
+      "loss": 0.5446,
+      "step": 4001
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2122371351832302e-06,
+      "loss": 0.5854,
+      "step": 4002
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.207346756069686e-06,
+      "loss": 0.5841,
+      "step": 4003
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2024593912915945e-06,
+      "loss": 0.5708,
+      "step": 4004
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.197575043017793e-06,
+      "loss": 0.535,
+      "step": 4005
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.19269371341578e-06,
+      "loss": 0.5983,
+      "step": 4006
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1878154046517095e-06,
+      "loss": 0.5621,
+      "step": 4007
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1829401188904096e-06,
+      "loss": 0.5479,
+      "step": 4008
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1780678582953483e-06,
+      "loss": 0.5981,
+      "step": 4009
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1731986250286593e-06,
+      "loss": 0.5496,
+      "step": 4010
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.16833242125114e-06,
+      "loss": 0.5107,
+      "step": 4011
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.163469249122234e-06,
+      "loss": 0.5406,
+      "step": 4012
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1586091108000416e-06,
+      "loss": 0.5555,
+      "step": 4013
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1537520084413175e-06,
+      "loss": 0.587,
+      "step": 4014
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1488979442014657e-06,
+      "loss": 0.5664,
+      "step": 4015
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1440469202345534e-06,
+      "loss": 0.5728,
+      "step": 4016
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.13919893869329e-06,
+      "loss": 0.5059,
+      "step": 4017
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1343540017290242e-06,
+      "loss": 0.5076,
+      "step": 4018
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.129512111491777e-06,
+      "loss": 0.5408,
+      "step": 4019
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.124673270130202e-06,
+      "loss": 0.5232,
+      "step": 4020
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.119837479791601e-06,
+      "loss": 0.5743,
+      "step": 4021
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1150047426219253e-06,
+      "loss": 0.5026,
+      "step": 4022
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.110175060765771e-06,
+      "loss": 0.5037,
+      "step": 4023
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.105348436366373e-06,
+      "loss": 0.5246,
+      "step": 4024
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.1005248715656235e-06,
+      "loss": 0.5576,
+      "step": 4025
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0957043685040376e-06,
+      "loss": 0.575,
+      "step": 4026
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0908869293207822e-06,
+      "loss": 0.4527,
+      "step": 4027
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0860725561536687e-06,
+      "loss": 0.5712,
+      "step": 4028
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0812612511391406e-06,
+      "loss": 0.5151,
+      "step": 4029
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.076453016412283e-06,
+      "loss": 0.5221,
+      "step": 4030
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.071647854106816e-06,
+      "loss": 0.5226,
+      "step": 4031
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0668457663550955e-06,
+      "loss": 0.5593,
+      "step": 4032
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0620467552881216e-06,
+      "loss": 0.5368,
+      "step": 4033
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0572508230355245e-06,
+      "loss": 0.5537,
+      "step": 4034
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0524579717255554e-06,
+      "loss": 0.5615,
+      "step": 4035
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0476682034851204e-06,
+      "loss": 0.5291,
+      "step": 4036
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.042881520439741e-06,
+      "loss": 0.5441,
+      "step": 4037
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0380979247135777e-06,
+      "loss": 0.5334,
+      "step": 4038
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0333174184294166e-06,
+      "loss": 0.5448,
+      "step": 4039
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0285400037086763e-06,
+      "loss": 0.497,
+      "step": 4040
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.023765682671398e-06,
+      "loss": 0.4927,
+      "step": 4041
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.018994457436263e-06,
+      "loss": 0.5609,
+      "step": 4042
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0142263301205564e-06,
+      "loss": 0.5437,
+      "step": 4043
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.009461302840212e-06,
+      "loss": 0.5418,
+      "step": 4044
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0046993777097755e-06,
+      "loss": 0.505,
+      "step": 4045
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.9999405568424166e-06,
+      "loss": 0.5295,
+      "step": 4046
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.9951848423499296e-06,
+      "loss": 0.5397,
+      "step": 4047
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.9904322363427297e-06,
+      "loss": 0.5147,
+      "step": 4048
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.9856827409298496e-06,
+      "loss": 0.5091,
+      "step": 4049
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.980936358218951e-06,
+      "loss": 0.5204,
+      "step": 4050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.976193090316307e-06,
+      "loss": 0.5684,
+      "step": 4051
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.971452939326802e-06,
+      "loss": 0.5205,
+      "step": 4052
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.966715907353953e-06,
+      "loss": 0.5062,
+      "step": 4053
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.961981996499882e-06,
+      "loss": 0.4728,
+      "step": 4054
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.957251208865328e-06,
+      "loss": 0.53,
+      "step": 4055
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.952523546549645e-06,
+      "loss": 0.4894,
+      "step": 4056
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.947799011650796e-06,
+      "loss": 0.4957,
+      "step": 4057
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.9430776062653666e-06,
+      "loss": 0.5384,
+      "step": 4058
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.9383593324885427e-06,
+      "loss": 0.5387,
+      "step": 4059
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.933644192414126e-06,
+      "loss": 0.5133,
+      "step": 4060
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.5542,
+      "step": 4061
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.92422332174076e-06,
+      "loss": 0.5265,
+      "step": 4062
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.9195175953224554e-06,
+      "loss": 0.5407,
+      "step": 4063
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.914815010967842e-06,
+      "loss": 0.5716,
+      "step": 4064
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.9101155707637605e-06,
+      "loss": 0.4877,
+      "step": 4065
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.905419276795648e-06,
+      "loss": 0.579,
+      "step": 4066
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.90072613114756e-06,
+      "loss": 0.5339,
+      "step": 4067
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.8960361359021407e-06,
+      "loss": 0.5525,
+      "step": 4068
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.891349293140642e-06,
+      "loss": 0.5582,
+      "step": 4069
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.8866656049429166e-06,
+      "loss": 0.5777,
+      "step": 4070
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.8819850733874134e-06,
+      "loss": 0.5155,
+      "step": 4071
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.877307700551194e-06,
+      "loss": 0.5598,
+      "step": 4072
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.8726334885098983e-06,
+      "loss": 0.5302,
+      "step": 4073
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.867962439337775e-06,
+      "loss": 0.4922,
+      "step": 4074
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.863294555107673e-06,
+      "loss": 0.5627,
+      "step": 4075
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.8586298378910293e-06,
+      "loss": 0.5669,
+      "step": 4076
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.853968289757877e-06,
+      "loss": 0.5364,
+      "step": 4077
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.849309912776843e-06,
+      "loss": 0.5519,
+      "step": 4078
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.84465470901515e-06,
+      "loss": 0.5261,
+      "step": 4079
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.840002680538606e-06,
+      "loss": 0.5007,
+      "step": 4080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.835353829411622e-06,
+      "loss": 0.5171,
+      "step": 4081
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.8307081576971806e-06,
+      "loss": 0.5266,
+      "step": 4082
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.8260656674568722e-06,
+      "loss": 0.5194,
+      "step": 4083
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.821426360750864e-06,
+      "loss": 0.5374,
+      "step": 4084
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.816790239637914e-06,
+      "loss": 0.5825,
+      "step": 4085
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.812157306175367e-06,
+      "loss": 0.527,
+      "step": 4086
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.8075275624191513e-06,
+      "loss": 0.5641,
+      "step": 4087
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.8029010104237786e-06,
+      "loss": 0.4997,
+      "step": 4088
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.7982776522423537e-06,
+      "loss": 0.5544,
+      "step": 4089
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.79365748992655e-06,
+      "loss": 0.5541,
+      "step": 4090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.7890405255266286e-06,
+      "loss": 0.5213,
+      "step": 4091
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.7844267610914376e-06,
+      "loss": 0.5567,
+      "step": 4092
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.779816198668399e-06,
+      "loss": 0.6103,
+      "step": 4093
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.7752088403035127e-06,
+      "loss": 0.536,
+      "step": 4094
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.7706046880413596e-06,
+      "loss": 0.5698,
+      "step": 4095
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.7660037439250942e-06,
+      "loss": 0.5136,
+      "step": 4096
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.761406009996457e-06,
+      "loss": 0.5287,
+      "step": 4097
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.7568114882957553e-06,
+      "loss": 0.5309,
+      "step": 4098
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.752220180861864e-06,
+      "loss": 0.5557,
+      "step": 4099
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.7476320897322507e-06,
+      "loss": 0.5442,
+      "step": 4100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.7430472169429412e-06,
+      "loss": 0.5731,
+      "step": 4101
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.7384655645285385e-06,
+      "loss": 0.5543,
+      "step": 4102
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.733887134522214e-06,
+      "loss": 0.545,
+      "step": 4103
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.7293119289557114e-06,
+      "loss": 0.5685,
+      "step": 4104
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.72473994985934e-06,
+      "loss": 0.5534,
+      "step": 4105
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.7201711992619873e-06,
+      "loss": 0.6132,
+      "step": 4106
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.715605679191092e-06,
+      "loss": 0.4948,
+      "step": 4107
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.7110433916726686e-06,
+      "loss": 0.5364,
+      "step": 4108
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.7064843387313024e-06,
+      "loss": 0.5237,
+      "step": 4109
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.7019285223901326e-06,
+      "loss": 0.5497,
+      "step": 4110
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6973759446708693e-06,
+      "loss": 0.5931,
+      "step": 4111
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6928266075937827e-06,
+      "loss": 0.5136,
+      "step": 4112
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.688280513177699e-06,
+      "loss": 0.5184,
+      "step": 4113
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6837376634400215e-06,
+      "loss": 0.5467,
+      "step": 4114
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.679198060396702e-06,
+      "loss": 0.5104,
+      "step": 4115
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6746617060622427e-06,
+      "loss": 0.514,
+      "step": 4116
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6701286024497265e-06,
+      "loss": 0.6003,
+      "step": 4117
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.665598751570777e-06,
+      "loss": 0.5048,
+      "step": 4118
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.66107215543558e-06,
+      "loss": 0.5427,
+      "step": 4119
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6565488160528764e-06,
+      "loss": 0.5476,
+      "step": 4120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.652028735429961e-06,
+      "loss": 0.5478,
+      "step": 4121
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6475119155726804e-06,
+      "loss": 0.5229,
+      "step": 4122
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6429983584854436e-06,
+      "loss": 0.5442,
+      "step": 4123
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.638488066171201e-06,
+      "loss": 0.486,
+      "step": 4124
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.633981040631459e-06,
+      "loss": 0.5275,
+      "step": 4125
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.629477283866274e-06,
+      "loss": 0.5044,
+      "step": 4126
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.62497679787425e-06,
+      "loss": 0.555,
+      "step": 4127
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.620479584652542e-06,
+      "loss": 0.58,
+      "step": 4128
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6159856461968524e-06,
+      "loss": 0.5094,
+      "step": 4129
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6114949845014248e-06,
+      "loss": 0.5799,
+      "step": 4130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6070076015590618e-06,
+      "loss": 0.5898,
+      "step": 4131
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6025234993610993e-06,
+      "loss": 0.4839,
+      "step": 4132
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5980426798974156e-06,
+      "loss": 0.5415,
+      "step": 4133
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5935651451564437e-06,
+      "loss": 0.5391,
+      "step": 4134
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5890908971251506e-06,
+      "loss": 0.5454,
+      "step": 4135
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5846199377890468e-06,
+      "loss": 0.5287,
+      "step": 4136
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5801522691321836e-06,
+      "loss": 0.583,
+      "step": 4137
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5756878931371475e-06,
+      "loss": 0.582,
+      "step": 4138
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.571226811785077e-06,
+      "loss": 0.5209,
+      "step": 4139
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5667690270556332e-06,
+      "loss": 0.546,
+      "step": 4140
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5623145409270244e-06,
+      "loss": 0.5901,
+      "step": 4141
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.557863355375988e-06,
+      "loss": 0.5442,
+      "step": 4142
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5534154723778027e-06,
+      "loss": 0.5262,
+      "step": 4143
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.548970893906275e-06,
+      "loss": 0.5305,
+      "step": 4144
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.544529621933759e-06,
+      "loss": 0.5584,
+      "step": 4145
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.540091658431122e-06,
+      "loss": 0.4891,
+      "step": 4146
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.535657005367773e-06,
+      "loss": 0.5272,
+      "step": 4147
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.531225664711657e-06,
+      "loss": 0.5534,
+      "step": 4148
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5267976384292426e-06,
+      "loss": 0.5486,
+      "step": 4149
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.522372928485526e-06,
+      "loss": 0.5509,
+      "step": 4150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5179515368440364e-06,
+      "loss": 0.5507,
+      "step": 4151
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.513533465466825e-06,
+      "loss": 0.5248,
+      "step": 4152
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5091187163144825e-06,
+      "loss": 0.5456,
+      "step": 4153
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5047072913461068e-06,
+      "loss": 0.5758,
+      "step": 4154
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.50029919251933e-06,
+      "loss": 0.5674,
+      "step": 4155
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4958944217903136e-06,
+      "loss": 0.5425,
+      "step": 4156
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4914929811137333e-06,
+      "loss": 0.6002,
+      "step": 4157
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.487094872442791e-06,
+      "loss": 0.5563,
+      "step": 4158
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4827000977292083e-06,
+      "loss": 0.5352,
+      "step": 4159
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.47830865892323e-06,
+      "loss": 0.5533,
+      "step": 4160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.473920557973615e-06,
+      "loss": 0.5499,
+      "step": 4161
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4695357968276533e-06,
+      "loss": 0.5571,
+      "step": 4162
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4651543774311327e-06,
+      "loss": 0.5474,
+      "step": 4163
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4607763017283794e-06,
+      "loss": 0.5188,
+      "step": 4164
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4564015716622216e-06,
+      "loss": 0.5705,
+      "step": 4165
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4520301891740082e-06,
+      "loss": 0.5585,
+      "step": 4166
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.447662156203603e-06,
+      "loss": 0.5775,
+      "step": 4167
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.443297474689379e-06,
+      "loss": 0.4967,
+      "step": 4168
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.438936146568225e-06,
+      "loss": 0.5248,
+      "step": 4169
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.434578173775549e-06,
+      "loss": 0.5841,
+      "step": 4170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4302235582452547e-06,
+      "loss": 0.5314,
+      "step": 4171
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.425872301909764e-06,
+      "loss": 0.5844,
+      "step": 4172
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4215244067000133e-06,
+      "loss": 0.5193,
+      "step": 4173
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.41717987454544e-06,
+      "loss": 0.5618,
+      "step": 4174
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4128387073739912e-06,
+      "loss": 0.518,
+      "step": 4175
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.40850090711212e-06,
+      "loss": 0.5536,
+      "step": 4176
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.404166475684785e-06,
+      "loss": 0.5004,
+      "step": 4177
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3998354150154555e-06,
+      "loss": 0.5407,
+      "step": 4178
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3955077270261017e-06,
+      "loss": 0.5513,
+      "step": 4179
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.391183413637186e-06,
+      "loss": 0.5199,
+      "step": 4180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.386862476767694e-06,
+      "loss": 0.5069,
+      "step": 4181
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.382544918335098e-06,
+      "loss": 0.5331,
+      "step": 4182
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.378230740255375e-06,
+      "loss": 0.5663,
+      "step": 4183
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3739199444430027e-06,
+      "loss": 0.5511,
+      "step": 4184
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3696125328109585e-06,
+      "loss": 0.552,
+      "step": 4185
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3653085072707127e-06,
+      "loss": 0.5012,
+      "step": 4186
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3610078697322426e-06,
+      "loss": 0.5511,
+      "step": 4187
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.356710622104017e-06,
+      "loss": 0.5295,
+      "step": 4188
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.352416766292993e-06,
+      "loss": 0.5819,
+      "step": 4189
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3481263042046353e-06,
+      "loss": 0.5618,
+      "step": 4190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3438392377428964e-06,
+      "loss": 0.5163,
+      "step": 4191
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.5118,
+      "step": 4192
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3352752993075477e-06,
+      "loss": 0.6123,
+      "step": 4193
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.330998431134305e-06,
+      "loss": 0.5011,
+      "step": 4194
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.326724966188416e-06,
+      "loss": 0.553,
+      "step": 4195
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3224549063662925e-06,
+      "loss": 0.564,
+      "step": 4196
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3181882535628264e-06,
+      "loss": 0.4883,
+      "step": 4197
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3139250096714116e-06,
+      "loss": 0.5891,
+      "step": 4198
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3096651765839208e-06,
+      "loss": 0.5535,
+      "step": 4199
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3054087561907133e-06,
+      "loss": 0.5735,
+      "step": 4200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.301155750380637e-06,
+      "loss": 0.5471,
+      "step": 4201
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.2969061610410224e-06,
+      "loss": 0.5133,
+      "step": 4202
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.292659990057683e-06,
+      "loss": 0.5577,
+      "step": 4203
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.2884172393149217e-06,
+      "loss": 0.572,
+      "step": 4204
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.2841779106955155e-06,
+      "loss": 0.5629,
+      "step": 4205
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2799420060807266e-06,
+      "loss": 0.5753,
+      "step": 4206
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.275709527350296e-06,
+      "loss": 0.5733,
+      "step": 4207
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2714804763824448e-06,
+      "loss": 0.5318,
+      "step": 4208
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2672548550538833e-06,
+      "loss": 0.5334,
+      "step": 4209
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.263032665239778e-06,
+      "loss": 0.4834,
+      "step": 4210
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2588139088137885e-06,
+      "loss": 0.5235,
+      "step": 4211
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2545985876480514e-06,
+      "loss": 0.532,
+      "step": 4212
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.250386703613172e-06,
+      "loss": 0.5457,
+      "step": 4213
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2461782585782346e-06,
+      "loss": 0.5076,
+      "step": 4214
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.241973254410794e-06,
+      "loss": 0.5201,
+      "step": 4215
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.23777169297688e-06,
+      "loss": 0.5291,
+      "step": 4216
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2335735761409962e-06,
+      "loss": 0.4877,
+      "step": 4217
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2293789057661153e-06,
+      "loss": 0.525,
+      "step": 4218
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2251876837136776e-06,
+      "loss": 0.5031,
+      "step": 4219
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2209999118436044e-06,
+      "loss": 0.5904,
+      "step": 4220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2168155920142733e-06,
+      "loss": 0.5586,
+      "step": 4221
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2126347260825363e-06,
+      "loss": 0.5292,
+      "step": 4222
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2084573159037113e-06,
+      "loss": 0.5579,
+      "step": 4223
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.204283363331583e-06,
+      "loss": 0.5479,
+      "step": 4224
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.200112870218399e-06,
+      "loss": 0.596,
+      "step": 4225
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1959458384148823e-06,
+      "loss": 0.5198,
+      "step": 4226
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.191782269770203e-06,
+      "loss": 0.5156,
+      "step": 4227
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.187622166132004e-06,
+      "loss": 0.5316,
+      "step": 4228
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.183465529346396e-06,
+      "loss": 0.5112,
+      "step": 4229
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1793123612579413e-06,
+      "loss": 0.5625,
+      "step": 4230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1751626637096668e-06,
+      "loss": 0.5531,
+      "step": 4231
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.171016438543059e-06,
+      "loss": 0.5698,
+      "step": 4232
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.166873687598061e-06,
+      "loss": 0.516,
+      "step": 4233
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1627344127130857e-06,
+      "loss": 0.5583,
+      "step": 4234
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1585986157249862e-06,
+      "loss": 0.5218,
+      "step": 4235
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.15446629846908e-06,
+      "loss": 0.5567,
+      "step": 4236
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1503374627791464e-06,
+      "loss": 0.5449,
+      "step": 4237
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1462121104874124e-06,
+      "loss": 0.5688,
+      "step": 4238
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1420902434245605e-06,
+      "loss": 0.5358,
+      "step": 4239
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.137971863419728e-06,
+      "loss": 0.5067,
+      "step": 4240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1338569723005034e-06,
+      "loss": 0.5594,
+      "step": 4241
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.129745571892924e-06,
+      "loss": 0.524,
+      "step": 4242
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1256376640214904e-06,
+      "loss": 0.513,
+      "step": 4243
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1215332505091346e-06,
+      "loss": 0.5749,
+      "step": 4244
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.117432333177255e-06,
+      "loss": 0.5435,
+      "step": 4245
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.113334913845689e-06,
+      "loss": 0.5193,
+      "step": 4246
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.109240994332724e-06,
+      "loss": 0.5332,
+      "step": 4247
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.105150576455094e-06,
+      "loss": 0.541,
+      "step": 4248
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1010636620279802e-06,
+      "loss": 0.5445,
+      "step": 4249
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0969802528650052e-06,
+      "loss": 0.526,
+      "step": 4250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0929003507782442e-06,
+      "loss": 0.5811,
+      "step": 4251
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.088823957578212e-06,
+      "loss": 0.5143,
+      "step": 4252
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0847510750738566e-06,
+      "loss": 0.4923,
+      "step": 4253
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.080681705072586e-06,
+      "loss": 0.5609,
+      "step": 4254
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.076615849380238e-06,
+      "loss": 0.5504,
+      "step": 4255
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.072553509801091e-06,
+      "loss": 0.5106,
+      "step": 4256
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0684946881378677e-06,
+      "loss": 0.5667,
+      "step": 4257
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0644393861917234e-06,
+      "loss": 0.5423,
+      "step": 4258
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0603876057622607e-06,
+      "loss": 0.5177,
+      "step": 4259
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.056339348647515e-06,
+      "loss": 0.6032,
+      "step": 4260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0522946166439494e-06,
+      "loss": 0.5395,
+      "step": 4261
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.048253411546477e-06,
+      "loss": 0.5668,
+      "step": 4262
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0442157351484394e-06,
+      "loss": 0.5689,
+      "step": 4263
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0401815892416098e-06,
+      "loss": 0.5541,
+      "step": 4264
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.036150975616198e-06,
+      "loss": 0.5336,
+      "step": 4265
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0321238960608456e-06,
+      "loss": 0.5442,
+      "step": 4266
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0281003523626242e-06,
+      "loss": 0.5797,
+      "step": 4267
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0240803463070425e-06,
+      "loss": 0.5377,
+      "step": 4268
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0200638796780337e-06,
+      "loss": 0.5395,
+      "step": 4269
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0160509542579544e-06,
+      "loss": 0.5989,
+      "step": 4270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0120415718276055e-06,
+      "loss": 0.5191,
+      "step": 4271
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.008035734166199e-06,
+      "loss": 0.5236,
+      "step": 4272
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.004033443051393e-06,
+      "loss": 0.5793,
+      "step": 4273
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.00003470025925e-06,
+      "loss": 0.5744,
+      "step": 4274
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.996039507564269e-06,
+      "loss": 0.5662,
+      "step": 4275
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9920478667393797e-06,
+      "loss": 0.5531,
+      "step": 4276
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.988059779555923e-06,
+      "loss": 0.5322,
+      "step": 4277
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9840752477836722e-06,
+      "loss": 0.5364,
+      "step": 4278
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.980094273190818e-06,
+      "loss": 0.5084,
+      "step": 4279
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.976116857543974e-06,
+      "loss": 0.5421,
+      "step": 4280
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9721430026081735e-06,
+      "loss": 0.5205,
+      "step": 4281
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.968172710146871e-06,
+      "loss": 0.5801,
+      "step": 4282
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9642059819219405e-06,
+      "loss": 0.5065,
+      "step": 4283
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9602428196936695e-06,
+      "loss": 0.5076,
+      "step": 4284
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.956283225220773e-06,
+      "loss": 0.5541,
+      "step": 4285
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9523272002603745e-06,
+      "loss": 0.59,
+      "step": 4286
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9483747465680147e-06,
+      "loss": 0.6,
+      "step": 4287
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9444258658976513e-06,
+      "loss": 0.5626,
+      "step": 4288
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.940480560001653e-06,
+      "loss": 0.5436,
+      "step": 4289
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.936538830630813e-06,
+      "loss": 0.5624,
+      "step": 4290
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9326006795343198e-06,
+      "loss": 0.48,
+      "step": 4291
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9286661084597857e-06,
+      "loss": 0.5901,
+      "step": 4292
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.924735119153235e-06,
+      "loss": 0.5357,
+      "step": 4293
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9208077133590973e-06,
+      "loss": 0.4845,
+      "step": 4294
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.916883892820216e-06,
+      "loss": 0.5346,
+      "step": 4295
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.91296365927784e-06,
+      "loss": 0.536,
+      "step": 4296
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.909047014471628e-06,
+      "loss": 0.5629,
+      "step": 4297
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9051339601396469e-06,
+      "loss": 0.5791,
+      "step": 4298
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9012244980183692e-06,
+      "loss": 0.5652,
+      "step": 4299
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8973186298426715e-06,
+      "loss": 0.5039,
+      "step": 4300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.893416357345843e-06,
+      "loss": 0.544,
+      "step": 4301
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8895176822595695e-06,
+      "loss": 0.5555,
+      "step": 4302
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8856226063139426e-06,
+      "loss": 0.5894,
+      "step": 4303
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8817311312374565e-06,
+      "loss": 0.5428,
+      "step": 4304
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8778432587570084e-06,
+      "loss": 0.5726,
+      "step": 4305
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.873958990597895e-06,
+      "loss": 0.5468,
+      "step": 4306
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8700783284838208e-06,
+      "loss": 0.4866,
+      "step": 4307
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8662012741368784e-06,
+      "loss": 0.5537,
+      "step": 4308
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8623278292775638e-06,
+      "loss": 0.4742,
+      "step": 4309
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8584579956247783e-06,
+      "loss": 0.5429,
+      "step": 4310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8545917748958132e-06,
+      "loss": 0.5312,
+      "step": 4311
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8507291688063566e-06,
+      "loss": 0.5143,
+      "step": 4312
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8468701790704958e-06,
+      "loss": 0.5129,
+      "step": 4313
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8430148074007082e-06,
+      "loss": 0.4828,
+      "step": 4314
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8391630555078755e-06,
+      "loss": 0.5795,
+      "step": 4315
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8353149251012648e-06,
+      "loss": 0.5424,
+      "step": 4316
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8314704178885334e-06,
+      "loss": 0.5425,
+      "step": 4317
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8276295355757401e-06,
+      "loss": 0.518,
+      "step": 4318
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8237922798673302e-06,
+      "loss": 0.531,
+      "step": 4319
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8199586524661395e-06,
+      "loss": 0.6,
+      "step": 4320
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8161286550733926e-06,
+      "loss": 0.5622,
+      "step": 4321
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8123022893887065e-06,
+      "loss": 0.5728,
+      "step": 4322
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.808479557110081e-06,
+      "loss": 0.5708,
+      "step": 4323
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8046604599339168e-06,
+      "loss": 0.5036,
+      "step": 4324
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8008449995549815e-06,
+      "loss": 0.5275,
+      "step": 4325
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7970331776664418e-06,
+      "loss": 0.5594,
+      "step": 4326
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7932249959598514e-06,
+      "loss": 0.5283,
+      "step": 4327
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7894204561251416e-06,
+      "loss": 0.5533,
+      "step": 4328
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.785619559850632e-06,
+      "loss": 0.5042,
+      "step": 4329
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7818223088230213e-06,
+      "loss": 0.474,
+      "step": 4330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7780287047273937e-06,
+      "loss": 0.56,
+      "step": 4331
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7742387492472169e-06,
+      "loss": 0.5323,
+      "step": 4332
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7704524440643378e-06,
+      "loss": 0.5346,
+      "step": 4333
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7666697908589747e-06,
+      "loss": 0.5365,
+      "step": 4334
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.7628907913097416e-06,
+      "loss": 0.5664,
+      "step": 4335
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.75911544709362e-06,
+      "loss": 0.5445,
+      "step": 4336
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7553437598859712e-06,
+      "loss": 0.5617,
+      "step": 4337
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7515757313605352e-06,
+      "loss": 0.5837,
+      "step": 4338
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7478113631894244e-06,
+      "loss": 0.5338,
+      "step": 4339
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.744050657043137e-06,
+      "loss": 0.5079,
+      "step": 4340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7402936145905347e-06,
+      "loss": 0.5499,
+      "step": 4341
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7365402374988583e-06,
+      "loss": 0.5396,
+      "step": 4342
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.732790527433723e-06,
+      "loss": 0.5101,
+      "step": 4343
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.729044486059114e-06,
+      "loss": 0.4956,
+      "step": 4344
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7253021150373905e-06,
+      "loss": 0.5159,
+      "step": 4345
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7215634160292826e-06,
+      "loss": 0.5435,
+      "step": 4346
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7178283906938898e-06,
+      "loss": 0.5527,
+      "step": 4347
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7140970406886802e-06,
+      "loss": 0.5233,
+      "step": 4348
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7103693676694977e-06,
+      "loss": 0.5335,
+      "step": 4349
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7066453732905497e-06,
+      "loss": 0.4982,
+      "step": 4350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.7029250592044032e-06,
+      "loss": 0.5095,
+      "step": 4351
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.699208427062008e-06,
+      "loss": 0.5613,
+      "step": 4352
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6954954785126665e-06,
+      "loss": 0.5378,
+      "step": 4353
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6917862152040599e-06,
+      "loss": 0.5162,
+      "step": 4354
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.688080638782218e-06,
+      "loss": 0.5166,
+      "step": 4355
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6843787508915421e-06,
+      "loss": 0.5136,
+      "step": 4356
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6806805531748038e-06,
+      "loss": 0.5947,
+      "step": 4357
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6769860472731259e-06,
+      "loss": 0.5506,
+      "step": 4358
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.673295234825998e-06,
+      "loss": 0.5687,
+      "step": 4359
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6696081174712696e-06,
+      "loss": 0.529,
+      "step": 4360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6659246968451526e-06,
+      "loss": 0.5645,
+      "step": 4361
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6622449745822145e-06,
+      "loss": 0.5127,
+      "step": 4362
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6585689523153859e-06,
+      "loss": 0.5178,
+      "step": 4363
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6548966316759518e-06,
+      "loss": 0.5671,
+      "step": 4364
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6512280142935545e-06,
+      "loss": 0.5355,
+      "step": 4365
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6475631017962002e-06,
+      "loss": 0.5672,
+      "step": 4366
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6439018958102404e-06,
+      "loss": 0.5518,
+      "step": 4367
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6402443979603899e-06,
+      "loss": 0.5709,
+      "step": 4368
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6365906098697127e-06,
+      "loss": 0.5048,
+      "step": 4369
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.632940533159626e-06,
+      "loss": 0.5386,
+      "step": 4370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.629294169449912e-06,
+      "loss": 0.5566,
+      "step": 4371
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6256515203586876e-06,
+      "loss": 0.492,
+      "step": 4372
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6220125875024296e-06,
+      "loss": 0.5286,
+      "step": 4373
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6183773724959695e-06,
+      "loss": 0.5046,
+      "step": 4374
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6147458769524838e-06,
+      "loss": 0.5515,
+      "step": 4375
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6111181024834999e-06,
+      "loss": 0.5435,
+      "step": 4376
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6074940506988933e-06,
+      "loss": 0.5206,
+      "step": 4377
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6038737232068879e-06,
+      "loss": 0.5186,
+      "step": 4378
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6002571216140529e-06,
+      "loss": 0.5274,
+      "step": 4379
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.596644247525313e-06,
+      "loss": 0.5787,
+      "step": 4380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5930351025439227e-06,
+      "loss": 0.4961,
+      "step": 4381
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5894296882714977e-06,
+      "loss": 0.5383,
+      "step": 4382
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5858280063079889e-06,
+      "loss": 0.5308,
+      "step": 4383
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5822300582516948e-06,
+      "loss": 0.5395,
+      "step": 4384
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5786358456992523e-06,
+      "loss": 0.5333,
+      "step": 4385
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.575045370245647e-06,
+      "loss": 0.5392,
+      "step": 4386
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5714586334841986e-06,
+      "loss": 0.5363,
+      "step": 4387
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5678756370065794e-06,
+      "loss": 0.5522,
+      "step": 4388
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.564296382402789e-06,
+      "loss": 0.5264,
+      "step": 4389
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5607208712611698e-06,
+      "loss": 0.5436,
+      "step": 4390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5571491051684107e-06,
+      "loss": 0.4924,
+      "step": 4391
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5535810857095313e-06,
+      "loss": 0.5723,
+      "step": 4392
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5500168144678896e-06,
+      "loss": 0.55,
+      "step": 4393
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5464562930251813e-06,
+      "loss": 0.5111,
+      "step": 4394
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5428995229614375e-06,
+      "loss": 0.5073,
+      "step": 4395
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.539346505855026e-06,
+      "loss": 0.5354,
+      "step": 4396
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5357972432826507e-06,
+      "loss": 0.5272,
+      "step": 4397
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5322517368193402e-06,
+      "loss": 0.4855,
+      "step": 4398
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.528709988038468e-06,
+      "loss": 0.5377,
+      "step": 4399
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.525171998511733e-06,
+      "loss": 0.5354,
+      "step": 4400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.521637769809169e-06,
+      "loss": 0.5803,
+      "step": 4401
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.518107303499139e-06,
+      "loss": 0.5161,
+      "step": 4402
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5145806011483378e-06,
+      "loss": 0.5563,
+      "step": 4403
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5110576643217857e-06,
+      "loss": 0.5103,
+      "step": 4404
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.507538494582842e-06,
+      "loss": 0.5382,
+      "step": 4405
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5040230934931832e-06,
+      "loss": 0.5575,
+      "step": 4406
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5005114626128158e-06,
+      "loss": 0.5364,
+      "step": 4407
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4970036035000811e-06,
+      "loss": 0.5571,
+      "step": 4408
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4934995177116386e-06,
+      "loss": 0.5329,
+      "step": 4409
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.489999206802475e-06,
+      "loss": 0.5598,
+      "step": 4410
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4865026723259023e-06,
+      "loss": 0.5407,
+      "step": 4411
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4830099158335565e-06,
+      "loss": 0.4787,
+      "step": 4412
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4795209388753996e-06,
+      "loss": 0.5213,
+      "step": 4413
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4760357429997163e-06,
+      "loss": 0.51,
+      "step": 4414
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.472554329753103e-06,
+      "loss": 0.5351,
+      "step": 4415
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.469076700680494e-06,
+      "loss": 0.5332,
+      "step": 4416
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4656028573251324e-06,
+      "loss": 0.5322,
+      "step": 4417
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4621328012285852e-06,
+      "loss": 0.563,
+      "step": 4418
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4586665339307392e-06,
+      "loss": 0.5368,
+      "step": 4419
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4552040569698e-06,
+      "loss": 0.5207,
+      "step": 4420
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.451745371882287e-06,
+      "loss": 0.5326,
+      "step": 4421
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4482904802030452e-06,
+      "loss": 0.5425,
+      "step": 4422
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4448393834652296e-06,
+      "loss": 0.6016,
+      "step": 4423
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4413920832003125e-06,
+      "loss": 0.5443,
+      "step": 4424
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4379485809380823e-06,
+      "loss": 0.5296,
+      "step": 4425
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4345088782066407e-06,
+      "loss": 0.5362,
+      "step": 4426
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4310729765324038e-06,
+      "loss": 0.4699,
+      "step": 4427
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.427640877440104e-06,
+      "loss": 0.5237,
+      "step": 4428
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4242125824527787e-06,
+      "loss": 0.512,
+      "step": 4429
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4207880930917872e-06,
+      "loss": 0.5235,
+      "step": 4430
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4173674108767955e-06,
+      "loss": 0.5606,
+      "step": 4431
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4139505373257722e-06,
+      "loss": 0.5571,
+      "step": 4432
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4105374739550093e-06,
+      "loss": 0.4857,
+      "step": 4433
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4071282222790983e-06,
+      "loss": 0.5131,
+      "step": 4434
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.403722783810949e-06,
+      "loss": 0.534,
+      "step": 4435
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.400321160061765e-06,
+      "loss": 0.5603,
+      "step": 4436
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3969233525410653e-06,
+      "loss": 0.5304,
+      "step": 4437
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3935293627566804e-06,
+      "loss": 0.551,
+      "step": 4438
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3901391922147367e-06,
+      "loss": 0.4847,
+      "step": 4439
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3867528424196708e-06,
+      "loss": 0.5213,
+      "step": 4440
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.383370314874225e-06,
+      "loss": 0.5177,
+      "step": 4441
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3799916110794397e-06,
+      "loss": 0.5564,
+      "step": 4442
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3766167325346635e-06,
+      "loss": 0.5891,
+      "step": 4443
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3732456807375515e-06,
+      "loss": 0.5119,
+      "step": 4444
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3698784571840485e-06,
+      "loss": 0.512,
+      "step": 4445
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3665150633684077e-06,
+      "loss": 0.5106,
+      "step": 4446
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3631555007831875e-06,
+      "loss": 0.5564,
+      "step": 4447
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.359799770919238e-06,
+      "loss": 0.5287,
+      "step": 4448
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3564478752657106e-06,
+      "loss": 0.5378,
+      "step": 4449
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3530998153100584e-06,
+      "loss": 0.5628,
+      "step": 4450
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3497555925380256e-06,
+      "loss": 0.5394,
+      "step": 4451
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3464152084336667e-06,
+      "loss": 0.5625,
+      "step": 4452
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3430786644793158e-06,
+      "loss": 0.4894,
+      "step": 4453
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.5151,
+      "step": 4454
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3364171029414951e-06,
+      "loss": 0.5376,
+      "step": 4455
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3330920883141885e-06,
+      "loss": 0.5683,
+      "step": 4456
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.329770919749216e-06,
+      "loss": 0.5425,
+      "step": 4457
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3264535987203909e-06,
+      "loss": 0.5622,
+      "step": 4458
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3231401266998211e-06,
+      "loss": 0.5491,
+      "step": 4459
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3198305051579063e-06,
+      "loss": 0.4679,
+      "step": 4460
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3165247355633437e-06,
+      "loss": 0.5523,
+      "step": 4461
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3132228193831054e-06,
+      "loss": 0.4999,
+      "step": 4462
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3099247580824692e-06,
+      "loss": 0.4975,
+      "step": 4463
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3066305531249957e-06,
+      "loss": 0.5139,
+      "step": 4464
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.303340205972533e-06,
+      "loss": 0.5154,
+      "step": 4465
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3000537180852213e-06,
+      "loss": 0.5872,
+      "step": 4466
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2967710909214837e-06,
+      "loss": 0.5296,
+      "step": 4467
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2934923259380305e-06,
+      "loss": 0.5476,
+      "step": 4468
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2902174245898668e-06,
+      "loss": 0.5005,
+      "step": 4469
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.286946388330269e-06,
+      "loss": 0.5519,
+      "step": 4470
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.283679218610806e-06,
+      "loss": 0.5479,
+      "step": 4471
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2804159168813346e-06,
+      "loss": 0.5487,
+      "step": 4472
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2771564845899887e-06,
+      "loss": 0.5147,
+      "step": 4473
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2739009231831866e-06,
+      "loss": 0.5121,
+      "step": 4474
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2706492341056297e-06,
+      "loss": 0.5444,
+      "step": 4475
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2674014188003002e-06,
+      "loss": 0.5203,
+      "step": 4476
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2641574787084643e-06,
+      "loss": 0.5634,
+      "step": 4477
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2609174152696657e-06,
+      "loss": 0.529,
+      "step": 4478
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2576812299217246e-06,
+      "loss": 0.5743,
+      "step": 4479
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2544489241007463e-06,
+      "loss": 0.591,
+      "step": 4480
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2512204992411125e-06,
+      "loss": 0.5713,
+      "step": 4481
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2479959567754819e-06,
+      "loss": 0.5274,
+      "step": 4482
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2447752981347893e-06,
+      "loss": 0.5217,
+      "step": 4483
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2415585247482498e-06,
+      "loss": 0.5426,
+      "step": 4484
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2383456380433467e-06,
+      "loss": 0.5196,
+      "step": 4485
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2351366394458496e-06,
+      "loss": 0.5346,
+      "step": 4486
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2319315303797963e-06,
+      "loss": 0.549,
+      "step": 4487
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2287303122674933e-06,
+      "loss": 0.5257,
+      "step": 4488
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2255329865295307e-06,
+      "loss": 0.6004,
+      "step": 4489
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2223395545847672e-06,
+      "loss": 0.5329,
+      "step": 4490
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2191500178503323e-06,
+      "loss": 0.5399,
+      "step": 4491
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2159643777416275e-06,
+      "loss": 0.5195,
+      "step": 4492
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2127826356723227e-06,
+      "loss": 0.5438,
+      "step": 4493
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2096047930543653e-06,
+      "loss": 0.5811,
+      "step": 4494
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2064308512979684e-06,
+      "loss": 0.5264,
+      "step": 4495
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2032608118116062e-06,
+      "loss": 0.5484,
+      "step": 4496
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.200094676002036e-06,
+      "loss": 0.5249,
+      "step": 4497
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1969324452742714e-06,
+      "loss": 0.5304,
+      "step": 4498
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1937741210315968e-06,
+      "loss": 0.5222,
+      "step": 4499
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.190619704675564e-06,
+      "loss": 0.539,
+      "step": 4500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1874691976059894e-06,
+      "loss": 0.5748,
+      "step": 4501
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.184322601220953e-06,
+      "loss": 0.5496,
+      "step": 4502
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1811799169168037e-06,
+      "loss": 0.5492,
+      "step": 4503
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1780411460881524e-06,
+      "loss": 0.5586,
+      "step": 4504
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1749062901278708e-06,
+      "loss": 0.5683,
+      "step": 4505
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1717753504270957e-06,
+      "loss": 0.5366,
+      "step": 4506
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.168648328375225e-06,
+      "loss": 0.5213,
+      "step": 4507
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1655252253599225e-06,
+      "loss": 0.5279,
+      "step": 4508
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1624060427671046e-06,
+      "loss": 0.5162,
+      "step": 4509
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1592907819809529e-06,
+      "loss": 0.5214,
+      "step": 4510
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1561794443839103e-06,
+      "loss": 0.5426,
+      "step": 4511
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1530720313566768e-06,
+      "loss": 0.5442,
+      "step": 4512
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1499685442782093e-06,
+      "loss": 0.4796,
+      "step": 4513
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1468689845257243e-06,
+      "loss": 0.5725,
+      "step": 4514
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1437733534746931e-06,
+      "loss": 0.4916,
+      "step": 4515
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1406816524988517e-06,
+      "loss": 0.575,
+      "step": 4516
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1375938829701794e-06,
+      "loss": 0.5313,
+      "step": 4517
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1345100462589176e-06,
+      "loss": 0.5324,
+      "step": 4518
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.131430143733565e-06,
+      "loss": 0.5428,
+      "step": 4519
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.128354176760873e-06,
+      "loss": 0.5288,
+      "step": 4520
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1252821467058417e-06,
+      "loss": 0.587,
+      "step": 4521
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1222140549317294e-06,
+      "loss": 0.5238,
+      "step": 4522
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1191499028000453e-06,
+      "loss": 0.5171,
+      "step": 4523
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1160896916705466e-06,
+      "loss": 0.5473,
+      "step": 4524
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1130334229012518e-06,
+      "loss": 0.5608,
+      "step": 4525
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1099810978484182e-06,
+      "loss": 0.5596,
+      "step": 4526
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1069327178665568e-06,
+      "loss": 0.5448,
+      "step": 4527
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.103888284308433e-06,
+      "loss": 0.5168,
+      "step": 4528
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1008477985250555e-06,
+      "loss": 0.5916,
+      "step": 4529
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0978112618656823e-06,
+      "loss": 0.5203,
+      "step": 4530
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.094778675677819e-06,
+      "loss": 0.5778,
+      "step": 4531
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.091750041307218e-06,
+      "loss": 0.5781,
+      "step": 4532
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.088725360097882e-06,
+      "loss": 0.5986,
+      "step": 4533
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0857046333920518e-06,
+      "loss": 0.5693,
+      "step": 4534
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0826878625302173e-06,
+      "loss": 0.5309,
+      "step": 4535
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0796750488511154e-06,
+      "loss": 0.5808,
+      "step": 4536
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0766661936917245e-06,
+      "loss": 0.5598,
+      "step": 4537
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0736612983872651e-06,
+      "loss": 0.5482,
+      "step": 4538
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0706603642712021e-06,
+      "loss": 0.5393,
+      "step": 4539
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0676633926752423e-06,
+      "loss": 0.5435,
+      "step": 4540
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0646703849293316e-06,
+      "loss": 0.5485,
+      "step": 4541
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0616813423616668e-06,
+      "loss": 0.5267,
+      "step": 4542
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0586962662986678e-06,
+      "loss": 0.5714,
+      "step": 4543
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0557151580650105e-06,
+      "loss": 0.4806,
+      "step": 4544
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0527380189836023e-06,
+      "loss": 0.5465,
+      "step": 4545
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.0497648503755908e-06,
+      "loss": 0.5747,
+      "step": 4546
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0467956535603596e-06,
+      "loss": 0.5454,
+      "step": 4547
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0438304298555336e-06,
+      "loss": 0.5558,
+      "step": 4548
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.04086918057697e-06,
+      "loss": 0.5434,
+      "step": 4549
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0379119070387678e-06,
+      "loss": 0.5246,
+      "step": 4550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0349586105532606e-06,
+      "loss": 0.5335,
+      "step": 4551
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0320092924310066e-06,
+      "loss": 0.5207,
+      "step": 4552
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0290639539808157e-06,
+      "loss": 0.5467,
+      "step": 4553
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0261225965097199e-06,
+      "loss": 0.5345,
+      "step": 4554
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0231852213229888e-06,
+      "loss": 0.5641,
+      "step": 4555
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0202518297241237e-06,
+      "loss": 0.539,
+      "step": 4556
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.017322423014856e-06,
+      "loss": 0.4816,
+      "step": 4557
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0143970024951555e-06,
+      "loss": 0.5377,
+      "step": 4558
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0114755694632194e-06,
+      "loss": 0.4809,
+      "step": 4559
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0085581252154674e-06,
+      "loss": 0.5151,
+      "step": 4560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0056446710465628e-06,
+      "loss": 0.5031,
+      "step": 4561
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.0027352082493902e-06,
+      "loss": 0.5567,
+      "step": 4562
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.998297381150635e-07,
+      "loss": 0.5529,
+      "step": 4563
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.969282619329279e-07,
+      "loss": 0.5218,
+      "step": 4564
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.940307809905536e-07,
+      "loss": 0.558,
+      "step": 4565
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.911372965737364e-07,
+      "loss": 0.5632,
+      "step": 4566
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.882478099665048e-07,
+      "loss": 0.5158,
+      "step": 4567
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.853623224511078e-07,
+      "loss": 0.51,
+      "step": 4568
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.824808353080173e-07,
+      "loss": 0.5754,
+      "step": 4569
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.796033498159385e-07,
+      "loss": 0.5638,
+      "step": 4570
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.767298672517921e-07,
+      "loss": 0.5784,
+      "step": 4571
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.738603888907329e-07,
+      "loss": 0.5173,
+      "step": 4572
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.70994916006125e-07,
+      "loss": 0.5289,
+      "step": 4573
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.681334498695649e-07,
+      "loss": 0.5572,
+      "step": 4574
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.652759917508703e-07,
+      "loss": 0.5796,
+      "step": 4575
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.624225429180767e-07,
+      "loss": 0.5412,
+      "step": 4576
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.595731046374424e-07,
+      "loss": 0.5685,
+      "step": 4577
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.567276781734458e-07,
+      "loss": 0.5393,
+      "step": 4578
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.538862647887848e-07,
+      "loss": 0.5499,
+      "step": 4579
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.510488657443773e-07,
+      "loss": 0.5614,
+      "step": 4580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.482154822993583e-07,
+      "loss": 0.5137,
+      "step": 4581
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.453861157110811e-07,
+      "loss": 0.5231,
+      "step": 4582
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.425607672351167e-07,
+      "loss": 0.5358,
+      "step": 4583
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.397394381252556e-07,
+      "loss": 0.5346,
+      "step": 4584
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.369221296335007e-07,
+      "loss": 0.5271,
+      "step": 4585
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.341088430100719e-07,
+      "loss": 0.5344,
+      "step": 4586
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.312995795034063e-07,
+      "loss": 0.4883,
+      "step": 4587
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.28494340360151e-07,
+      "loss": 0.5018,
+      "step": 4588
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.256931268251756e-07,
+      "loss": 0.5541,
+      "step": 4589
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.228959401415527e-07,
+      "loss": 0.5765,
+      "step": 4590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.201027815505737e-07,
+      "loss": 0.5058,
+      "step": 4591
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.173136522917458e-07,
+      "loss": 0.5316,
+      "step": 4592
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.145285536027815e-07,
+      "loss": 0.5431,
+      "step": 4593
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.117474867196086e-07,
+      "loss": 0.4954,
+      "step": 4594
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.089704528763655e-07,
+      "loss": 0.5576,
+      "step": 4595
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.061974533053974e-07,
+      "loss": 0.5971,
+      "step": 4596
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.034284892372636e-07,
+      "loss": 0.5484,
+      "step": 4597
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.006635619007309e-07,
+      "loss": 0.4864,
+      "step": 4598
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.979026725227724e-07,
+      "loss": 0.5389,
+      "step": 4599
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.951458223285747e-07,
+      "loss": 0.5113,
+      "step": 4600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.923930125415292e-07,
+      "loss": 0.5557,
+      "step": 4601
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.89644244383231e-07,
+      "loss": 0.6069,
+      "step": 4602
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.868995190734875e-07,
+      "loss": 0.5611,
+      "step": 4603
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.841588378303067e-07,
+      "loss": 0.5199,
+      "step": 4604
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.814222018699047e-07,
+      "loss": 0.5348,
+      "step": 4605
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.786896124067069e-07,
+      "loss": 0.5626,
+      "step": 4606
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.759610706533317e-07,
+      "loss": 0.5397,
+      "step": 4607
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.732365778206097e-07,
+      "loss": 0.5618,
+      "step": 4608
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.705161351175762e-07,
+      "loss": 0.5991,
+      "step": 4609
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.67799743751463e-07,
+      "loss": 0.4942,
+      "step": 4610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.65087404927708e-07,
+      "loss": 0.5976,
+      "step": 4611
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.623791198499499e-07,
+      "loss": 0.5119,
+      "step": 4612
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.596748897200247e-07,
+      "loss": 0.5704,
+      "step": 4613
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.569747157379782e-07,
+      "loss": 0.5326,
+      "step": 4614
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.542785991020508e-07,
+      "loss": 0.5843,
+      "step": 4615
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.515865410086744e-07,
+      "loss": 0.5064,
+      "step": 4616
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.488985426524954e-07,
+      "loss": 0.5346,
+      "step": 4617
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.462146052263487e-07,
+      "loss": 0.5857,
+      "step": 4618
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.435347299212681e-07,
+      "loss": 0.5093,
+      "step": 4619
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.408589179264859e-07,
+      "loss": 0.5684,
+      "step": 4620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.381871704294331e-07,
+      "loss": 0.5176,
+      "step": 4621
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.355194886157326e-07,
+      "loss": 0.5307,
+      "step": 4622
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.328558736692116e-07,
+      "loss": 0.5512,
+      "step": 4623
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.301963267718783e-07,
+      "loss": 0.5164,
+      "step": 4624
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.275408491039493e-07,
+      "loss": 0.5521,
+      "step": 4625
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.248894418438303e-07,
+      "loss": 0.5471,
+      "step": 4626
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.222421061681185e-07,
+      "loss": 0.6081,
+      "step": 4627
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.195988432516077e-07,
+      "loss": 0.5609,
+      "step": 4628
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.169596542672831e-07,
+      "loss": 0.564,
+      "step": 4629
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.143245403863187e-07,
+      "loss": 0.5327,
+      "step": 4630
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.116935027780893e-07,
+      "loss": 0.5832,
+      "step": 4631
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.090665426101518e-07,
+      "loss": 0.5218,
+      "step": 4632
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.06443661048254e-07,
+      "loss": 0.4863,
+      "step": 4633
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.038248592563414e-07,
+      "loss": 0.5368,
+      "step": 4634
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.012101383965409e-07,
+      "loss": 0.5798,
+      "step": 4635
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.985994996291735e-07,
+      "loss": 0.5549,
+      "step": 4636
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.959929441127456e-07,
+      "loss": 0.5297,
+      "step": 4637
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.93390473003951e-07,
+      "loss": 0.5924,
+      "step": 4638
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.907920874576769e-07,
+      "loss": 0.5723,
+      "step": 4639
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.881977886269931e-07,
+      "loss": 0.5085,
+      "step": 4640
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.856075776631533e-07,
+      "loss": 0.5204,
+      "step": 4641
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.830214557156024e-07,
+      "loss": 0.5587,
+      "step": 4642
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.804394239319668e-07,
+      "loss": 0.5106,
+      "step": 4643
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.778614834580612e-07,
+      "loss": 0.5332,
+      "step": 4644
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.752876354378814e-07,
+      "loss": 0.5464,
+      "step": 4645
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.727178810136094e-07,
+      "loss": 0.536,
+      "step": 4646
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.701522213256085e-07,
+      "loss": 0.515,
+      "step": 4647
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.675906575124293e-07,
+      "loss": 0.5906,
+      "step": 4648
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.65033190710801e-07,
+      "loss": 0.5297,
+      "step": 4649
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.624798220556307e-07,
+      "loss": 0.5166,
+      "step": 4650
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.599305526800183e-07,
+      "loss": 0.604,
+      "step": 4651
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.573853837152323e-07,
+      "loss": 0.5538,
+      "step": 4652
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.548443162907348e-07,
+      "loss": 0.5137,
+      "step": 4653
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.523073515341528e-07,
+      "loss": 0.5875,
+      "step": 4654
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.497744905713012e-07,
+      "loss": 0.571,
+      "step": 4655
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.472457345261763e-07,
+      "loss": 0.5237,
+      "step": 4656
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.447210845209463e-07,
+      "loss": 0.5494,
+      "step": 4657
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.422005416759614e-07,
+      "loss": 0.5283,
+      "step": 4658
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.396841071097483e-07,
+      "loss": 0.5689,
+      "step": 4659
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.37171781939009e-07,
+      "loss": 0.5065,
+      "step": 4660
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.34663567278624e-07,
+      "loss": 0.5673,
+      "step": 4661
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.321594642416475e-07,
+      "loss": 0.5135,
+      "step": 4662
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.296594739393126e-07,
+      "loss": 0.5176,
+      "step": 4663
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.27163597481022e-07,
+      "loss": 0.5393,
+      "step": 4664
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.246718359743598e-07,
+      "loss": 0.5792,
+      "step": 4665
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.221841905250781e-07,
+      "loss": 0.5274,
+      "step": 4666
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.19700662237105e-07,
+      "loss": 0.5943,
+      "step": 4667
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.17221252212541e-07,
+      "loss": 0.5136,
+      "step": 4668
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.147459615516572e-07,
+      "loss": 0.5264,
+      "step": 4669
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.122747913529049e-07,
+      "loss": 0.5313,
+      "step": 4670
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.098077427128947e-07,
+      "loss": 0.5429,
+      "step": 4671
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.073448167264141e-07,
+      "loss": 0.5538,
+      "step": 4672
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.048860144864234e-07,
+      "loss": 0.512,
+      "step": 4673
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.024313370840518e-07,
+      "loss": 0.5466,
+      "step": 4674
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.999807856085949e-07,
+      "loss": 0.5399,
+      "step": 4675
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.975343611475194e-07,
+      "loss": 0.5391,
+      "step": 4676
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.950920647864612e-07,
+      "loss": 0.5327,
+      "step": 4677
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.926538976092223e-07,
+      "loss": 0.5106,
+      "step": 4678
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.902198606977783e-07,
+      "loss": 0.5551,
+      "step": 4679
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.877899551322609e-07,
+      "loss": 0.4639,
+      "step": 4680
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.853641819909784e-07,
+      "loss": 0.5038,
+      "step": 4681
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.829425423504022e-07,
+      "loss": 0.5574,
+      "step": 4682
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.805250372851679e-07,
+      "loss": 0.5493,
+      "step": 4683
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.781116678680777e-07,
+      "loss": 0.5331,
+      "step": 4684
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.757024351700969e-07,
+      "loss": 0.6175,
+      "step": 4685
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.732973402603571e-07,
+      "loss": 0.5288,
+      "step": 4686
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.708963842061566e-07,
+      "loss": 0.5567,
+      "step": 4687
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.684995680729489e-07,
+      "loss": 0.535,
+      "step": 4688
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.661068929243541e-07,
+      "loss": 0.5198,
+      "step": 4689
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.637183598221597e-07,
+      "loss": 0.537,
+      "step": 4690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.613339698263088e-07,
+      "loss": 0.5442,
+      "step": 4691
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.589537239949073e-07,
+      "loss": 0.5444,
+      "step": 4692
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.565776233842247e-07,
+      "loss": 0.5589,
+      "step": 4693
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.542056690486853e-07,
+      "loss": 0.4823,
+      "step": 4694
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.518378620408828e-07,
+      "loss": 0.5521,
+      "step": 4695
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.49474203411562e-07,
+      "loss": 0.5555,
+      "step": 4696
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.471146942096274e-07,
+      "loss": 0.525,
+      "step": 4697
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.44759335482148e-07,
+      "loss": 0.5213,
+      "step": 4698
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.424081282743455e-07,
+      "loss": 0.5949,
+      "step": 4699
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.40061073629602e-07,
+      "loss": 0.5715,
+      "step": 4700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.377181725894554e-07,
+      "loss": 0.501,
+      "step": 4701
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.353794261936008e-07,
+      "loss": 0.5356,
+      "step": 4702
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.330448354798901e-07,
+      "loss": 0.5307,
+      "step": 4703
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.307144014843314e-07,
+      "loss": 0.5231,
+      "step": 4704
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.283881252410873e-07,
+      "loss": 0.5592,
+      "step": 4705
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.260660077824754e-07,
+      "loss": 0.5298,
+      "step": 4706
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.237480501389681e-07,
+      "loss": 0.5205,
+      "step": 4707
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.214342533391926e-07,
+      "loss": 0.5171,
+      "step": 4708
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.191246184099264e-07,
+      "loss": 0.5018,
+      "step": 4709
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.168191463761052e-07,
+      "loss": 0.552,
+      "step": 4710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.145178382608108e-07,
+      "loss": 0.518,
+      "step": 4711
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.12220695085286e-07,
+      "loss": 0.5749,
+      "step": 4712
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.099277178689189e-07,
+      "loss": 0.5693,
+      "step": 4713
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.076389076292466e-07,
+      "loss": 0.5178,
+      "step": 4714
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.053542653819666e-07,
+      "loss": 0.5685,
+      "step": 4715
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.5359,
+      "step": 4716
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.007974889180912e-07,
+      "loss": 0.5714,
+      "step": 4717
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.985253567236305e-07,
+      "loss": 0.4968,
+      "step": 4718
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.962573965658247e-07,
+      "loss": 0.5413,
+      "step": 4719
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.939936094511156e-07,
+      "loss": 0.4975,
+      "step": 4720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.917339963840896e-07,
+      "loss": 0.5,
+      "step": 4721
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.894785583674812e-07,
+      "loss": 0.5225,
+      "step": 4722
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.872272964021719e-07,
+      "loss": 0.5442,
+      "step": 4723
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.849802114871927e-07,
+      "loss": 0.5402,
+      "step": 4724
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.827373046197171e-07,
+      "loss": 0.5465,
+      "step": 4725
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.804985767950689e-07,
+      "loss": 0.5305,
+      "step": 4726
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.782640290067132e-07,
+      "loss": 0.5027,
+      "step": 4727
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.760336622462615e-07,
+      "loss": 0.5674,
+      "step": 4728
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.73807477503473e-07,
+      "loss": 0.5198,
+      "step": 4729
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.715854757662487e-07,
+      "loss": 0.5224,
+      "step": 4730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.693676580206286e-07,
+      "loss": 0.5415,
+      "step": 4731
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.671540252508056e-07,
+      "loss": 0.5315,
+      "step": 4732
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.649445784391061e-07,
+      "loss": 0.5511,
+      "step": 4733
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.62739318566009e-07,
+      "loss": 0.5431,
+      "step": 4734
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.605382466101239e-07,
+      "loss": 0.5077,
+      "step": 4735
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.583413635482082e-07,
+      "loss": 0.5914,
+      "step": 4736
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.561486703551622e-07,
+      "loss": 0.5292,
+      "step": 4737
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.539601680040219e-07,
+      "loss": 0.5282,
+      "step": 4738
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.517758574659682e-07,
+      "loss": 0.5278,
+      "step": 4739
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.495957397103191e-07,
+      "loss": 0.5733,
+      "step": 4740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.474198157045307e-07,
+      "loss": 0.5604,
+      "step": 4741
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.452480864142007e-07,
+      "loss": 0.5633,
+      "step": 4742
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.43080552803067e-07,
+      "loss": 0.5086,
+      "step": 4743
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.409172158330012e-07,
+      "loss": 0.5532,
+      "step": 4744
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.38758076464011e-07,
+      "loss": 0.5399,
+      "step": 4745
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.366031356542511e-07,
+      "loss": 0.5461,
+      "step": 4746
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.344523943600033e-07,
+      "loss": 0.5089,
+      "step": 4747
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.323058535356907e-07,
+      "loss": 0.5478,
+      "step": 4748
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.301635141338701e-07,
+      "loss": 0.5459,
+      "step": 4749
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.280253771052335e-07,
+      "loss": 0.5452,
+      "step": 4750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.25891443398614e-07,
+      "loss": 0.5571,
+      "step": 4751
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.237617139609697e-07,
+      "loss": 0.5279,
+      "step": 4752
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.216361897373989e-07,
+      "loss": 0.5224,
+      "step": 4753
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.19514871671134e-07,
+      "loss": 0.5238,
+      "step": 4754
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.173977607035407e-07,
+      "loss": 0.5349,
+      "step": 4755
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.152848577741155e-07,
+      "loss": 0.5007,
+      "step": 4756
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.131761638204879e-07,
+      "loss": 0.5731,
+      "step": 4757
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.11071679778422e-07,
+      "loss": 0.5167,
+      "step": 4758
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.089714065818108e-07,
+      "loss": 0.526,
+      "step": 4759
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.06875345162684e-07,
+      "loss": 0.5138,
+      "step": 4760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.047834964511922e-07,
+      "loss": 0.5869,
+      "step": 4761
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.026958613756272e-07,
+      "loss": 0.4833,
+      "step": 4762
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.006124408624047e-07,
+      "loss": 0.5501,
+      "step": 4763
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.985332358360728e-07,
+      "loss": 0.5692,
+      "step": 4764
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.964582472193069e-07,
+      "loss": 0.5625,
+      "step": 4765
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.943874759329126e-07,
+      "loss": 0.5369,
+      "step": 4766
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.923209228958236e-07,
+      "loss": 0.5261,
+      "step": 4767
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.902585890251032e-07,
+      "loss": 0.5324,
+      "step": 4768
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.882004752359426e-07,
+      "loss": 0.5602,
+      "step": 4769
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.861465824416533e-07,
+      "loss": 0.517,
+      "step": 4770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.840969115536831e-07,
+      "loss": 0.5662,
+      "step": 4771
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.82051463481602e-07,
+      "loss": 0.5154,
+      "step": 4772
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.800102391331052e-07,
+      "loss": 0.5251,
+      "step": 4773
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.779732394140158e-07,
+      "loss": 0.504,
+      "step": 4774
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.759404652282795e-07,
+      "loss": 0.5239,
+      "step": 4775
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.739119174779705e-07,
+      "loss": 0.5423,
+      "step": 4776
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.718875970632864e-07,
+      "loss": 0.5099,
+      "step": 4777
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6986750488254296e-07,
+      "loss": 0.5409,
+      "step": 4778
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6785164183218877e-07,
+      "loss": 0.5064,
+      "step": 4779
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.658400088067905e-07,
+      "loss": 0.5353,
+      "step": 4780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.638326066990384e-07,
+      "loss": 0.5515,
+      "step": 4781
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.618294363997444e-07,
+      "loss": 0.5185,
+      "step": 4782
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.5983049879784523e-07,
+      "loss": 0.565,
+      "step": 4783
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.5783579478039466e-07,
+      "loss": 0.6028,
+      "step": 4784
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.558453252325745e-07,
+      "loss": 0.5568,
+      "step": 4785
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.5385909103768275e-07,
+      "loss": 0.5504,
+      "step": 4786
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.518770930771366e-07,
+      "loss": 0.5586,
+      "step": 4787
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.4989933223047675e-07,
+      "loss": 0.4476,
+      "step": 4788
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.4792580937536244e-07,
+      "loss": 0.4856,
+      "step": 4789
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.4595652538757305e-07,
+      "loss": 0.5264,
+      "step": 4790
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.439914811410051e-07,
+      "loss": 0.5346,
+      "step": 4791
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.4203067750767324e-07,
+      "loss": 0.4974,
+      "step": 4792
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.400741153577137e-07,
+      "loss": 0.5907,
+      "step": 4793
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.381217955593797e-07,
+      "loss": 0.5439,
+      "step": 4794
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3617371897903716e-07,
+      "loss": 0.6088,
+      "step": 4795
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.342298864811745e-07,
+      "loss": 0.5499,
+      "step": 4796
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.32290298928395e-07,
+      "loss": 0.4932,
+      "step": 4797
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3035495718141695e-07,
+      "loss": 0.5905,
+      "step": 4798
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.284238620990766e-07,
+      "loss": 0.5422,
+      "step": 4799
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.264970145383218e-07,
+      "loss": 0.506,
+      "step": 4800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.2457441535422195e-07,
+      "loss": 0.5281,
+      "step": 4801
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.2265606539995675e-07,
+      "loss": 0.526,
+      "step": 4802
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.207419655268186e-07,
+      "loss": 0.5993,
+      "step": 4803
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.1883211658421795e-07,
+      "loss": 0.5439,
+      "step": 4804
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.169265194196781e-07,
+      "loss": 0.5334,
+      "step": 4805
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.150251748788314e-07,
+      "loss": 0.5539,
+      "step": 4806
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.131280838054308e-07,
+      "loss": 0.5307,
+      "step": 4807
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.1123524704133276e-07,
+      "loss": 0.5472,
+      "step": 4808
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0934666542651213e-07,
+      "loss": 0.5749,
+      "step": 4809
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.07462339799054e-07,
+      "loss": 0.547,
+      "step": 4810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.05582270995154e-07,
+      "loss": 0.57,
+      "step": 4811
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0370645984911916e-07,
+      "loss": 0.5407,
+      "step": 4812
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0183490719336693e-07,
+      "loss": 0.4994,
+      "step": 4813
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.999676138584241e-07,
+      "loss": 0.5899,
+      "step": 4814
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.9810458067293114e-07,
+      "loss": 0.5205,
+      "step": 4815
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.962458084636334e-07,
+      "loss": 0.5336,
+      "step": 4816
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.9439129805538544e-07,
+      "loss": 0.5883,
+      "step": 4817
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.925410502711557e-07,
+      "loss": 0.5444,
+      "step": 4818
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.9069506593201636e-07,
+      "loss": 0.5598,
+      "step": 4819
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8885334585714864e-07,
+      "loss": 0.5296,
+      "step": 4820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8701589086384216e-07,
+      "loss": 0.4872,
+      "step": 4821
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.851827017674925e-07,
+      "loss": 0.4851,
+      "step": 4822
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.833537793816022e-07,
+      "loss": 0.5485,
+      "step": 4823
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.815291245177866e-07,
+      "loss": 0.5236,
+      "step": 4824
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.797087379857556e-07,
+      "loss": 0.4952,
+      "step": 4825
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.778926205933342e-07,
+      "loss": 0.5117,
+      "step": 4826
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7608077314645e-07,
+      "loss": 0.5223,
+      "step": 4827
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.742731964491364e-07,
+      "loss": 0.5302,
+      "step": 4828
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.724698913035296e-07,
+      "loss": 0.5618,
+      "step": 4829
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.706708585098728e-07,
+      "loss": 0.5398,
+      "step": 4830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6887609886650966e-07,
+      "loss": 0.6,
+      "step": 4831
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.67085613169893e-07,
+      "loss": 0.5773,
+      "step": 4832
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.652994022145773e-07,
+      "loss": 0.5697,
+      "step": 4833
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.635174667932129e-07,
+      "loss": 0.5435,
+      "step": 4834
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.617398076965639e-07,
+      "loss": 0.5789,
+      "step": 4835
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.599664257134905e-07,
+      "loss": 0.5557,
+      "step": 4836
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.581973216309553e-07,
+      "loss": 0.5299,
+      "step": 4837
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.5643249623402263e-07,
+      "loss": 0.4942,
+      "step": 4838
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.5467195030585933e-07,
+      "loss": 0.5673,
+      "step": 4839
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.5291568462773043e-07,
+      "loss": 0.525,
+      "step": 4840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.51163699979008e-07,
+      "loss": 0.5637,
+      "step": 4841
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.494159971371547e-07,
+      "loss": 0.5267,
+      "step": 4842
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.4767257687774225e-07,
+      "loss": 0.5072,
+      "step": 4843
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.459334399744374e-07,
+      "loss": 0.5404,
+      "step": 4844
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.441985871990039e-07,
+      "loss": 0.5259,
+      "step": 4845
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.4246801932131034e-07,
+      "loss": 0.5321,
+      "step": 4846
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": 0.5245,
+      "step": 4847
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.3901974132908854e-07,
+      "loss": 0.5822,
+      "step": 4848
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.373020327447851e-07,
+      "loss": 0.5668,
+      "step": 4849
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.3558861211866465e-07,
+      "loss": 0.5139,
+      "step": 4850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.3387948021107785e-07,
+      "loss": 0.5061,
+      "step": 4851
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.321746377804791e-07,
+      "loss": 0.5412,
+      "step": 4852
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.3047408558341545e-07,
+      "loss": 0.539,
+      "step": 4853
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.28777824374531e-07,
+      "loss": 0.5426,
+      "step": 4854
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.2708585490656475e-07,
+      "loss": 0.4918,
+      "step": 4855
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.2539817793035277e-07,
+      "loss": 0.5526,
+      "step": 4856
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.23714794194826e-07,
+      "loss": 0.535,
+      "step": 4857
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.220357044470102e-07,
+      "loss": 0.5069,
+      "step": 4858
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.203609094320215e-07,
+      "loss": 0.5518,
+      "step": 4859
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1869040989307766e-07,
+      "loss": 0.5162,
+      "step": 4860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.170242065714846e-07,
+      "loss": 0.4732,
+      "step": 4861
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1536230020664413e-07,
+      "loss": 0.5216,
+      "step": 4862
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1370469153604976e-07,
+      "loss": 0.5844,
+      "step": 4863
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.120513812952886e-07,
+      "loss": 0.4976,
+      "step": 4864
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.104023702180403e-07,
+      "loss": 0.5416,
+      "step": 4865
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0875765903607857e-07,
+      "loss": 0.5201,
+      "step": 4866
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0711724847926615e-07,
+      "loss": 0.5123,
+      "step": 4867
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0548113927555634e-07,
+      "loss": 0.5947,
+      "step": 4868
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.038493321509983e-07,
+      "loss": 0.5601,
+      "step": 4869
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.022218278297273e-07,
+      "loss": 0.4899,
+      "step": 4870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0059862703397447e-07,
+      "loss": 0.5025,
+      "step": 4871
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9897973048405473e-07,
+      "loss": 0.5517,
+      "step": 4872
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.973651388983756e-07,
+      "loss": 0.5225,
+      "step": 4873
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.957548529934384e-07,
+      "loss": 0.5524,
+      "step": 4874
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9414887348382825e-07,
+      "loss": 0.524,
+      "step": 4875
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9254720108222167e-07,
+      "loss": 0.5422,
+      "step": 4876
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9094983649938234e-07,
+      "loss": 0.5408,
+      "step": 4877
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.893567804441633e-07,
+      "loss": 0.5124,
+      "step": 4878
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.8776803362350667e-07,
+      "loss": 0.544,
+      "step": 4879
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.861835967424409e-07,
+      "loss": 0.5744,
+      "step": 4880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.846034705040801e-07,
+      "loss": 0.4874,
+      "step": 4881
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.830276556096312e-07,
+      "loss": 0.5309,
+      "step": 4882
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.8145615275838147e-07,
+      "loss": 0.5621,
+      "step": 4883
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.798889626477086e-07,
+      "loss": 0.5372,
+      "step": 4884
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.783260859730752e-07,
+      "loss": 0.5713,
+      "step": 4885
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.767675234280298e-07,
+      "loss": 0.5557,
+      "step": 4886
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.7521327570420367e-07,
+      "loss": 0.5373,
+      "step": 4887
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.736633434913216e-07,
+      "loss": 0.4645,
+      "step": 4888
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.721177274771847e-07,
+      "loss": 0.5083,
+      "step": 4889
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.705764283476797e-07,
+      "loss": 0.5039,
+      "step": 4890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6903944678678295e-07,
+      "loss": 0.5477,
+      "step": 4891
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6750678347655103e-07,
+      "loss": 0.535,
+      "step": 4892
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.659784390971254e-07,
+      "loss": 0.5567,
+      "step": 4893
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.64454414326728e-07,
+      "loss": 0.5613,
+      "step": 4894
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.629347098416679e-07,
+      "loss": 0.513,
+      "step": 4895
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.614193263163356e-07,
+      "loss": 0.4794,
+      "step": 4896
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.599082644232054e-07,
+      "loss": 0.543,
+      "step": 4897
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.584015248328275e-07,
+      "loss": 0.4995,
+      "step": 4898
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.568991082138428e-07,
+      "loss": 0.5401,
+      "step": 4899
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.554010152329689e-07,
+      "loss": 0.5458,
+      "step": 4900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.539072465550052e-07,
+      "loss": 0.537,
+      "step": 4901
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5241780284283256e-07,
+      "loss": 0.5356,
+      "step": 4902
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.509326847574134e-07,
+      "loss": 0.5585,
+      "step": 4903
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4945189295778715e-07,
+      "loss": 0.5398,
+      "step": 4904
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.479754281010804e-07,
+      "loss": 0.5282,
+      "step": 4905
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.465032908424925e-07,
+      "loss": 0.4974,
+      "step": 4906
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.45035481835304e-07,
+      "loss": 0.5925,
+      "step": 4907
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.435720017308785e-07,
+      "loss": 0.5463,
+      "step": 4908
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4211285117865413e-07,
+      "loss": 0.5751,
+      "step": 4909
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.406580308261508e-07,
+      "loss": 0.5496,
+      "step": 4910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.392075413189632e-07,
+      "loss": 0.5226,
+      "step": 4911
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3776138330076858e-07,
+      "loss": 0.5206,
+      "step": 4912
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3631955741331925e-07,
+      "loss": 0.5543,
+      "step": 4913
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3488206429644667e-07,
+      "loss": 0.5725,
+      "step": 4914
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3344890458805613e-07,
+      "loss": 0.6321,
+      "step": 4915
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3202007892413447e-07,
+      "loss": 0.5482,
+      "step": 4916
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3059558793874337e-07,
+      "loss": 0.5172,
+      "step": 4917
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.291754322640194e-07,
+      "loss": 0.5362,
+      "step": 4918
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.2775961253017732e-07,
+      "loss": 0.5484,
+      "step": 4919
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.2634812936550787e-07,
+      "loss": 0.5299,
+      "step": 4920
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.2494098339637338e-07,
+      "loss": 0.5031,
+      "step": 4921
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.2353817524721988e-07,
+      "loss": 0.5189,
+      "step": 4922
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.221397055405583e-07,
+      "loss": 0.553,
+      "step": 4923
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.2074557489698224e-07,
+      "loss": 0.5306,
+      "step": 4924
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.1935578393515678e-07,
+      "loss": 0.5396,
+      "step": 4925
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.179703332718186e-07,
+      "loss": 0.5045,
+      "step": 4926
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.1658922352178367e-07,
+      "loss": 0.5243,
+      "step": 4927
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.1521245529793732e-07,
+      "loss": 0.5689,
+      "step": 4928
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.138400292112397e-07,
+      "loss": 0.5294,
+      "step": 4929
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.1247194587072585e-07,
+      "loss": 0.5451,
+      "step": 4930
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.1110820588350234e-07,
+      "loss": 0.5795,
+      "step": 4931
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.09748809854744e-07,
+      "loss": 0.5179,
+      "step": 4932
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0839375838770493e-07,
+      "loss": 0.5693,
+      "step": 4933
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0704305208370857e-07,
+      "loss": 0.571,
+      "step": 4934
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0569669154214988e-07,
+      "loss": 0.5872,
+      "step": 4935
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.043546773604943e-07,
+      "loss": 0.4789,
+      "step": 4936
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0301701013427765e-07,
+      "loss": 0.564,
+      "step": 4937
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0168369045711288e-07,
+      "loss": 0.575,
+      "step": 4938
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.003547189206767e-07,
+      "loss": 0.5253,
+      "step": 4939
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9903009611471958e-07,
+      "loss": 0.5916,
+      "step": 4940
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9770982262706239e-07,
+      "loss": 0.5609,
+      "step": 4941
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.963938990435932e-07,
+      "loss": 0.5617,
+      "step": 4942
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9508232594827148e-07,
+      "loss": 0.5502,
+      "step": 4943
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9377510392312838e-07,
+      "loss": 0.5535,
+      "step": 4944
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9247223354826095e-07,
+      "loss": 0.5821,
+      "step": 4945
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.911737154018356e-07,
+      "loss": 0.52,
+      "step": 4946
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.898795500600892e-07,
+      "loss": 0.5288,
+      "step": 4947
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8858973809732562e-07,
+      "loss": 0.5211,
+      "step": 4948
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8730428008591596e-07,
+      "loss": 0.5184,
+      "step": 4949
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8602317659630164e-07,
+      "loss": 0.5541,
+      "step": 4950
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8474642819698796e-07,
+      "loss": 0.5531,
+      "step": 4951
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.83474035454555e-07,
+      "loss": 0.5597,
+      "step": 4952
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8220599893364e-07,
+      "loss": 0.465,
+      "step": 4953
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8094231919695393e-07,
+      "loss": 0.5458,
+      "step": 4954
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7968299680527157e-07,
+      "loss": 0.5424,
+      "step": 4955
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.78428032317437e-07,
+      "loss": 0.4824,
+      "step": 4956
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7717742629035695e-07,
+      "loss": 0.5542,
+      "step": 4957
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7593117927900527e-07,
+      "loss": 0.5035,
+      "step": 4958
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.746892918364218e-07,
+      "loss": 0.5712,
+      "step": 4959
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7345176451371127e-07,
+      "loss": 0.5525,
+      "step": 4960
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7221859786004547e-07,
+      "loss": 0.5707,
+      "step": 4961
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7098979242265667e-07,
+      "loss": 0.4912,
+      "step": 4962
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6976534874684648e-07,
+      "loss": 0.5532,
+      "step": 4963
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.685452673759791e-07,
+      "loss": 0.5259,
+      "step": 4964
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6732954885148255e-07,
+      "loss": 0.5572,
+      "step": 4965
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6611819371284754e-07,
+      "loss": 0.5504,
+      "step": 4966
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6491120249763292e-07,
+      "loss": 0.5258,
+      "step": 4967
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.637085757414547e-07,
+      "loss": 0.5159,
+      "step": 4968
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6251031397799932e-07,
+      "loss": 0.529,
+      "step": 4969
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6131641773900807e-07,
+      "loss": 0.5438,
+      "step": 4970
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.601268875542916e-07,
+      "loss": 0.5255,
+      "step": 4971
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5894172395171992e-07,
+      "loss": 0.4786,
+      "step": 4972
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.577609274572267e-07,
+      "loss": 0.4998,
+      "step": 4973
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.565844985948073e-07,
+      "loss": 0.5825,
+      "step": 4974
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5541243788651737e-07,
+      "loss": 0.5236,
+      "step": 4975
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.542447458524754e-07,
+      "loss": 0.5123,
+      "step": 4976
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5308142301086348e-07,
+      "loss": 0.5556,
+      "step": 4977
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.5612,
+      "step": 4978
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5076788696794632e-07,
+      "loss": 0.5582,
+      "step": 4979
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4961767479330646e-07,
+      "loss": 0.5252,
+      "step": 4980
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.484718338644231e-07,
+      "loss": 0.5376,
+      "step": 4981
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4733036468977946e-07,
+      "loss": 0.5652,
+      "step": 4982
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4619326777591702e-07,
+      "loss": 0.5312,
+      "step": 4983
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4506054362743993e-07,
+      "loss": 0.5635,
+      "step": 4984
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4393219274700942e-07,
+      "loss": 0.5873,
+      "step": 4985
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.428082156353494e-07,
+      "loss": 0.5664,
+      "step": 4986
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4168861279123758e-07,
+      "loss": 0.5414,
+      "step": 4987
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4057338471151428e-07,
+      "loss": 0.5677,
+      "step": 4988
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3946253189107917e-07,
+      "loss": 0.5092,
+      "step": 4989
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3835605482288794e-07,
+      "loss": 0.4858,
+      "step": 4990
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3725395399795448e-07,
+      "loss": 0.5263,
+      "step": 4991
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3615622990535204e-07,
+      "loss": 0.551,
+      "step": 4992
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3506288303221094e-07,
+      "loss": 0.5019,
+      "step": 4993
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3397391386372083e-07,
+      "loss": 0.5336,
+      "step": 4994
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.328893228831263e-07,
+      "loss": 0.5424,
+      "step": 4995
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3180911057172785e-07,
+      "loss": 0.551,
+      "step": 4996
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3073327740888764e-07,
+      "loss": 0.5263,
+      "step": 4997
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2966182387201931e-07,
+      "loss": 0.5394,
+      "step": 4998
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.285947504365981e-07,
+      "loss": 0.566,
+      "step": 4999
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2753205757615184e-07,
+      "loss": 0.5173,
+      "step": 5000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2647374576226446e-07,
+      "loss": 0.5324,
+      "step": 5001
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2541981546457582e-07,
+      "loss": 0.4969,
+      "step": 5002
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2437026715078627e-07,
+      "loss": 0.498,
+      "step": 5003
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.233251012866443e-07,
+      "loss": 0.5219,
+      "step": 5004
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2228431833595788e-07,
+      "loss": 0.4976,
+      "step": 5005
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.212479187605897e-07,
+      "loss": 0.5702,
+      "step": 5006
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2021590302045528e-07,
+      "loss": 0.5148,
+      "step": 5007
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1918827157352708e-07,
+      "loss": 0.5177,
+      "step": 5008
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1816502487583148e-07,
+      "loss": 0.5589,
+      "step": 5009
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1714616338144746e-07,
+      "loss": 0.5477,
+      "step": 5010
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1613168754250892e-07,
+      "loss": 0.5815,
+      "step": 5011
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1512159780920684e-07,
+      "loss": 0.5336,
+      "step": 5012
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1411589462977823e-07,
+      "loss": 0.5514,
+      "step": 5013
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1311457845052165e-07,
+      "loss": 0.534,
+      "step": 5014
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1211764971578277e-07,
+      "loss": 0.5295,
+      "step": 5015
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1112510886796435e-07,
+      "loss": 0.5301,
+      "step": 5016
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1013695634751965e-07,
+      "loss": 0.5875,
+      "step": 5017
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0915319259295565e-07,
+      "loss": 0.5417,
+      "step": 5018
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0817381804083093e-07,
+      "loss": 0.5654,
+      "step": 5019
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0719883312575674e-07,
+      "loss": 0.4664,
+      "step": 5020
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0622823828039696e-07,
+      "loss": 0.5377,
+      "step": 5021
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0526203393546597e-07,
+      "loss": 0.5895,
+      "step": 5022
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0430022051973188e-07,
+      "loss": 0.5525,
+      "step": 5023
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0334279846001106e-07,
+      "loss": 0.5171,
+      "step": 5024
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0238976818117364e-07,
+      "loss": 0.5685,
+      "step": 5025
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0144113010614021e-07,
+      "loss": 0.4989,
+      "step": 5026
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0049688465588292e-07,
+      "loss": 0.5308,
+      "step": 5027
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.955703224942326e-08,
+      "loss": 0.5387,
+      "step": 5028
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.862157330383537e-08,
+      "loss": 0.5822,
+      "step": 5029
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.769050823424053e-08,
+      "loss": 0.5347,
+      "step": 5030
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.676383745381268e-08,
+      "loss": 0.5365,
+      "step": 5031
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.584156137377399e-08,
+      "loss": 0.5171,
+      "step": 5032
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.492368040340038e-08,
+      "loss": 0.5393,
+      "step": 5033
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.401019495001162e-08,
+      "loss": 0.557,
+      "step": 5034
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.31011054189801e-08,
+      "loss": 0.6245,
+      "step": 5035
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.219641221372755e-08,
+      "loss": 0.498,
+      "step": 5036
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.129611573572506e-08,
+      "loss": 0.5227,
+      "step": 5037
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.040021638449081e-08,
+      "loss": 0.5332,
+      "step": 5038
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.950871455759347e-08,
+      "loss": 0.5665,
+      "step": 5039
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.862161065064878e-08,
+      "loss": 0.532,
+      "step": 5040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.773890505732186e-08,
+      "loss": 0.5124,
+      "step": 5041
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.686059816932601e-08,
+      "loss": 0.5393,
+      "step": 5042
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.598669037642282e-08,
+      "loss": 0.5271,
+      "step": 5043
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.511718206641873e-08,
+      "loss": 0.5476,
+      "step": 5044
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.425207362517284e-08,
+      "loss": 0.5056,
+      "step": 5045
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.33913654365892e-08,
+      "loss": 0.5436,
+      "step": 5046
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.253505788261673e-08,
+      "loss": 0.5897,
+      "step": 5047
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.1683151343257e-08,
+      "loss": 0.5698,
+      "step": 5048
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.083564619655316e-08,
+      "loss": 0.508,
+      "step": 5049
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.99925428185988e-08,
+      "loss": 0.5774,
+      "step": 5050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.915384158353246e-08,
+      "loss": 0.542,
+      "step": 5051
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.831954286353971e-08,
+      "loss": 0.4804,
+      "step": 5052
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.74896470288522e-08,
+      "loss": 0.5383,
+      "step": 5053
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.666415444774866e-08,
+      "loss": 0.5175,
+      "step": 5054
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.584306548655163e-08,
+      "loss": 0.4965,
+      "step": 5055
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.502638050963295e-08,
+      "loss": 0.4941,
+      "step": 5056
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.421409987940609e-08,
+      "loss": 0.5615,
+      "step": 5057
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.340622395633157e-08,
+      "loss": 0.5499,
+      "step": 5058
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.260275309891817e-08,
+      "loss": 0.5484,
+      "step": 5059
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.180368766371515e-08,
+      "loss": 0.5966,
+      "step": 5060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.100902800532105e-08,
+      "loss": 0.5174,
+      "step": 5061
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.021877447637493e-08,
+      "loss": 0.5378,
+      "step": 5062
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.943292742756513e-08,
+      "loss": 0.5579,
+      "step": 5063
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.865148720762161e-08,
+      "loss": 0.5565,
+      "step": 5064
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.787445416331917e-08,
+      "loss": 0.5654,
+      "step": 5065
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.710182863947534e-08,
+      "loss": 0.5625,
+      "step": 5066
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.633361097895696e-08,
+      "loss": 0.4559,
+      "step": 5067
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.556980152267023e-08,
+      "loss": 0.5936,
+      "step": 5068
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.481040060956512e-08,
+      "loss": 0.4945,
+      "step": 5069
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.405540857663872e-08,
+      "loss": 0.4919,
+      "step": 5070
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.330482575892638e-08,
+      "loss": 0.527,
+      "step": 5071
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.255865248951276e-08,
+      "loss": 0.5401,
+      "step": 5072
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.181688909952077e-08,
+      "loss": 0.547,
+      "step": 5073
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.107953591811821e-08,
+      "loss": 0.5452,
+      "step": 5074
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.03465932725178e-08,
+      "loss": 0.5593,
+      "step": 5075
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.961806148797156e-08,
+      "loss": 0.568,
+      "step": 5076
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.889394088777645e-08,
+      "loss": 0.4913,
+      "step": 5077
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.817423179327097e-08,
+      "loss": 0.5255,
+      "step": 5078
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.745893452383522e-08,
+      "loss": 0.5027,
+      "step": 5079
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.6748049396894154e-08,
+      "loss": 0.5652,
+      "step": 5080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.604157672791211e-08,
+      "loss": 0.5338,
+      "step": 5081
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.5339516830397176e-08,
+      "loss": 0.5712,
+      "step": 5082
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.464187001589572e-08,
+      "loss": 0.5472,
+      "step": 5083
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.394863659400229e-08,
+      "loss": 0.5515,
+      "step": 5084
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.325981687234527e-08,
+      "loss": 0.5234,
+      "step": 5085
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.257541115660014e-08,
+      "loss": 0.5565,
+      "step": 5086
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.189541975048173e-08,
+      "loss": 0.5233,
+      "step": 5087
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.121984295574423e-08,
+      "loss": 0.499,
+      "step": 5088
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.0548681072185576e-08,
+      "loss": 0.5584,
+      "step": 5089
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9881934397643106e-08,
+      "loss": 0.5451,
+      "step": 5090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.921960322799457e-08,
+      "loss": 0.5844,
+      "step": 5091
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.856168785715931e-08,
+      "loss": 0.5708,
+      "step": 5092
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.7908188577096e-08,
+      "loss": 0.5534,
+      "step": 5093
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.7259105677804893e-08,
+      "loss": 0.4705,
+      "step": 5094
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6614439447324466e-08,
+      "loss": 0.5676,
+      "step": 5095
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.597419017173588e-08,
+      "loss": 0.5605,
+      "step": 5096
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.533835813515852e-08,
+      "loss": 0.5455,
+      "step": 5097
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.470694361975003e-08,
+      "loss": 0.5677,
+      "step": 5098
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.407994690571182e-08,
+      "loss": 0.5542,
+      "step": 5099
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.345736827128133e-08,
+      "loss": 0.5066,
+      "step": 5100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.2839207992736445e-08,
+      "loss": 0.582,
+      "step": 5101
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.22254663443944e-08,
+      "loss": 0.5184,
+      "step": 5102
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.161614359861288e-08,
+      "loss": 0.5419,
+      "step": 5103
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.101124002578671e-08,
+      "loss": 0.5272,
+      "step": 5104
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.041075589434895e-08,
+      "loss": 0.5405,
+      "step": 5105
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.981469147077643e-08,
+      "loss": 0.469,
+      "step": 5106
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.922304701957758e-08,
+      "loss": 0.5201,
+      "step": 5107
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.863582280330347e-08,
+      "loss": 0.5701,
+      "step": 5108
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.805301908254455e-08,
+      "loss": 0.5105,
+      "step": 5109
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.747463611592839e-08,
+      "loss": 0.5679,
+      "step": 5110
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.690067416011855e-08,
+      "loss": 0.5198,
+      "step": 5111
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.63311334698202e-08,
+      "loss": 0.5692,
+      "step": 5112
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.576601429777449e-08,
+      "loss": 0.536,
+      "step": 5113
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.520531689476192e-08,
+      "loss": 0.5559,
+      "step": 5114
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.464904150959903e-08,
+      "loss": 0.4949,
+      "step": 5115
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.409718838914056e-08,
+      "loss": 0.5672,
+      "step": 5116
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.354975777827951e-08,
+      "loss": 0.526,
+      "step": 5117
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3006749919946015e-08,
+      "loss": 0.4974,
+      "step": 5118
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.24681650551073e-08,
+      "loss": 0.5386,
+      "step": 5119
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1934003422767754e-08,
+      "loss": 0.5282,
+      "step": 5120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.140426525996998e-08,
+      "loss": 0.5427,
+      "step": 5121
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.087895080179149e-08,
+      "loss": 0.5255,
+      "step": 5122
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0358060281350244e-08,
+      "loss": 0.5526,
+      "step": 5123
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.98415939297958e-08,
+      "loss": 0.5075,
+      "step": 5124
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.9329551976318148e-08,
+      "loss": 0.5738,
+      "step": 5125
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8821934648144424e-08,
+      "loss": 0.4839,
+      "step": 5126
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.831874217053665e-08,
+      "loss": 0.5401,
+      "step": 5127
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.781997476679288e-08,
+      "loss": 0.5416,
+      "step": 5128
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7325632658247168e-08,
+      "loss": 0.4914,
+      "step": 5129
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.683571606427293e-08,
+      "loss": 0.515,
+      "step": 5130
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6350225202277367e-08,
+      "loss": 0.5303,
+      "step": 5131
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.586916028770259e-08,
+      "loss": 0.5495,
+      "step": 5132
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5392521534027824e-08,
+      "loss": 0.5102,
+      "step": 5133
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.492030915276944e-08,
+      "loss": 0.5462,
+      "step": 5134
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4452523353477586e-08,
+      "loss": 0.5298,
+      "step": 5135
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3989164343739546e-08,
+      "loss": 0.5482,
+      "step": 5136
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3530232329176393e-08,
+      "loss": 0.5426,
+      "step": 5137
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3075727513446332e-08,
+      "loss": 0.5123,
+      "step": 5138
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2625650098241357e-08,
+      "loss": 0.5375,
+      "step": 5139
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2180000283290593e-08,
+      "loss": 0.5484,
+      "step": 5140
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1738778266356954e-08,
+      "loss": 0.5288,
+      "step": 5141
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.130198424323937e-08,
+      "loss": 0.5921,
+      "step": 5142
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0869618407771685e-08,
+      "loss": 0.5221,
+      "step": 5143
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.044168095182153e-08,
+      "loss": 0.5149,
+      "step": 5144
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.001817206529144e-08,
+      "loss": 0.6417,
+      "step": 5145
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9599091936121085e-08,
+      "loss": 0.5741,
+      "step": 5146
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9184440750281697e-08,
+      "loss": 0.554,
+      "step": 5147
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8774218691782753e-08,
+      "loss": 0.5422,
+      "step": 5148
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.836842594266419e-08,
+      "loss": 0.5496,
+      "step": 5149
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7967062683001968e-08,
+      "loss": 0.5557,
+      "step": 5150
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.757012909090694e-08,
+      "loss": 0.5274,
+      "step": 5151
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7177625342525982e-08,
+      "loss": 0.5578,
+      "step": 5152
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6789551612035325e-08,
+      "loss": 0.4846,
+      "step": 5153
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6405908071650545e-08,
+      "loss": 0.5547,
+      "step": 5154
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6026694891616566e-08,
+      "loss": 0.4958,
+      "step": 5155
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.565191224021656e-08,
+      "loss": 0.575,
+      "step": 5156
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.528156028376526e-08,
+      "loss": 0.5465,
+      "step": 5157
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4915639186610097e-08,
+      "loss": 0.5373,
+      "step": 5158
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4554149111135618e-08,
+      "loss": 0.5394,
+      "step": 5159
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4197090217755727e-08,
+      "loss": 0.4982,
+      "step": 5160
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3844462664923674e-08,
+      "loss": 0.5036,
+      "step": 5161
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3496266609119846e-08,
+      "loss": 0.4999,
+      "step": 5162
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.315250220486286e-08,
+      "loss": 0.5598,
+      "step": 5163
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2813169604701804e-08,
+      "loss": 0.5397,
+      "step": 5164
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2478268959221773e-08,
+      "loss": 0.5445,
+      "step": 5165
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2147800417038336e-08,
+      "loss": 0.5574,
+      "step": 5166
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.182176412480196e-08,
+      "loss": 0.5515,
+      "step": 5167
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.150016022719691e-08,
+      "loss": 0.5019,
+      "step": 5168
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1182988866936806e-08,
+      "loss": 0.4977,
+      "step": 5169
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.087025018477461e-08,
+      "loss": 0.5188,
+      "step": 5170
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0561944319489314e-08,
+      "loss": 0.4802,
+      "step": 5171
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.025807140789703e-08,
+      "loss": 0.4771,
+      "step": 5172
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.958631584846556e-09,
+      "loss": 0.5118,
+      "step": 5173
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.663624983217156e-09,
+      "loss": 0.5368,
+      "step": 5174
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.37305173392189e-09,
+      "loss": 0.545,
+      "step": 5175
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.086911965908718e-09,
+      "loss": 0.538,
+      "step": 5176
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.80520580615496e-09,
+      "loss": 0.5173,
+      "step": 5177
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.527933379671726e-09,
+      "loss": 0.5144,
+      "step": 5178
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.255094809502817e-09,
+      "loss": 0.5709,
+      "step": 5179
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.986690216722493e-09,
+      "loss": 0.505,
+      "step": 5180
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.72271972044214e-09,
+      "loss": 0.4979,
+      "step": 5181
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.46318343780028e-09,
+      "loss": 0.543,
+      "step": 5182
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.208081483970342e-09,
+      "loss": 0.547,
+      "step": 5183
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.957413972157323e-09,
+      "loss": 0.5628,
+      "step": 5184
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.711181013597801e-09,
+      "loss": 0.5344,
+      "step": 5185
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.469382717563255e-09,
+      "loss": 0.5389,
+      "step": 5186
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.232019191352301e-09,
+      "loss": 0.5688,
+      "step": 5187
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.999090540300678e-09,
+      "loss": 0.5613,
+      "step": 5188
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.770596867772371e-09,
+      "loss": 0.5112,
+      "step": 5189
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5465382751640486e-09,
+      "loss": 0.525,
+      "step": 5190
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.3269148619072844e-09,
+      "loss": 0.5557,
+      "step": 5191
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.111726725460786e-09,
+      "loss": 0.5543,
+      "step": 5192
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.900973961319277e-09,
+      "loss": 0.527,
+      "step": 5193
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.694656663004616e-09,
+      "loss": 0.5694,
+      "step": 5194
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4927749220757824e-09,
+      "loss": 0.6059,
+      "step": 5195
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.295328828118894e-09,
+      "loss": 0.5301,
+      "step": 5196
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.102318468753863e-09,
+      "loss": 0.5584,
+      "step": 5197
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.913743929631064e-09,
+      "loss": 0.5606,
+      "step": 5198
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7296052944346685e-09,
+      "loss": 0.5949,
+      "step": 5199
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.5499026448782006e-09,
+      "loss": 0.5504,
+      "step": 5200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.3746360607067597e-09,
+      "loss": 0.4956,
+      "step": 5201
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.2038056196981306e-09,
+      "loss": 0.5433,
+      "step": 5202
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.037411397659451e-09,
+      "loss": 0.5396,
+      "step": 5203
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.8754534684316548e-09,
+      "loss": 0.568,
+      "step": 5204
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7179319038850292e-09,
+      "loss": 0.5395,
+      "step": 5205
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.564846773922547e-09,
+      "loss": 0.5029,
+      "step": 5206
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4161981464787544e-09,
+      "loss": 0.5391,
+      "step": 5207
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.271986087516442e-09,
+      "loss": 0.5099,
+      "step": 5208
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.132210661033307e-09,
+      "loss": 0.5517,
+      "step": 5209
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9968719290552887e-09,
+      "loss": 0.547,
+      "step": 5210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.865969951643232e-09,
+      "loss": 0.543,
+      "step": 5211
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7395047868840053e-09,
+      "loss": 0.4879,
+      "step": 5212
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6174764909004936e-09,
+      "loss": 0.5096,
+      "step": 5213
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4998851178438245e-09,
+      "loss": 0.5449,
+      "step": 5214
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3867307198955903e-09,
+      "loss": 0.5701,
+      "step": 5215
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2780133472711786e-09,
+      "loss": 0.5475,
+      "step": 5216
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1737330482142206e-09,
+      "loss": 0.5441,
+      "step": 5217
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0738898690021426e-09,
+      "loss": 0.5403,
+      "step": 5218
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.78483853940615e-10,
+      "loss": 0.4891,
+      "step": 5219
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.875150453668824e-10,
+      "loss": 0.5184,
+      "step": 5220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.009834836519848e-10,
+      "loss": 0.5043,
+      "step": 5221
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.188892071929854e-10,
+      "loss": 0.5707,
+      "step": 5222
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.412322524207426e-10,
+      "loss": 0.5381,
+      "step": 5223
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.680126537976893e-10,
+      "loss": 0.5727,
+      "step": 5224
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.992304438156126e-10,
+      "loss": 0.5433,
+      "step": 5225
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.34885652997874e-10,
+      "loss": 0.5573,
+      "step": 5226
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.749783098971893e-10,
+      "loss": 0.5636,
+      "step": 5227
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.195084411000693e-10,
+      "loss": 0.5306,
+      "step": 5228
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.684760712201584e-10,
+      "loss": 0.5341,
+      "step": 5229
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.2188122290600633e-10,
+      "loss": 0.5407,
+      "step": 5230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.7972391683218626e-10,
+      "loss": 0.5105,
+      "step": 5231
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.420041717081766e-10,
+      "loss": 0.558,
+      "step": 5232
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0872200427280987e-10,
+      "loss": 0.5327,
+      "step": 5233
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.9877429295383e-11,
+      "loss": 0.5515,
+      "step": 5234
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.5470459574547e-11,
+      "loss": 0.5636,
+      "step": 5235
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.550110594385814e-11,
+      "loss": 0.5037,
+      "step": 5236
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.996937726289616e-11,
+      "loss": 0.4922,
+      "step": 5237
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.875280423925603e-12,
+      "loss": 0.5624,
+      "step": 5238
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.218820351895801e-12,
+      "loss": 0.5206,
+      "step": 5239
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.5882,
+      "step": 5240
+    },
+    {
+      "epoch": 2.0,
+      "step": 5240,
+      "total_flos": 5.227172297898459e+18,
+      "train_loss": 0.6100755021772312,
+      "train_runtime": 104912.1493,
+      "train_samples_per_second": 3.197,
+      "train_steps_per_second": 0.05
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5240,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 100000.0,
+  "total_flos": 5.227172297898459e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}