diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,40194 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 6694,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.950248756218906e-08,
+      "loss": 1.3606,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9900497512437812e-07,
+      "loss": 1.3045,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9850746268656716e-07,
+      "loss": 1.2961,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9800995024875624e-07,
+      "loss": 1.2721,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.975124378109453e-07,
+      "loss": 1.3593,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.970149253731343e-07,
+      "loss": 1.3115,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.965174129353235e-07,
+      "loss": 1.2711,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.960199004975125e-07,
+      "loss": 0.3181,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.955223880597015e-07,
+      "loss": 1.2937,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.950248756218907e-07,
+      "loss": 1.2661,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0945273631840796e-06,
+      "loss": 1.3382,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1940298507462686e-06,
+      "loss": 1.2536,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2935323383084578e-06,
+      "loss": 1.2679,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.393034825870647e-06,
+      "loss": 1.3007,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4925373134328358e-06,
+      "loss": 1.2138,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.592039800995025e-06,
+      "loss": 1.2234,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6915422885572142e-06,
+      "loss": 1.1784,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.791044776119403e-06,
+      "loss": 1.1565,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8905472636815921e-06,
+      "loss": 1.1222,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9900497512437813e-06,
+      "loss": 1.1738,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.08955223880597e-06,
+      "loss": 1.1648,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1890547263681593e-06,
+      "loss": 1.0938,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2885572139303485e-06,
+      "loss": 1.1435,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3880597014925373e-06,
+      "loss": 1.1123,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4875621890547264e-06,
+      "loss": 0.3232,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5870646766169156e-06,
+      "loss": 1.124,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.686567164179105e-06,
+      "loss": 1.1012,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.786069651741294e-06,
+      "loss": 1.1673,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.885572139303483e-06,
+      "loss": 1.0285,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9850746268656716e-06,
+      "loss": 0.3138,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0845771144278608e-06,
+      "loss": 0.9779,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.18407960199005e-06,
+      "loss": 0.96,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.283582089552239e-06,
+      "loss": 0.3216,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3830845771144283e-06,
+      "loss": 1.0646,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4825870646766175e-06,
+      "loss": 1.0382,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.582089552238806e-06,
+      "loss": 1.0102,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.681592039800995e-06,
+      "loss": 1.0203,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7810945273631843e-06,
+      "loss": 0.9679,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8805970149253735e-06,
+      "loss": 1.0153,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.980099502487563e-06,
+      "loss": 0.9807,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.079601990049751e-06,
+      "loss": 0.9862,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.17910447761194e-06,
+      "loss": 0.9919,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.278606965174129e-06,
+      "loss": 0.9817,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.378109452736319e-06,
+      "loss": 1.0082,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.477611940298508e-06,
+      "loss": 0.9557,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.577114427860697e-06,
+      "loss": 0.9361,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.676616915422886e-06,
+      "loss": 0.3052,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7761194029850745e-06,
+      "loss": 0.9273,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.875621890547264e-06,
+      "loss": 0.9096,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.975124378109453e-06,
+      "loss": 0.9665,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.074626865671642e-06,
+      "loss": 0.9244,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.174129353233831e-06,
+      "loss": 0.9766,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2736318407960205e-06,
+      "loss": 0.9334,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.37313432835821e-06,
+      "loss": 0.9471,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.472636815920398e-06,
+      "loss": 0.9232,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.572139303482588e-06,
+      "loss": 0.9666,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.671641791044776e-06,
+      "loss": 0.9858,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.771144278606966e-06,
+      "loss": 0.9552,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.870646766169155e-06,
+      "loss": 0.8953,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.970149253731343e-06,
+      "loss": 0.9543,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.069651741293533e-06,
+      "loss": 0.9927,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.1691542288557215e-06,
+      "loss": 0.3273,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2686567164179116e-06,
+      "loss": 0.9166,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.3681592039801e-06,
+      "loss": 0.9408,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.46766169154229e-06,
+      "loss": 0.862,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.567164179104478e-06,
+      "loss": 0.877,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.9992,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.766169154228857e-06,
+      "loss": 0.9527,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.865671641791045e-06,
+      "loss": 0.9683,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.965174129353235e-06,
+      "loss": 0.9617,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.064676616915423e-06,
+      "loss": 0.9075,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.164179104477612e-06,
+      "loss": 0.8986,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.263681592039802e-06,
+      "loss": 0.9124,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.36318407960199e-06,
+      "loss": 0.9219,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.46268656716418e-06,
+      "loss": 0.9927,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5621890547263685e-06,
+      "loss": 0.9603,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.661691542288557e-06,
+      "loss": 0.9477,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.761194029850747e-06,
+      "loss": 0.9299,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.860696517412935e-06,
+      "loss": 0.9536,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.960199004975125e-06,
+      "loss": 0.8829,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.059701492537314e-06,
+      "loss": 0.9059,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.159203980099502e-06,
+      "loss": 0.9583,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.258706467661692e-06,
+      "loss": 0.9307,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.35820895522388e-06,
+      "loss": 0.9032,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.45771144278607e-06,
+      "loss": 0.9969,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.557213930348259e-06,
+      "loss": 0.9841,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.656716417910447e-06,
+      "loss": 0.9389,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.756218905472637e-06,
+      "loss": 0.9294,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.855721393034826e-06,
+      "loss": 0.8854,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.955223880597016e-06,
+      "loss": 0.9261,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.054726368159204e-06,
+      "loss": 0.908,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.154228855721394e-06,
+      "loss": 0.8723,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.253731343283582e-06,
+      "loss": 0.3313,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.353233830845772e-06,
+      "loss": 0.8874,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.45273631840796e-06,
+      "loss": 0.9134,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.552238805970149e-06,
+      "loss": 0.3244,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.651741293532339e-06,
+      "loss": 0.9174,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.751243781094527e-06,
+      "loss": 0.9459,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.850746268656717e-06,
+      "loss": 0.893,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.950248756218906e-06,
+      "loss": 0.9314,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0049751243781096e-05,
+      "loss": 0.886,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0149253731343284e-05,
+      "loss": 0.9595,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0248756218905474e-05,
+      "loss": 0.8465,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0348258706467663e-05,
+      "loss": 0.8694,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0447761194029851e-05,
+      "loss": 0.9153,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0547263681592041e-05,
+      "loss": 0.2894,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.064676616915423e-05,
+      "loss": 0.9399,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.074626865671642e-05,
+      "loss": 0.9258,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0845771144278608e-05,
+      "loss": 0.9579,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0945273631840796e-05,
+      "loss": 0.8831,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1044776119402986e-05,
+      "loss": 0.9207,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1144278606965176e-05,
+      "loss": 0.9341,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1243781094527364e-05,
+      "loss": 0.9256,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1343283582089553e-05,
+      "loss": 0.8057,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1442786069651741e-05,
+      "loss": 0.8431,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1542288557213931e-05,
+      "loss": 0.928,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1641791044776121e-05,
+      "loss": 0.8919,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.174129353233831e-05,
+      "loss": 0.823,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1840796019900498e-05,
+      "loss": 0.8742,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1940298507462686e-05,
+      "loss": 0.9232,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2039800995024878e-05,
+      "loss": 0.9351,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2139303482587066e-05,
+      "loss": 0.8621,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2238805970149255e-05,
+      "loss": 0.9794,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2338308457711443e-05,
+      "loss": 0.859,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2437810945273631e-05,
+      "loss": 0.9318,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2537313432835823e-05,
+      "loss": 0.8986,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2636815920398011e-05,
+      "loss": 0.8815,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.27363184079602e-05,
+      "loss": 0.9093,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2835820895522388e-05,
+      "loss": 0.8025,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.293532338308458e-05,
+      "loss": 0.9085,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3034825870646768e-05,
+      "loss": 0.8597,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3134328358208957e-05,
+      "loss": 0.8985,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3233830845771145e-05,
+      "loss": 0.9001,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.8973,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3432835820895525e-05,
+      "loss": 0.964,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3532338308457713e-05,
+      "loss": 0.8707,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3631840796019902e-05,
+      "loss": 0.8683,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.373134328358209e-05,
+      "loss": 0.9674,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3830845771144278e-05,
+      "loss": 0.8707,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.393034825870647e-05,
+      "loss": 0.926,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4029850746268658e-05,
+      "loss": 0.9075,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4129353233830847e-05,
+      "loss": 0.8583,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4228855721393035e-05,
+      "loss": 0.807,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4328358208955224e-05,
+      "loss": 0.9084,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4427860696517415e-05,
+      "loss": 0.2973,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4527363184079604e-05,
+      "loss": 0.9142,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4626865671641792e-05,
+      "loss": 0.8766,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.472636815920398e-05,
+      "loss": 0.9239,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4825870646766169e-05,
+      "loss": 0.9494,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.492537313432836e-05,
+      "loss": 0.923,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5024875621890549e-05,
+      "loss": 0.843,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5124378109452737e-05,
+      "loss": 0.9015,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5223880597014925e-05,
+      "loss": 0.9238,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5323383084577114e-05,
+      "loss": 0.831,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5422885572139307e-05,
+      "loss": 0.8664,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5522388059701494e-05,
+      "loss": 0.2865,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5621890547263684e-05,
+      "loss": 0.8739,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.572139303482587e-05,
+      "loss": 0.8777,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.582089552238806e-05,
+      "loss": 0.903,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.592039800995025e-05,
+      "loss": 0.8855,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.601990049751244e-05,
+      "loss": 0.9174,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6119402985074627e-05,
+      "loss": 0.858,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6218905472636817e-05,
+      "loss": 0.8789,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6318407960199004e-05,
+      "loss": 0.8982,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6417910447761197e-05,
+      "loss": 0.9387,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6517412935323384e-05,
+      "loss": 0.9386,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6616915422885574e-05,
+      "loss": 0.8751,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.671641791044776e-05,
+      "loss": 0.9078,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.681592039800995e-05,
+      "loss": 0.8999,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.691542288557214e-05,
+      "loss": 0.8117,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.701492537313433e-05,
+      "loss": 0.8826,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7114427860696518e-05,
+      "loss": 0.8844,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7213930348258708e-05,
+      "loss": 0.9159,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7313432835820894e-05,
+      "loss": 0.9354,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7412935323383088e-05,
+      "loss": 0.9398,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7512437810945274e-05,
+      "loss": 0.9289,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7611940298507464e-05,
+      "loss": 0.9588,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.771144278606965e-05,
+      "loss": 0.8851,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7810945273631844e-05,
+      "loss": 0.8507,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.791044776119403e-05,
+      "loss": 0.9247,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.800995024875622e-05,
+      "loss": 0.8878,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8109452736318408e-05,
+      "loss": 0.9995,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8208955223880598e-05,
+      "loss": 0.9144,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8308457711442788e-05,
+      "loss": 0.9688,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8407960199004978e-05,
+      "loss": 0.938,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8507462686567165e-05,
+      "loss": 0.8115,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8606965174129355e-05,
+      "loss": 0.9234,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8706467661691545e-05,
+      "loss": 0.9043,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8805970149253735e-05,
+      "loss": 0.906,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.890547263681592e-05,
+      "loss": 0.905,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.900497512437811e-05,
+      "loss": 0.9079,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9104477611940298e-05,
+      "loss": 0.8712,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9203980099502488e-05,
+      "loss": 0.9036,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9303482587064678e-05,
+      "loss": 0.8844,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9402985074626868e-05,
+      "loss": 0.8656,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9502487562189055e-05,
+      "loss": 0.9382,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9601990049751245e-05,
+      "loss": 0.872,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9701492537313435e-05,
+      "loss": 0.8964,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9800995024875625e-05,
+      "loss": 0.839,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.990049751243781e-05,
+      "loss": 0.8864,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.8875,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999882947974e-05,
+      "loss": 0.8694,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999531791923e-05,
+      "loss": 0.9206,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999989465319296e-05,
+      "loss": 0.9261,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999981271681304e-05,
+      "loss": 0.9142,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997073700718e-05,
+      "loss": 0.9182,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999957861299375e-05,
+      "loss": 0.865,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999942644560915e-05,
+      "loss": 0.9259,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999925086795364e-05,
+      "loss": 0.8731,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999905188006825e-05,
+      "loss": 0.8656,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999882948199963e-05,
+      "loss": 0.8186,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999858367379978e-05,
+      "loss": 0.9108,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999831445552628e-05,
+      "loss": 0.9758,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999802182724214e-05,
+      "loss": 0.9759,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999977057890159e-05,
+      "loss": 0.9402,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999736634092154e-05,
+      "loss": 0.9062,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999970034830385e-05,
+      "loss": 0.9724,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999661721545173e-05,
+      "loss": 0.8514,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999620753825163e-05,
+      "loss": 0.9226,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999577445153416e-05,
+      "loss": 0.8011,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999531795540068e-05,
+      "loss": 0.898,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999483804995807e-05,
+      "loss": 0.9074,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999943347353187e-05,
+      "loss": 0.9067,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999938080116003e-05,
+      "loss": 0.8686,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999932578789263e-05,
+      "loss": 0.8734,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999268433742542e-05,
+      "loss": 0.9557,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999208738723195e-05,
+      "loss": 0.8192,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999146702848565e-05,
+      "loss": 0.9587,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999082326133174e-05,
+      "loss": 0.839,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999015608592086e-05,
+      "loss": 0.8191,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999894655024093e-05,
+      "loss": 0.9274,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999887515109587e-05,
+      "loss": 0.3089,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998801411173616e-05,
+      "loss": 0.8904,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998725330491436e-05,
+      "loss": 0.9338,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998646909067138e-05,
+      "loss": 0.9278,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998566146919083e-05,
+      "loss": 0.9232,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998483044066176e-05,
+      "loss": 0.8574,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998397600527874e-05,
+      "loss": 0.8934,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998309816324174e-05,
+      "loss": 0.8704,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998219691475634e-05,
+      "loss": 0.91,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999812722600335e-05,
+      "loss": 0.8859,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998032419928965e-05,
+      "loss": 0.8616,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997935273274682e-05,
+      "loss": 0.8263,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999783578606323e-05,
+      "loss": 0.8352,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999773395831791e-05,
+      "loss": 0.8525,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997629790062557e-05,
+      "loss": 0.8565,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997523281321562e-05,
+      "loss": 0.8993,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999741443211985e-05,
+      "loss": 0.8724,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999730324248291e-05,
+      "loss": 0.8972,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997189712436768e-05,
+      "loss": 0.8641,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997073842008007e-05,
+      "loss": 0.9023,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996955631223744e-05,
+      "loss": 0.3743,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996835080111662e-05,
+      "loss": 0.8684,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996712188699975e-05,
+      "loss": 0.8398,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996586957017457e-05,
+      "loss": 0.3438,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996459385093425e-05,
+      "loss": 0.8526,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996329472957742e-05,
+      "loss": 0.8863,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996197220640826e-05,
+      "loss": 0.8865,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999606262817363e-05,
+      "loss": 0.8552,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995925695587666e-05,
+      "loss": 0.86,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995786422914993e-05,
+      "loss": 0.88,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995644810188213e-05,
+      "loss": 0.8156,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995500857440477e-05,
+      "loss": 0.9213,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995354564705488e-05,
+      "loss": 0.8865,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999520593201749e-05,
+      "loss": 0.9792,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995054959411284e-05,
+      "loss": 0.8744,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994901646922206e-05,
+      "loss": 0.8939,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999474599458615e-05,
+      "loss": 0.9335,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999458800243956e-05,
+      "loss": 0.7747,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999442767051942e-05,
+      "loss": 0.9242,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994264998863255e-05,
+      "loss": 0.8728,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999409998750916e-05,
+      "loss": 0.8901,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993932636495762e-05,
+      "loss": 0.8683,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993762945862232e-05,
+      "loss": 0.8209,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99935909156483e-05,
+      "loss": 0.3209,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999341654589424e-05,
+      "loss": 0.922,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999323983664087e-05,
+      "loss": 0.8715,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999306078792956e-05,
+      "loss": 0.8537,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992879399802227e-05,
+      "loss": 0.8919,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992695672301328e-05,
+      "loss": 0.8113,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992509605469884e-05,
+      "loss": 0.8933,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992321199351446e-05,
+      "loss": 0.9036,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992130453990127e-05,
+      "loss": 0.8603,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999193736943058e-05,
+      "loss": 0.8762,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991741945718e-05,
+      "loss": 0.9099,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991544182898147e-05,
+      "loss": 0.841,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991344081017312e-05,
+      "loss": 0.8948,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999114164012234e-05,
+      "loss": 0.8651,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990936860260624e-05,
+      "loss": 0.8945,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99907297414801e-05,
+      "loss": 0.9455,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999052028382926e-05,
+      "loss": 0.8644,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990308487357143e-05,
+      "loss": 0.8918,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999009435211332e-05,
+      "loss": 0.8516,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998987787814793e-05,
+      "loss": 0.8825,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998965906551165e-05,
+      "loss": 0.8922,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99894379142557e-05,
+      "loss": 0.8671,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989214424431855e-05,
+      "loss": 0.3441,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988988596092435e-05,
+      "loss": 0.9234,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998876042929031e-05,
+      "loss": 0.9387,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988529924078884e-05,
+      "loss": 0.8231,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998829708051213e-05,
+      "loss": 0.7832,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988061898644558e-05,
+      "loss": 0.95,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987824378531222e-05,
+      "loss": 0.8675,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987584520227722e-05,
+      "loss": 0.2902,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998734232379022e-05,
+      "loss": 0.8159,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987097789275403e-05,
+      "loss": 0.8889,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986850916740526e-05,
+      "loss": 0.9098,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998660170624338e-05,
+      "loss": 0.9353,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998635015784231e-05,
+      "loss": 0.834,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986096271596196e-05,
+      "loss": 0.8848,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985840047564478e-05,
+      "loss": 0.8834,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985581485807146e-05,
+      "loss": 0.9315,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985320586384722e-05,
+      "loss": 0.9239,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985057349358287e-05,
+      "loss": 0.9668,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984791774789464e-05,
+      "loss": 0.8652,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984523862740426e-05,
+      "loss": 0.3256,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984253613273893e-05,
+      "loss": 0.9012,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998398102645313e-05,
+      "loss": 0.9038,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998370610234195e-05,
+      "loss": 0.8928,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983428841004714e-05,
+      "loss": 0.837,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983149242506335e-05,
+      "loss": 0.9348,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982867306912262e-05,
+      "loss": 0.8396,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99825830342885e-05,
+      "loss": 0.8899,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99822964247016e-05,
+      "loss": 0.8575,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982007478218656e-05,
+      "loss": 0.8696,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981716194907312e-05,
+      "loss": 0.8597,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981422574835757e-05,
+      "loss": 0.8723,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998112661807273e-05,
+      "loss": 0.2979,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998082832468752e-05,
+      "loss": 0.87,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980527694749952e-05,
+      "loss": 0.8706,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980224728330405e-05,
+      "loss": 0.8857,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997991942549981e-05,
+      "loss": 0.9036,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979611786329638e-05,
+      "loss": 1.0028,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979301810891908e-05,
+      "loss": 0.8594,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978989499259183e-05,
+      "loss": 0.9394,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978674851504578e-05,
+      "loss": 0.8937,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978357867701758e-05,
+      "loss": 0.9239,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978038547924924e-05,
+      "loss": 0.8812,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977716892248834e-05,
+      "loss": 0.8402,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977392900748786e-05,
+      "loss": 0.8818,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997706657350063e-05,
+      "loss": 0.9149,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997673791058076e-05,
+      "loss": 0.9102,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976406912066118e-05,
+      "loss": 0.797,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976073578034192e-05,
+      "loss": 0.9711,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975737908563017e-05,
+      "loss": 0.8188,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997539990373117e-05,
+      "loss": 0.8555,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975059563617786e-05,
+      "loss": 0.8893,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974716888302535e-05,
+      "loss": 0.8928,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997437187786564e-05,
+      "loss": 0.9064,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974024532387873e-05,
+      "loss": 0.8513,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973674851950547e-05,
+      "loss": 0.8751,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973322836635517e-05,
+      "loss": 0.8614,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99729684865252e-05,
+      "loss": 0.9059,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997261180170255e-05,
+      "loss": 0.8836,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972252782251065e-05,
+      "loss": 0.8978,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971891428254793e-05,
+      "loss": 0.8244,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971527739798326e-05,
+      "loss": 0.8612,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971161716966813e-05,
+      "loss": 0.9135,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970793359845935e-05,
+      "loss": 0.8308,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997042266852193e-05,
+      "loss": 0.925,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970049643081574e-05,
+      "loss": 0.9564,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969674283612196e-05,
+      "loss": 0.9028,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996929659020167e-05,
+      "loss": 0.8568,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996891656293841e-05,
+      "loss": 0.8369,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996853420191139e-05,
+      "loss": 0.8835,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996814950721012e-05,
+      "loss": 0.9209,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967762478924654e-05,
+      "loss": 0.8939,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99673731171456e-05,
+      "loss": 0.8956,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996698142196411e-05,
+      "loss": 0.893,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966587393471884e-05,
+      "loss": 0.8015,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966191031761158e-05,
+      "loss": 0.8516,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996579233692473e-05,
+      "loss": 0.8583,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996539130905593e-05,
+      "loss": 0.8133,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964987948248645e-05,
+      "loss": 0.8492,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964582254597297e-05,
+      "loss": 0.8152,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996417422819687e-05,
+      "loss": 0.8624,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963763869142874e-05,
+      "loss": 0.8153,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963351177531382e-05,
+      "loss": 0.9109,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962936153459006e-05,
+      "loss": 0.8232,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962518797022906e-05,
+      "loss": 0.8677,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962099108320786e-05,
+      "loss": 0.9196,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961677087450893e-05,
+      "loss": 0.9195,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996125273451203e-05,
+      "loss": 0.3331,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960826049603535e-05,
+      "loss": 0.9038,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99603970328253e-05,
+      "loss": 0.928,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959965684277755e-05,
+      "loss": 0.8677,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959532004061884e-05,
+      "loss": 0.8075,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959095992279212e-05,
+      "loss": 0.8438,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958657649031815e-05,
+      "loss": 0.8915,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958216974422307e-05,
+      "loss": 0.8357,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995777396855385e-05,
+      "loss": 0.863,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957328631530157e-05,
+      "loss": 0.8385,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956880963455482e-05,
+      "loss": 0.8876,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956430964434627e-05,
+      "loss": 0.8058,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995597863457294e-05,
+      "loss": 0.8906,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955523973976307e-05,
+      "loss": 0.8494,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955066982751168e-05,
+      "loss": 0.8952,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995460766100451e-05,
+      "loss": 0.8104,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954146008843863e-05,
+      "loss": 0.8651,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953682026377296e-05,
+      "loss": 0.9033,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953215713713433e-05,
+      "loss": 0.8607,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995274707096144e-05,
+      "loss": 0.851,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952276098231027e-05,
+      "loss": 0.8815,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995180279563245e-05,
+      "loss": 0.8386,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995132716327651e-05,
+      "loss": 0.8695,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950849201274557e-05,
+      "loss": 0.8913,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950368909738485e-05,
+      "loss": 0.8986,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949886288780727e-05,
+      "loss": 0.9308,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949401338514273e-05,
+      "loss": 0.8829,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948914059052646e-05,
+      "loss": 0.8715,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948424450509926e-05,
+      "loss": 0.8321,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947932513000724e-05,
+      "loss": 0.9067,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994743824664021e-05,
+      "loss": 0.799,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99469416515441e-05,
+      "loss": 0.8826,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946442727828633e-05,
+      "loss": 0.8929,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945941475610623e-05,
+      "loss": 0.8366,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994543789500741e-05,
+      "loss": 0.849,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944931986136886e-05,
+      "loss": 0.8916,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994442374911748e-05,
+      "loss": 0.8685,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943913184068184e-05,
+      "loss": 0.8472,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943400291108513e-05,
+      "loss": 0.8856,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942885070358545e-05,
+      "loss": 0.8445,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942367521938887e-05,
+      "loss": 0.8267,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941847645970705e-05,
+      "loss": 0.8628,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99413254425757e-05,
+      "loss": 0.8112,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994080091187613e-05,
+      "loss": 0.9219,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940274053994778e-05,
+      "loss": 0.8276,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939744869054993e-05,
+      "loss": 0.7728,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939213357180657e-05,
+      "loss": 0.895,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938679518496194e-05,
+      "loss": 0.8452,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938143353126588e-05,
+      "loss": 0.7902,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993760486119735e-05,
+      "loss": 0.8457,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993706404283454e-05,
+      "loss": 0.8344,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936520898164775e-05,
+      "loss": 0.8262,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99359754273152e-05,
+      "loss": 0.8431,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935427630413517e-05,
+      "loss": 0.8593,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934877507587965e-05,
+      "loss": 0.8356,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993432505896733e-05,
+      "loss": 0.8,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933770284680943e-05,
+      "loss": 0.839,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933213184858676e-05,
+      "loss": 0.9707,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932653759630956e-05,
+      "loss": 0.9385,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932092009128735e-05,
+      "loss": 0.8003,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931527933483533e-05,
+      "loss": 0.8007,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930961532827396e-05,
+      "loss": 0.8159,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993039280729292e-05,
+      "loss": 0.8825,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929821757013248e-05,
+      "loss": 0.9268,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929248382122065e-05,
+      "loss": 0.8479,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928672682753602e-05,
+      "loss": 0.818,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928094659042628e-05,
+      "loss": 0.823,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927514311124465e-05,
+      "loss": 0.8101,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926931639134974e-05,
+      "loss": 0.8372,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992634664321056e-05,
+      "loss": 0.7628,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992575932348817e-05,
+      "loss": 0.8895,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925169680105304e-05,
+      "loss": 0.7889,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924577713199996e-05,
+      "loss": 0.8578,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992398342291083e-05,
+      "loss": 0.8899,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923386809376933e-05,
+      "loss": 0.8234,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922787872737967e-05,
+      "loss": 0.8526,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922186613134152e-05,
+      "loss": 0.9107,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921583030706245e-05,
+      "loss": 0.9177,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920977125595547e-05,
+      "loss": 0.9531,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99203688979439e-05,
+      "loss": 0.913,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919758347893694e-05,
+      "loss": 0.8694,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919145475587865e-05,
+      "loss": 0.8896,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918530281169884e-05,
+      "loss": 0.9306,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917912764783772e-05,
+      "loss": 0.8854,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991729292657409e-05,
+      "loss": 0.9506,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991667076668595e-05,
+      "loss": 0.8977,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916046285265e-05,
+      "loss": 0.9137,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991541948245743e-05,
+      "loss": 0.8502,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914790358409978e-05,
+      "loss": 0.8398,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914158913269926e-05,
+      "loss": 0.8653,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913525147185104e-05,
+      "loss": 0.9055,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912889060303867e-05,
+      "loss": 0.9052,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991225065277514e-05,
+      "loss": 0.7965,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991160992474836e-05,
+      "loss": 0.8651,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910966876373535e-05,
+      "loss": 0.917,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910321507801204e-05,
+      "loss": 0.7994,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909673819182453e-05,
+      "loss": 0.8626,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909023810668896e-05,
+      "loss": 0.7661,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990837148241272e-05,
+      "loss": 0.8012,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990771683456662e-05,
+      "loss": 0.8255,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907059867283868e-05,
+      "loss": 0.8367,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906400580718253e-05,
+      "loss": 0.8886,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990573897502412e-05,
+      "loss": 0.8525,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990507505035635e-05,
+      "loss": 0.9059,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990440880687038e-05,
+      "loss": 0.8513,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990374024472217e-05,
+      "loss": 0.8509,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903069364068233e-05,
+      "loss": 0.8256,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990239616506563e-05,
+      "loss": 0.8489,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990172064787196e-05,
+      "loss": 0.8149,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901042812645364e-05,
+      "loss": 0.8555,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990036265954452e-05,
+      "loss": 0.8963,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899680188728662e-05,
+      "loss": 0.8232,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898995400357556e-05,
+      "loss": 0.8014,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898308294591514e-05,
+      "loss": 0.9277,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897618871591387e-05,
+      "loss": 0.8682,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989692713151858e-05,
+      "loss": 0.8907,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989623307453502e-05,
+      "loss": 0.7797,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895536700803203e-05,
+      "loss": 0.8664,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894838010486138e-05,
+      "loss": 0.8705,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894137003747404e-05,
+      "loss": 0.7966,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893433680751105e-05,
+      "loss": 0.8267,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892728041661885e-05,
+      "loss": 0.9496,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989202008664495e-05,
+      "loss": 0.8741,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891309815866026e-05,
+      "loss": 0.8576,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989059722949139e-05,
+      "loss": 0.7725,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889882327687867e-05,
+      "loss": 0.9121,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988916511062281e-05,
+      "loss": 0.3335,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888445578464133e-05,
+      "loss": 0.3391,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887723731380274e-05,
+      "loss": 0.871,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988699956954022e-05,
+      "loss": 0.8743,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886273093113508e-05,
+      "loss": 0.8123,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885544302270197e-05,
+      "loss": 0.7631,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988481319718091e-05,
+      "loss": 0.8493,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98840797780168e-05,
+      "loss": 0.8262,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988334404494956e-05,
+      "loss": 0.858,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882605998151427e-05,
+      "loss": 0.9148,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881865637795187e-05,
+      "loss": 0.8971,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881122964054152e-05,
+      "loss": 0.8309,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880377977102196e-05,
+      "loss": 0.896,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879630677113714e-05,
+      "loss": 0.8344,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987888106426366e-05,
+      "loss": 0.8565,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878129138727518e-05,
+      "loss": 0.796,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877374900681313e-05,
+      "loss": 0.8117,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987661835030162e-05,
+      "loss": 0.9383,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987585948776555e-05,
+      "loss": 0.8527,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875098313250755e-05,
+      "loss": 0.7829,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874334826935427e-05,
+      "loss": 0.8732,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873569028998303e-05,
+      "loss": 0.8531,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987280091961866e-05,
+      "loss": 0.8369,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872030498976316e-05,
+      "loss": 0.8961,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987125776725163e-05,
+      "loss": 0.8331,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870482724625495e-05,
+      "loss": 0.8073,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986970537127936e-05,
+      "loss": 0.8016,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868925707395203e-05,
+      "loss": 0.8724,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868143733155545e-05,
+      "loss": 0.8292,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867359448743456e-05,
+      "loss": 0.8906,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986657285434253e-05,
+      "loss": 0.8594,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986578395013692e-05,
+      "loss": 0.8787,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986499273631131e-05,
+      "loss": 0.7982,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864199213050927e-05,
+      "loss": 0.8341,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986340338054153e-05,
+      "loss": 0.7631,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862605238969437e-05,
+      "loss": 0.7985,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986180478852149e-05,
+      "loss": 0.8592,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861002029385085e-05,
+      "loss": 0.8825,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986019696174814e-05,
+      "loss": 0.8559,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859389585799135e-05,
+      "loss": 0.7929,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985857990172707e-05,
+      "loss": 0.8606,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857767909721505e-05,
+      "loss": 0.9461,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856953609972526e-05,
+      "loss": 0.8618,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985613700267076e-05,
+      "loss": 0.8091,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855318088007393e-05,
+      "loss": 0.8048,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854496866174117e-05,
+      "loss": 0.7644,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853673337363196e-05,
+      "loss": 0.7746,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852847501767416e-05,
+      "loss": 0.8517,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852019359580107e-05,
+      "loss": 0.8537,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985118891099515e-05,
+      "loss": 0.8907,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850356156206945e-05,
+      "loss": 0.3665,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984952109541045e-05,
+      "loss": 0.8608,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848683728801153e-05,
+      "loss": 0.3365,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847844056575086e-05,
+      "loss": 0.8623,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847002078928817e-05,
+      "loss": 0.8145,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846157796059466e-05,
+      "loss": 0.3532,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845311208164674e-05,
+      "loss": 0.9018,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984446231544263e-05,
+      "loss": 0.8355,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843611118092067e-05,
+      "loss": 0.8278,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842757616312253e-05,
+      "loss": 0.8515,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841901810302997e-05,
+      "loss": 0.782,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841043700264644e-05,
+      "loss": 0.8947,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840183286398083e-05,
+      "loss": 0.9051,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839320568904742e-05,
+      "loss": 0.8031,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838455547986583e-05,
+      "loss": 0.3389,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983758822384611e-05,
+      "loss": 0.9393,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836718596686375e-05,
+      "loss": 0.9519,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835846666710952e-05,
+      "loss": 0.8209,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983497243412397e-05,
+      "loss": 0.7893,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834095899130088e-05,
+      "loss": 0.8529,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833217061934505e-05,
+      "loss": 0.847,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832335922742963e-05,
+      "loss": 0.8099,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831452481761736e-05,
+      "loss": 0.8283,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830566739197647e-05,
+      "loss": 0.8688,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829678695258044e-05,
+      "loss": 0.7699,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982878835015083e-05,
+      "loss": 0.3481,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982789570408443e-05,
+      "loss": 0.7989,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827000757267826e-05,
+      "loss": 0.6851,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826103509910525e-05,
+      "loss": 0.8796,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825203962222573e-05,
+      "loss": 0.8032,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982430211441456e-05,
+      "loss": 0.9323,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823397966697607e-05,
+      "loss": 0.902,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822491519283388e-05,
+      "loss": 0.9195,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821582772384104e-05,
+      "loss": 0.8716,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982067172621249e-05,
+      "loss": 0.828,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819758380981828e-05,
+      "loss": 0.9187,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818842736905937e-05,
+      "loss": 0.7779,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817924794199177e-05,
+      "loss": 0.867,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981700455307644e-05,
+      "loss": 0.8808,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816082013753148e-05,
+      "loss": 0.9421,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815157176445286e-05,
+      "loss": 0.783,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814230041369353e-05,
+      "loss": 0.836,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813300608742402e-05,
+      "loss": 0.7691,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812368878782008e-05,
+      "loss": 0.8712,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98114348517063e-05,
+      "loss": 0.8606,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810498527733934e-05,
+      "loss": 0.8755,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809559907084112e-05,
+      "loss": 0.8073,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808618989976562e-05,
+      "loss": 0.9041,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980767577663156e-05,
+      "loss": 0.9194,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806730267269917e-05,
+      "loss": 0.9396,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980578246211298e-05,
+      "loss": 0.8402,
+      "step": 609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980483236138263e-05,
+      "loss": 0.8633,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803879965301296e-05,
+      "loss": 0.8017,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980292527409193e-05,
+      "loss": 0.8451,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801968287978043e-05,
+      "loss": 0.9076,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801009007183653e-05,
+      "loss": 0.7988,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980004743193334e-05,
+      "loss": 0.8776,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979908356245221e-05,
+      "loss": 0.8866,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798117398965914e-05,
+      "loss": 0.8179,
+      "step": 617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797148941700628e-05,
+      "loss": 0.7292,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796178190883074e-05,
+      "loss": 0.8553,
+      "step": 619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979520514674051e-05,
+      "loss": 0.8539,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794229809500728e-05,
+      "loss": 0.8575,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793252179392062e-05,
+      "loss": 0.8835,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792272256643374e-05,
+      "loss": 0.8215,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979129004148407e-05,
+      "loss": 0.833,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790305534144096e-05,
+      "loss": 0.8251,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789318734853916e-05,
+      "loss": 0.7992,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788329643844557e-05,
+      "loss": 0.8114,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978733826134756e-05,
+      "loss": 0.787,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978634458759502e-05,
+      "loss": 0.8583,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785348622819555e-05,
+      "loss": 0.9006,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784350367254322e-05,
+      "loss": 0.81,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783349821133022e-05,
+      "loss": 0.8511,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782346984689883e-05,
+      "loss": 0.8774,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781341858159676e-05,
+      "loss": 0.88,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97803344417777e-05,
+      "loss": 0.8799,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779324735779806e-05,
+      "loss": 0.7913,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977831274040236e-05,
+      "loss": 0.8416,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777298455882276e-05,
+      "loss": 0.8213,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776281882457007e-05,
+      "loss": 0.8468,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977526302036453e-05,
+      "loss": 0.8208,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774241869843375e-05,
+      "loss": 0.8466,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773218431132582e-05,
+      "loss": 0.8222,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772192704471758e-05,
+      "loss": 0.8063,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977116469010102e-05,
+      "loss": 0.7508,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770134388261037e-05,
+      "loss": 0.8067,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769101799192997e-05,
+      "loss": 0.7897,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976806692313864e-05,
+      "loss": 0.8905,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767029760340238e-05,
+      "loss": 0.8349,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765990311040588e-05,
+      "loss": 0.8597,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764948575483033e-05,
+      "loss": 0.8506,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976390455391145e-05,
+      "loss": 0.8192,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976285824657024e-05,
+      "loss": 0.8306,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761809653704357e-05,
+      "loss": 0.8479,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760758775559275e-05,
+      "loss": 0.8516,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975970561238101e-05,
+      "loss": 0.8258,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758650164416116e-05,
+      "loss": 0.8065,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757592431911668e-05,
+      "loss": 0.8709,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97565324151153e-05,
+      "loss": 0.8749,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975547011427515e-05,
+      "loss": 0.7469,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754405529639923e-05,
+      "loss": 0.8709,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753338661458833e-05,
+      "loss": 0.7824,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752269509981637e-05,
+      "loss": 0.8607,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751198075458627e-05,
+      "loss": 0.937,
+      "step": 663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975012435814064e-05,
+      "loss": 0.8847,
+      "step": 664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749048358279027e-05,
+      "loss": 0.7706,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974797007612569e-05,
+      "loss": 0.8191,
+      "step": 666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746889511933058e-05,
+      "loss": 0.8352,
+      "step": 667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974580666595409e-05,
+      "loss": 0.8548,
+      "step": 668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744721538442295e-05,
+      "loss": 0.7965,
+      "step": 669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97436341296517e-05,
+      "loss": 0.866,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974254443983687e-05,
+      "loss": 0.9027,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741452469252908e-05,
+      "loss": 0.8235,
+      "step": 672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974035821815545e-05,
+      "loss": 0.3389,
+      "step": 673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739261686800662e-05,
+      "loss": 0.8011,
+      "step": 674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738162875445247e-05,
+      "loss": 0.8376,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737061784346442e-05,
+      "loss": 0.3444,
+      "step": 676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735958413762012e-05,
+      "loss": 0.7621,
+      "step": 677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734852763950272e-05,
+      "loss": 0.9085,
+      "step": 678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973374483517005e-05,
+      "loss": 0.8909,
+      "step": 679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973263462768072e-05,
+      "loss": 0.8287,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731522141742178e-05,
+      "loss": 0.3362,
+      "step": 681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730407377614876e-05,
+      "loss": 0.8538,
+      "step": 682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729290335559773e-05,
+      "loss": 0.8477,
+      "step": 683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972817101583838e-05,
+      "loss": 0.7835,
+      "step": 684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727049418712733e-05,
+      "loss": 0.8285,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725925544445398e-05,
+      "loss": 0.804,
+      "step": 686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972479939329948e-05,
+      "loss": 0.8029,
+      "step": 687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972367096553862e-05,
+      "loss": 0.7579,
+      "step": 688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722540261426984e-05,
+      "loss": 0.8714,
+      "step": 689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721407281229273e-05,
+      "loss": 0.8247,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720272025210728e-05,
+      "loss": 0.8654,
+      "step": 691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719134493637112e-05,
+      "loss": 0.9327,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717994686774724e-05,
+      "loss": 0.7931,
+      "step": 693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97168526048904e-05,
+      "loss": 0.8818,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971570824825151e-05,
+      "loss": 0.7888,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714561617125948e-05,
+      "loss": 0.8533,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713412711782144e-05,
+      "loss": 0.8123,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971226153248907e-05,
+      "loss": 0.8706,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711108079516208e-05,
+      "loss": 0.8354,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709952353133594e-05,
+      "loss": 0.8458,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970879435361179e-05,
+      "loss": 0.8121,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970763408122188e-05,
+      "loss": 0.8825,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97064715362355e-05,
+      "loss": 0.8189,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97053067189248e-05,
+      "loss": 0.844,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704139629562466e-05,
+      "loss": 0.8849,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702970268421724e-05,
+      "loss": 0.8482,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701798635776323e-05,
+      "loss": 0.8626,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700624731900548e-05,
+      "loss": 0.8278,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699448557069213e-05,
+      "loss": 0.8357,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698270111557667e-05,
+      "loss": 0.8225,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969708939564179e-05,
+      "loss": 0.3561,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969590640959799e-05,
+      "loss": 0.8286,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694721153703206e-05,
+      "loss": 0.826,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969353362823492e-05,
+      "loss": 0.8459,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969234383347113e-05,
+      "loss": 0.8701,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969115176969037e-05,
+      "loss": 0.8374,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968995743717171e-05,
+      "loss": 0.8521,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688760836194755e-05,
+      "loss": 0.8883,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968756196703962e-05,
+      "loss": 0.8361,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686360829986974e-05,
+      "loss": 0.8349,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968515742531801e-05,
+      "loss": 0.7905,
+      "step": 721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683951753314443e-05,
+      "loss": 0.8895,
+      "step": 722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968274381425853e-05,
+      "loss": 0.8547,
+      "step": 723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968153360843305e-05,
+      "loss": 0.869,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680321136121328e-05,
+      "loss": 0.8287,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679106397607195e-05,
+      "loss": 0.7691,
+      "step": 726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677889393175035e-05,
+      "loss": 0.8309,
+      "step": 727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676670123109753e-05,
+      "loss": 0.8081,
+      "step": 728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967544858769678e-05,
+      "loss": 0.8906,
+      "step": 729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967422478722209e-05,
+      "loss": 0.3405,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967299872197217e-05,
+      "loss": 0.8784,
+      "step": 731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967177039223405e-05,
+      "loss": 0.8488,
+      "step": 732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670539798295296e-05,
+      "loss": 0.8324,
+      "step": 733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669306940443983e-05,
+      "loss": 0.8046,
+      "step": 734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668071818968736e-05,
+      "loss": 0.8537,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666834434158696e-05,
+      "loss": 0.8122,
+      "step": 736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966559478630355e-05,
+      "loss": 0.8028,
+      "step": 737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664352875693493e-05,
+      "loss": 0.8337,
+      "step": 738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966310870261926e-05,
+      "loss": 0.8304,
+      "step": 739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966186226737213e-05,
+      "loss": 0.8423,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966061357024389e-05,
+      "loss": 0.8827,
+      "step": 741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659362611526864e-05,
+      "loss": 0.8465,
+      "step": 742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965810939151391e-05,
+      "loss": 0.8937,
+      "step": 743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965685391049841e-05,
+      "loss": 0.8528,
+      "step": 744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655596168774284e-05,
+      "loss": 0.8292,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654336166635963e-05,
+      "loss": 0.8581,
+      "step": 746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653073904378422e-05,
+      "loss": 0.8975,
+      "step": 747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651809382297165e-05,
+      "loss": 0.8601,
+      "step": 748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965054260068822e-05,
+      "loss": 0.328,
+      "step": 749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964927355984815e-05,
+      "loss": 0.8021,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648002260074034e-05,
+      "loss": 0.8426,
+      "step": 751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646728701663495e-05,
+      "loss": 0.854,
+      "step": 752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645452884914675e-05,
+      "loss": 0.9165,
+      "step": 753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964417481012625e-05,
+      "loss": 0.889,
+      "step": 754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964289447759742e-05,
+      "loss": 0.8216,
+      "step": 755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641611887627925e-05,
+      "loss": 0.7601,
+      "step": 756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964032704051801e-05,
+      "loss": 0.9002,
+      "step": 757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639039936568473e-05,
+      "loss": 0.7837,
+      "step": 758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637750576080626e-05,
+      "loss": 0.789,
+      "step": 759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963645895935632e-05,
+      "loss": 0.8858,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635165086697914e-05,
+      "loss": 0.8071,
+      "step": 761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963386895840832e-05,
+      "loss": 0.9202,
+      "step": 762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632570574790968e-05,
+      "loss": 0.8837,
+      "step": 763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631269936149807e-05,
+      "loss": 0.8391,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962996704278933e-05,
+      "loss": 0.7923,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628661895014542e-05,
+      "loss": 0.8334,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627354493130986e-05,
+      "loss": 0.8213,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626044837444736e-05,
+      "loss": 0.8002,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962473292826238e-05,
+      "loss": 0.8438,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623418765891045e-05,
+      "loss": 0.8084,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622102350638377e-05,
+      "loss": 0.8614,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962078368281256e-05,
+      "loss": 0.815,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96194627627223e-05,
+      "loss": 0.7956,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618139590676824e-05,
+      "loss": 0.8162,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616814166985895e-05,
+      "loss": 0.8344,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615486491959803e-05,
+      "loss": 0.8994,
+      "step": 776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961415656590936e-05,
+      "loss": 0.8437,
+      "step": 777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612824389145898e-05,
+      "loss": 0.8601,
+      "step": 778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611489961981302e-05,
+      "loss": 0.3536,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610153284727956e-05,
+      "loss": 0.855,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960881435769878e-05,
+      "loss": 0.3419,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960747318120723e-05,
+      "loss": 0.864,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606129755567275e-05,
+      "loss": 0.8272,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960478408109342e-05,
+      "loss": 0.874,
+      "step": 784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603436158100686e-05,
+      "loss": 0.8137,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602085986904635e-05,
+      "loss": 0.8024,
+      "step": 786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600733567821347e-05,
+      "loss": 0.8031,
+      "step": 787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599378901167423e-05,
+      "loss": 0.795,
+      "step": 788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598021987260002e-05,
+      "loss": 0.7892,
+      "step": 789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596662826416736e-05,
+      "loss": 0.8936,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595301418955823e-05,
+      "loss": 0.8387,
+      "step": 791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593937765195958e-05,
+      "loss": 0.8209,
+      "step": 792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592571865456388e-05,
+      "loss": 0.8038,
+      "step": 793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591203720056872e-05,
+      "loss": 0.8044,
+      "step": 794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589833329317703e-05,
+      "loss": 0.8345,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958846069355969e-05,
+      "loss": 0.7583,
+      "step": 796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587085813104168e-05,
+      "loss": 0.8767,
+      "step": 797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585708688273016e-05,
+      "loss": 0.8091,
+      "step": 798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584329319388614e-05,
+      "loss": 0.8112,
+      "step": 799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958294770677388e-05,
+      "loss": 0.931,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581563850752257e-05,
+      "loss": 0.8394,
+      "step": 801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958017775164771e-05,
+      "loss": 0.372,
+      "step": 802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578789409784727e-05,
+      "loss": 0.7803,
+      "step": 803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577398825488333e-05,
+      "loss": 0.8478,
+      "step": 804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957600599908406e-05,
+      "loss": 0.8434,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574610930897982e-05,
+      "loss": 0.8591,
+      "step": 806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573213621256684e-05,
+      "loss": 0.831,
+      "step": 807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571814070487286e-05,
+      "loss": 0.8832,
+      "step": 808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570412278917427e-05,
+      "loss": 0.8367,
+      "step": 809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956900824687527e-05,
+      "loss": 0.8514,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567601974689508e-05,
+      "loss": 0.8812,
+      "step": 811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566193462689355e-05,
+      "loss": 0.8381,
+      "step": 812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956478271120455e-05,
+      "loss": 0.8002,
+      "step": 813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956336972056535e-05,
+      "loss": 0.87,
+      "step": 814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561954491102545e-05,
+      "loss": 0.8457,
+      "step": 815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560537023147447e-05,
+      "loss": 0.7638,
+      "step": 816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559117317031893e-05,
+      "loss": 0.849,
+      "step": 817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557695373088235e-05,
+      "loss": 0.8533,
+      "step": 818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556271191649365e-05,
+      "loss": 0.8468,
+      "step": 819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554844773048683e-05,
+      "loss": 0.841,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553416117620122e-05,
+      "loss": 0.8552,
+      "step": 821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551985225698133e-05,
+      "loss": 0.8598,
+      "step": 822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95505520976177e-05,
+      "loss": 0.8658,
+      "step": 823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549116733714317e-05,
+      "loss": 0.3535,
+      "step": 824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954767913432401e-05,
+      "loss": 0.8564,
+      "step": 825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546239299783333e-05,
+      "loss": 0.8703,
+      "step": 826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954479723042935e-05,
+      "loss": 0.7663,
+      "step": 827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543352926599657e-05,
+      "loss": 0.859,
+      "step": 828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541906388632374e-05,
+      "loss": 0.9178,
+      "step": 829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540457616866136e-05,
+      "loss": 0.7754,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539006611640113e-05,
+      "loss": 0.8335,
+      "step": 831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537553373293988e-05,
+      "loss": 0.8161,
+      "step": 832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953609790216797e-05,
+      "loss": 0.7864,
+      "step": 833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534640198602792e-05,
+      "loss": 0.8964,
+      "step": 834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533180262939704e-05,
+      "loss": 0.8287,
+      "step": 835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953171809552049e-05,
+      "loss": 0.8767,
+      "step": 836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953025369668744e-05,
+      "loss": 0.8406,
+      "step": 837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528787066783384e-05,
+      "loss": 0.8531,
+      "step": 838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527318206151662e-05,
+      "loss": 0.7919,
+      "step": 839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525847115136142e-05,
+      "loss": 0.809,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952437379408121e-05,
+      "loss": 0.7823,
+      "step": 841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952289824333178e-05,
+      "loss": 0.8581,
+      "step": 842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521420463233283e-05,
+      "loss": 0.8408,
+      "step": 843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951994045413167e-05,
+      "loss": 0.8814,
+      "step": 844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951845821637342e-05,
+      "loss": 0.7581,
+      "step": 845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951697375030553e-05,
+      "loss": 0.863,
+      "step": 846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515487056275525e-05,
+      "loss": 0.8884,
+      "step": 847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951399813463144e-05,
+      "loss": 0.8107,
+      "step": 848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512506985721836e-05,
+      "loss": 0.8538,
+      "step": 849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95110136098958e-05,
+      "loss": 0.8186,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509518007502936e-05,
+      "loss": 0.8456,
+      "step": 851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508020178893377e-05,
+      "loss": 0.7893,
+      "step": 852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950652012441776e-05,
+      "loss": 0.8453,
+      "step": 853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505017844427264e-05,
+      "loss": 0.8639,
+      "step": 854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950351333927357e-05,
+      "loss": 0.7969,
+      "step": 855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502006609308896e-05,
+      "loss": 0.894,
+      "step": 856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500497654885972e-05,
+      "loss": 0.8445,
+      "step": 857
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498986476358044e-05,
+      "loss": 0.8349,
+      "step": 858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497473074078893e-05,
+      "loss": 0.8799,
+      "step": 859
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949595744840281e-05,
+      "loss": 0.8396,
+      "step": 860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949443959968461e-05,
+      "loss": 0.8411,
+      "step": 861
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492919528279622e-05,
+      "loss": 0.3507,
+      "step": 862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491397234543704e-05,
+      "loss": 0.8398,
+      "step": 863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489872718833238e-05,
+      "loss": 0.9419,
+      "step": 864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948834598150511e-05,
+      "loss": 0.8876,
+      "step": 865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486817022916738e-05,
+      "loss": 0.3473,
+      "step": 866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948528584342606e-05,
+      "loss": 0.7626,
+      "step": 867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948375244339153e-05,
+      "loss": 0.8888,
+      "step": 868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482216823172126e-05,
+      "loss": 0.8343,
+      "step": 869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948067898312733e-05,
+      "loss": 0.7994,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479138923617176e-05,
+      "loss": 0.8397,
+      "step": 871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477596645002183e-05,
+      "loss": 0.8389,
+      "step": 872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476052147643413e-05,
+      "loss": 0.8423,
+      "step": 873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474505431902434e-05,
+      "loss": 0.8123,
+      "step": 874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947295649814134e-05,
+      "loss": 0.8023,
+      "step": 875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471405346722747e-05,
+      "loss": 0.8354,
+      "step": 876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469851978009778e-05,
+      "loss": 0.902,
+      "step": 877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468296392366092e-05,
+      "loss": 0.8446,
+      "step": 878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946673859015585e-05,
+      "loss": 0.8192,
+      "step": 879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946517857174374e-05,
+      "loss": 0.8343,
+      "step": 880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463616337494974e-05,
+      "loss": 0.8608,
+      "step": 881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462051887775277e-05,
+      "loss": 0.7858,
+      "step": 882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946048522295089e-05,
+      "loss": 0.7863,
+      "step": 883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945891634338857e-05,
+      "loss": 0.8408,
+      "step": 884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945734524945561e-05,
+      "loss": 0.8232,
+      "step": 885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455771941519806e-05,
+      "loss": 0.8536,
+      "step": 886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945419641994947e-05,
+      "loss": 0.8971,
+      "step": 887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945261868511344e-05,
+      "loss": 0.3683,
+      "step": 888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451038737381078e-05,
+      "loss": 0.8595,
+      "step": 889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449456577122247e-05,
+      "loss": 0.375,
+      "step": 890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944787220470734e-05,
+      "loss": 0.8894,
+      "step": 891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446285620507263e-05,
+      "loss": 0.8337,
+      "step": 892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444696824893447e-05,
+      "loss": 0.7552,
+      "step": 893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944310581823783e-05,
+      "loss": 0.8337,
+      "step": 894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441512600912874e-05,
+      "loss": 0.7611,
+      "step": 895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439917173291563e-05,
+      "loss": 0.8328,
+      "step": 896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438319535747386e-05,
+      "loss": 0.8192,
+      "step": 897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943671968865436e-05,
+      "loss": 0.8036,
+      "step": 898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943511763238702e-05,
+      "loss": 0.7395,
+      "step": 899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94335133673204e-05,
+      "loss": 0.8409,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943190689383008e-05,
+      "loss": 0.773,
+      "step": 901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430298212292136e-05,
+      "loss": 0.8965,
+      "step": 902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428687323083167e-05,
+      "loss": 0.8657,
+      "step": 903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427074226580288e-05,
+      "loss": 0.8082,
+      "step": 904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942545892316113e-05,
+      "loss": 0.7819,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423841413203845e-05,
+      "loss": 0.8911,
+      "step": 906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422221697087098e-05,
+      "loss": 0.8205,
+      "step": 907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942059977519007e-05,
+      "loss": 0.8274,
+      "step": 908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418975647892462e-05,
+      "loss": 0.7638,
+      "step": 909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417349315574486e-05,
+      "loss": 0.7648,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941572077861688e-05,
+      "loss": 0.8177,
+      "step": 911
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414090037400876e-05,
+      "loss": 0.7916,
+      "step": 912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941245709230825e-05,
+      "loss": 0.7665,
+      "step": 913
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410821943721275e-05,
+      "loss": 0.7885,
+      "step": 914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409184592022755e-05,
+      "loss": 0.8427,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407545037595988e-05,
+      "loss": 0.3569,
+      "step": 916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405903280824807e-05,
+      "loss": 0.8199,
+      "step": 917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404259322093556e-05,
+      "loss": 0.8609,
+      "step": 918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402613161787086e-05,
+      "loss": 0.8037,
+      "step": 919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400964800290777e-05,
+      "loss": 0.8729,
+      "step": 920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399314237990512e-05,
+      "loss": 0.7628,
+      "step": 921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397661475272693e-05,
+      "loss": 0.8706,
+      "step": 922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396006512524246e-05,
+      "loss": 0.8209,
+      "step": 923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394349350132598e-05,
+      "loss": 0.8846,
+      "step": 924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392689988485696e-05,
+      "loss": 0.3346,
+      "step": 925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939102842797201e-05,
+      "loss": 0.8984,
+      "step": 926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389364668980516e-05,
+      "loss": 0.8091,
+      "step": 927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93876987119007e-05,
+      "loss": 0.798,
+      "step": 928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386030557122578e-05,
+      "loss": 0.3159,
+      "step": 929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938436020503667e-05,
+      "loss": 0.8439,
+      "step": 930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938268765603401e-05,
+      "loss": 0.8596,
+      "step": 931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381012910506146e-05,
+      "loss": 0.8333,
+      "step": 932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379335968845147e-05,
+      "loss": 0.8145,
+      "step": 933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937765683144359e-05,
+      "loss": 0.8838,
+      "step": 934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375975498694565e-05,
+      "loss": 0.836,
+      "step": 935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937429197099169e-05,
+      "loss": 0.8224,
+      "step": 936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372606248729075e-05,
+      "loss": 0.8663,
+      "step": 937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370918332301353e-05,
+      "loss": 0.8191,
+      "step": 938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936922822210368e-05,
+      "loss": 0.8718,
+      "step": 939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367535918531716e-05,
+      "loss": 0.3589,
+      "step": 940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365841421981632e-05,
+      "loss": 0.9024,
+      "step": 941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364144732850118e-05,
+      "loss": 0.7885,
+      "step": 942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362445851534378e-05,
+      "loss": 0.8171,
+      "step": 943
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360744778432124e-05,
+      "loss": 0.8454,
+      "step": 944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935904151394159e-05,
+      "loss": 0.9139,
+      "step": 945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357336058461506e-05,
+      "loss": 0.7914,
+      "step": 946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355628412391138e-05,
+      "loss": 0.8154,
+      "step": 947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353918576130242e-05,
+      "loss": 0.9018,
+      "step": 948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935220655007911e-05,
+      "loss": 0.8299,
+      "step": 949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935049233463852e-05,
+      "loss": 0.7977,
+      "step": 950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348775930209792e-05,
+      "loss": 0.8229,
+      "step": 951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934705733719473e-05,
+      "loss": 0.8374,
+      "step": 952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345336555995673e-05,
+      "loss": 0.9175,
+      "step": 953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343613587015457e-05,
+      "loss": 0.8135,
+      "step": 954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341888430657438e-05,
+      "loss": 0.8066,
+      "step": 955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340161087325483e-05,
+      "loss": 0.8642,
+      "step": 956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338431557423966e-05,
+      "loss": 0.7708,
+      "step": 957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336699841357784e-05,
+      "loss": 0.7597,
+      "step": 958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334965939532335e-05,
+      "loss": 0.8774,
+      "step": 959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333229852353526e-05,
+      "loss": 0.8534,
+      "step": 960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331491580227792e-05,
+      "loss": 0.8231,
+      "step": 961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329751123562065e-05,
+      "loss": 0.7521,
+      "step": 962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93280084827638e-05,
+      "loss": 0.7948,
+      "step": 963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326263658240942e-05,
+      "loss": 0.7927,
+      "step": 964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932451665040197e-05,
+      "loss": 0.8069,
+      "step": 965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932276745965587e-05,
+      "loss": 0.7965,
+      "step": 966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321016086412127e-05,
+      "loss": 0.8002,
+      "step": 967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931926253108075e-05,
+      "loss": 0.8298,
+      "step": 968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931750679407225e-05,
+      "loss": 0.8582,
+      "step": 969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315748875797647e-05,
+      "loss": 0.8309,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931398877666849e-05,
+      "loss": 0.8916,
+      "step": 971
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312226497096812e-05,
+      "loss": 0.8254,
+      "step": 972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931046203749518e-05,
+      "loss": 0.8616,
+      "step": 973
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308695398276657e-05,
+      "loss": 0.8179,
+      "step": 974
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930692657985482e-05,
+      "loss": 0.8384,
+      "step": 975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930515558264376e-05,
+      "loss": 0.3702,
+      "step": 976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303382407058064e-05,
+      "loss": 0.3599,
+      "step": 977
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301607053512856e-05,
+      "loss": 0.8635,
+      "step": 978
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929982952242374e-05,
+      "loss": 0.8619,
+      "step": 979
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298049814206844e-05,
+      "loss": 0.8408,
+      "step": 980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296267929278813e-05,
+      "loss": 0.8664,
+      "step": 981
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929448386805679e-05,
+      "loss": 0.3437,
+      "step": 982
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292697630958428e-05,
+      "loss": 0.8894,
+      "step": 983
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290909218401896e-05,
+      "loss": 0.7838,
+      "step": 984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289118630805867e-05,
+      "loss": 0.8515,
+      "step": 985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287325868589525e-05,
+      "loss": 0.8068,
+      "step": 986
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928553093217256e-05,
+      "loss": 0.8141,
+      "step": 987
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283733821975185e-05,
+      "loss": 0.82,
+      "step": 988
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281934538418094e-05,
+      "loss": 0.8381,
+      "step": 989
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280133081922518e-05,
+      "loss": 0.8407,
+      "step": 990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278329452910185e-05,
+      "loss": 0.8207,
+      "step": 991
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276523651803326e-05,
+      "loss": 0.8519,
+      "step": 992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927471567902469e-05,
+      "loss": 0.8565,
+      "step": 993
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927290553499753e-05,
+      "loss": 0.8859,
+      "step": 994
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927109322014561e-05,
+      "loss": 0.8885,
+      "step": 995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92692787348932e-05,
+      "loss": 0.8351,
+      "step": 996
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267462079665072e-05,
+      "loss": 0.8406,
+      "step": 997
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926564325488652e-05,
+      "loss": 0.8725,
+      "step": 998
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263822260983337e-05,
+      "loss": 0.3444,
+      "step": 999
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261999098381827e-05,
+      "loss": 0.8633,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926017376750879e-05,
+      "loss": 0.8393,
+      "step": 1001
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925834626879155e-05,
+      "loss": 0.7944,
+      "step": 1002
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256516602657936e-05,
+      "loss": 0.8498,
+      "step": 1003
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254684769536273e-05,
+      "loss": 0.8121,
+      "step": 1004
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252850769855405e-05,
+      "loss": 0.8221,
+      "step": 1005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251014604044674e-05,
+      "loss": 0.7843,
+      "step": 1006
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924917627253394e-05,
+      "loss": 0.87,
+      "step": 1007
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924733577575356e-05,
+      "loss": 0.8551,
+      "step": 1008
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245493114134405e-05,
+      "loss": 0.9195,
+      "step": 1009
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243648288107843e-05,
+      "loss": 0.8086,
+      "step": 1010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241801298105763e-05,
+      "loss": 0.8627,
+      "step": 1011
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239952144560546e-05,
+      "loss": 0.8035,
+      "step": 1012
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923810082790509e-05,
+      "loss": 0.8032,
+      "step": 1013
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92362473485728e-05,
+      "loss": 0.8538,
+      "step": 1014
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234391706997576e-05,
+      "loss": 0.844,
+      "step": 1015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232533903613833e-05,
+      "loss": 0.8396,
+      "step": 1016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923067393885649e-05,
+      "loss": 0.7646,
+      "step": 1017
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228811813160972e-05,
+      "loss": 0.8424,
+      "step": 1018
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226947526963214e-05,
+      "loss": 0.8881,
+      "step": 1019
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225081080699653e-05,
+      "loss": 0.8174,
+      "step": 1020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223212474807224e-05,
+      "loss": 0.8987,
+      "step": 1021
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922134170972338e-05,
+      "loss": 0.7932,
+      "step": 1022
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921946878588608e-05,
+      "loss": 0.3618,
+      "step": 1023
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921759370373377e-05,
+      "loss": 0.8847,
+      "step": 1024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215716463705433e-05,
+      "loss": 0.8541,
+      "step": 1025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213837066240518e-05,
+      "loss": 0.847,
+      "step": 1026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211955511779016e-05,
+      "loss": 0.8624,
+      "step": 1027
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210071800761396e-05,
+      "loss": 0.8304,
+      "step": 1028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208185933628648e-05,
+      "loss": 0.8141,
+      "step": 1029
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206297910822254e-05,
+      "loss": 0.8084,
+      "step": 1030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920440773278422e-05,
+      "loss": 0.8451,
+      "step": 1031
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202515399957037e-05,
+      "loss": 0.7273,
+      "step": 1032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200620912783705e-05,
+      "loss": 0.8526,
+      "step": 1033
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198724271707738e-05,
+      "loss": 0.8537,
+      "step": 1034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919682547717314e-05,
+      "loss": 0.8131,
+      "step": 1035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194924529624434e-05,
+      "loss": 0.7942,
+      "step": 1036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919302142950663e-05,
+      "loss": 0.8142,
+      "step": 1037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919111617726526e-05,
+      "loss": 0.8205,
+      "step": 1038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189208773346354e-05,
+      "loss": 0.8374,
+      "step": 1039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187299218196428e-05,
+      "loss": 0.8432,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185387512262533e-05,
+      "loss": 0.8591,
+      "step": 1041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183473655992198e-05,
+      "loss": 0.7897,
+      "step": 1042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181557649833467e-05,
+      "loss": 0.7927,
+      "step": 1043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179639494234884e-05,
+      "loss": 0.8649,
+      "step": 1044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91777191896455e-05,
+      "loss": 0.8656,
+      "step": 1045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917579673651486e-05,
+      "loss": 0.7953,
+      "step": 1046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173872135293025e-05,
+      "loss": 0.8603,
+      "step": 1047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171945386430547e-05,
+      "loss": 0.7689,
+      "step": 1048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170016490378485e-05,
+      "loss": 0.7277,
+      "step": 1049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168085447588404e-05,
+      "loss": 0.7662,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916615225851237e-05,
+      "loss": 0.8512,
+      "step": 1051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164216923602953e-05,
+      "loss": 0.7854,
+      "step": 1052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162279443313213e-05,
+      "loss": 0.7793,
+      "step": 1053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916033981809673e-05,
+      "loss": 0.7744,
+      "step": 1054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158398048407574e-05,
+      "loss": 0.7915,
+      "step": 1055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156454134700325e-05,
+      "loss": 0.8062,
+      "step": 1056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154508077430063e-05,
+      "loss": 0.7689,
+      "step": 1057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152559877052355e-05,
+      "loss": 0.7755,
+      "step": 1058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91506095340233e-05,
+      "loss": 0.7682,
+      "step": 1059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914865704879947e-05,
+      "loss": 0.8522,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146702421837952e-05,
+      "loss": 0.8588,
+      "step": 1061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144745653596334e-05,
+      "loss": 0.857,
+      "step": 1062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142786744532698e-05,
+      "loss": 0.8346,
+      "step": 1063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140825695105638e-05,
+      "loss": 0.7853,
+      "step": 1064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138862505774245e-05,
+      "loss": 0.9023,
+      "step": 1065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136897176998107e-05,
+      "loss": 0.8588,
+      "step": 1066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913492970923731e-05,
+      "loss": 0.8382,
+      "step": 1067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132960102952455e-05,
+      "loss": 0.8693,
+      "step": 1068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913098835860463e-05,
+      "loss": 0.8664,
+      "step": 1069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129014476655432e-05,
+      "loss": 0.8949,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912703845756695e-05,
+      "loss": 0.8687,
+      "step": 1071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912506030180178e-05,
+      "loss": 0.7449,
+      "step": 1072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912308000982302e-05,
+      "loss": 0.8453,
+      "step": 1073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121097582094256e-05,
+      "loss": 0.867,
+      "step": 1074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911911301907959e-05,
+      "loss": 0.8295,
+      "step": 1075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117126321243613e-05,
+      "loss": 0.8235,
+      "step": 1076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911513748905142e-05,
+      "loss": 0.8113,
+      "step": 1077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113146522968605e-05,
+      "loss": 0.3556,
+      "step": 1078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911115342346126e-05,
+      "loss": 0.8429,
+      "step": 1079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109158190995977e-05,
+      "loss": 0.8205,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910716082603985e-05,
+      "loss": 0.8547,
+      "step": 1081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910516132906047e-05,
+      "loss": 0.8429,
+      "step": 1082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103159700525926e-05,
+      "loss": 0.8234,
+      "step": 1083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910115594090481e-05,
+      "loss": 0.8522,
+      "step": 1084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099150050666206e-05,
+      "loss": 0.8164,
+      "step": 1085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097142030279705e-05,
+      "loss": 0.7993,
+      "step": 1086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909513188021539e-05,
+      "loss": 0.7966,
+      "step": 1087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093119600943846e-05,
+      "loss": 0.8305,
+      "step": 1088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091105192936156e-05,
+      "loss": 0.7921,
+      "step": 1089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089088656663902e-05,
+      "loss": 0.8661,
+      "step": 1090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908706999259916e-05,
+      "loss": 0.8855,
+      "step": 1091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085049201214515e-05,
+      "loss": 0.8013,
+      "step": 1092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083026282983036e-05,
+      "loss": 0.865,
+      "step": 1093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081001238378295e-05,
+      "loss": 0.8422,
+      "step": 1094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907897406787437e-05,
+      "loss": 0.7748,
+      "step": 1095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076944771945825e-05,
+      "loss": 0.8181,
+      "step": 1096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074913351067724e-05,
+      "loss": 0.8469,
+      "step": 1097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072879805715633e-05,
+      "loss": 0.8176,
+      "step": 1098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070844136365617e-05,
+      "loss": 0.7928,
+      "step": 1099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906880634349423e-05,
+      "loss": 0.7983,
+      "step": 1100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066766427578534e-05,
+      "loss": 0.8402,
+      "step": 1101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064724389096073e-05,
+      "loss": 0.8433,
+      "step": 1102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90626802285249e-05,
+      "loss": 0.803,
+      "step": 1103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906063394634356e-05,
+      "loss": 0.7808,
+      "step": 1104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90585855430311e-05,
+      "loss": 0.8624,
+      "step": 1105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056535019067056e-05,
+      "loss": 0.8614,
+      "step": 1106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905448237493147e-05,
+      "loss": 0.8147,
+      "step": 1107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052427611104864e-05,
+      "loss": 0.8138,
+      "step": 1108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050370728068272e-05,
+      "loss": 0.793,
+      "step": 1109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904831172630322e-05,
+      "loss": 0.3322,
+      "step": 1110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046250606291724e-05,
+      "loss": 0.8291,
+      "step": 1111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044187368516308e-05,
+      "loss": 0.8558,
+      "step": 1112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042122013459976e-05,
+      "loss": 0.7828,
+      "step": 1113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040054541606245e-05,
+      "loss": 0.8321,
+      "step": 1114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903798495343911e-05,
+      "loss": 0.86,
+      "step": 1115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035913249443072e-05,
+      "loss": 0.8199,
+      "step": 1116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033839430103128e-05,
+      "loss": 0.7795,
+      "step": 1117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031763495904768e-05,
+      "loss": 0.8893,
+      "step": 1118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029685447333977e-05,
+      "loss": 0.7231,
+      "step": 1119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902760528487723e-05,
+      "loss": 0.8282,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90255230090215e-05,
+      "loss": 0.8015,
+      "step": 1121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023438620254264e-05,
+      "loss": 0.8484,
+      "step": 1122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021352119063484e-05,
+      "loss": 0.8351,
+      "step": 1123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019263505937612e-05,
+      "loss": 0.8688,
+      "step": 1124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017172781365608e-05,
+      "loss": 0.7958,
+      "step": 1125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015079945836918e-05,
+      "loss": 0.8445,
+      "step": 1126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901298499984148e-05,
+      "loss": 0.8363,
+      "step": 1127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901088794386973e-05,
+      "loss": 0.8639,
+      "step": 1128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90087887784126e-05,
+      "loss": 0.824,
+      "step": 1129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006687503961507e-05,
+      "loss": 0.8086,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004584121008378e-05,
+      "loss": 0.8302,
+      "step": 1131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002478630045616e-05,
+      "loss": 0.7994,
+      "step": 1132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000371031566125e-05,
+      "loss": 0.7751,
+      "step": 1133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998261326063304e-05,
+      "loss": 0.8553,
+      "step": 1134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996149514031045e-05,
+      "loss": 0.8468,
+      "step": 1135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899403559596373e-05,
+      "loss": 0.8384,
+      "step": 1136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991919572356238e-05,
+      "loss": 0.3663,
+      "step": 1137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989801443703934e-05,
+      "loss": 0.8899,
+      "step": 1138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987681210502686e-05,
+      "loss": 0.7791,
+      "step": 1139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985558873248842e-05,
+      "loss": 0.8444,
+      "step": 1140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898343443243926e-05,
+      "loss": 0.7494,
+      "step": 1141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898130788857127e-05,
+      "loss": 0.7618,
+      "step": 1142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897917924214271e-05,
+      "loss": 0.8377,
+      "step": 1143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977048493651906e-05,
+      "loss": 0.8315,
+      "step": 1144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897491564359767e-05,
+      "loss": 0.3366,
+      "step": 1145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972780692479315e-05,
+      "loss": 0.7391,
+      "step": 1146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970643640796642e-05,
+      "loss": 0.7439,
+      "step": 1147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896850448904994e-05,
+      "loss": 0.7773,
+      "step": 1148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966363237739994e-05,
+      "loss": 0.8072,
+      "step": 1149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964219887368083e-05,
+      "loss": 0.7569,
+      "step": 1150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962074438435968e-05,
+      "loss": 0.8144,
+      "step": 1151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959926891445913e-05,
+      "loss": 0.7933,
+      "step": 1152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957777246900665e-05,
+      "loss": 0.7834,
+      "step": 1153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955625505303465e-05,
+      "loss": 0.8123,
+      "step": 1154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895347166715804e-05,
+      "loss": 0.7862,
+      "step": 1155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895131573296862e-05,
+      "loss": 0.8519,
+      "step": 1156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949157703239916e-05,
+      "loss": 0.8652,
+      "step": 1157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946997578477125e-05,
+      "loss": 0.7597,
+      "step": 1158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894483535918595e-05,
+      "loss": 0.8225,
+      "step": 1159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942671045872568e-05,
+      "loss": 0.8725,
+      "step": 1160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894050463904366e-05,
+      "loss": 0.8122,
+      "step": 1161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938336139206385e-05,
+      "loss": 0.7007,
+      "step": 1162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936165546868395e-05,
+      "loss": 0.8243,
+      "step": 1163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933992862537845e-05,
+      "loss": 0.7735,
+      "step": 1164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931818086723363e-05,
+      "loss": 0.7479,
+      "step": 1165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892964121993407e-05,
+      "loss": 0.7764,
+      "step": 1166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927462262679585e-05,
+      "loss": 0.8332,
+      "step": 1167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925281215470008e-05,
+      "loss": 0.7195,
+      "step": 1168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923098078815934e-05,
+      "loss": 0.7996,
+      "step": 1169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920912853228437e-05,
+      "loss": 0.8363,
+      "step": 1170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918725539219095e-05,
+      "loss": 0.7881,
+      "step": 1171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891653613729996e-05,
+      "loss": 0.8312,
+      "step": 1172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914344647983586e-05,
+      "loss": 0.7975,
+      "step": 1173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912151071783004e-05,
+      "loss": 0.8138,
+      "step": 1174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909955409211748e-05,
+      "loss": 0.761,
+      "step": 1175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890775766078382e-05,
+      "loss": 0.3784,
+      "step": 1176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905557827013733e-05,
+      "loss": 0.848,
+      "step": 1177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890335590841647e-05,
+      "loss": 0.8625,
+      "step": 1178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901151905507505e-05,
+      "loss": 0.7509,
+      "step": 1179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898945818802816e-05,
+      "loss": 0.8165,
+      "step": 1180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896737648818845e-05,
+      "loss": 0.8306,
+      "step": 1181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894527396072542e-05,
+      "loss": 0.7257,
+      "step": 1182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889231506108133e-05,
+      "loss": 0.8319,
+      "step": 1183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889010064436313e-05,
+      "loss": 0.7491,
+      "step": 1184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887884146436346e-05,
+      "loss": 0.7683,
+      "step": 1185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885665567819868e-05,
+      "loss": 0.7864,
+      "step": 1186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883444909033075e-05,
+      "loss": 0.8669,
+      "step": 1187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888122217059583e-05,
+      "loss": 0.847,
+      "step": 1188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878997353028484e-05,
+      "loss": 0.8002,
+      "step": 1189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887677045685188e-05,
+      "loss": 0.7578,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874541482587338e-05,
+      "loss": 0.815,
+      "step": 1191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887231043075668e-05,
+      "loss": 0.868,
+      "step": 1192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887007730188219e-05,
+      "loss": 0.3642,
+      "step": 1193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867842096486666e-05,
+      "loss": 0.7756,
+      "step": 1194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8865604815093373e-05,
+      "loss": 0.8363,
+      "step": 1195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863365458226065e-05,
+      "loss": 0.8234,
+      "step": 1196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861124026408987e-05,
+      "loss": 0.9086,
+      "step": 1197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858880520166865e-05,
+      "loss": 0.8666,
+      "step": 1198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885663494002492e-05,
+      "loss": 0.8566,
+      "step": 1199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854387286508845e-05,
+      "loss": 0.7528,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852137560144826e-05,
+      "loss": 0.7937,
+      "step": 1201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849885761459532e-05,
+      "loss": 0.8884,
+      "step": 1202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884763189098012e-05,
+      "loss": 0.7732,
+      "step": 1203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884537594923423e-05,
+      "loss": 0.8384,
+      "step": 1204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843117936749985e-05,
+      "loss": 0.7519,
+      "step": 1205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840857854056e-05,
+      "loss": 0.8012,
+      "step": 1206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838595701681364e-05,
+      "loss": 0.8187,
+      "step": 1207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836331480155662e-05,
+      "loss": 0.848,
+      "step": 1208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883406519000895e-05,
+      "loss": 0.8487,
+      "step": 1209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883179683177178e-05,
+      "loss": 0.8278,
+      "step": 1210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829526405975185e-05,
+      "loss": 0.7875,
+      "step": 1211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882725391315068e-05,
+      "loss": 0.7925,
+      "step": 1212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882497935383026e-05,
+      "loss": 0.8712,
+      "step": 1213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822702728546415e-05,
+      "loss": 0.8183,
+      "step": 1214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882042403783211e-05,
+      "loss": 0.817,
+      "step": 1215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818143282220793e-05,
+      "loss": 0.7926,
+      "step": 1216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815860462246402e-05,
+      "loss": 0.8496,
+      "step": 1217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8813575578443354e-05,
+      "loss": 0.783,
+      "step": 1218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881128863134655e-05,
+      "loss": 0.8361,
+      "step": 1219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880899962149137e-05,
+      "loss": 0.8695,
+      "step": 1220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806708549413684e-05,
+      "loss": 0.7349,
+      "step": 1221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804415415649838e-05,
+      "loss": 0.3793,
+      "step": 1222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802120220736666e-05,
+      "loss": 0.7763,
+      "step": 1223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879982296521148e-05,
+      "loss": 0.3256,
+      "step": 1224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797523649612085e-05,
+      "loss": 0.8357,
+      "step": 1225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879522227447675e-05,
+      "loss": 0.8165,
+      "step": 1226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879291884034424e-05,
+      "loss": 0.7844,
+      "step": 1227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87906133477538e-05,
+      "loss": 0.8014,
+      "step": 1228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788305797245155e-05,
+      "loss": 0.3607,
+      "step": 1229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878599618935851e-05,
+      "loss": 0.7469,
+      "step": 1230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783684524634553e-05,
+      "loss": 0.7628,
+      "step": 1231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781370803614456e-05,
+      "loss": 0.8103,
+      "step": 1232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877905502683987e-05,
+      "loss": 0.8453,
+      "step": 1233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776737194852927e-05,
+      "loss": 0.6707,
+      "step": 1234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774417308196242e-05,
+      "loss": 0.8322,
+      "step": 1235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877209536741291e-05,
+      "loss": 0.796,
+      "step": 1236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769771373046505e-05,
+      "loss": 0.7726,
+      "step": 1237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767445325641084e-05,
+      "loss": 0.7904,
+      "step": 1238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765117225741186e-05,
+      "loss": 0.8155,
+      "step": 1239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762787073891828e-05,
+      "loss": 0.7694,
+      "step": 1240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760454870638507e-05,
+      "loss": 0.9164,
+      "step": 1241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87581206165272e-05,
+      "loss": 0.848,
+      "step": 1242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755784312104367e-05,
+      "loss": 0.8247,
+      "step": 1243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875344595791695e-05,
+      "loss": 0.7054,
+      "step": 1244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751105554512356e-05,
+      "loss": 0.7356,
+      "step": 1245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748763102438496e-05,
+      "loss": 0.7856,
+      "step": 1246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746418602243742e-05,
+      "loss": 0.8509,
+      "step": 1247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744072054476947e-05,
+      "loss": 0.766,
+      "step": 1248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741723459687454e-05,
+      "loss": 0.8832,
+      "step": 1249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739372818425076e-05,
+      "loss": 0.861,
+      "step": 1250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737020131240103e-05,
+      "loss": 0.8111,
+      "step": 1251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734665398683317e-05,
+      "loss": 0.7925,
+      "step": 1252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732308621305967e-05,
+      "loss": 0.8609,
+      "step": 1253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8729949799659783e-05,
+      "loss": 0.8251,
+      "step": 1254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727588934296972e-05,
+      "loss": 0.8524,
+      "step": 1255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725226025770228e-05,
+      "loss": 0.8391,
+      "step": 1256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722861074632717e-05,
+      "loss": 0.7455,
+      "step": 1257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872049408143808e-05,
+      "loss": 0.8019,
+      "step": 1258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718125046740438e-05,
+      "loss": 0.8215,
+      "step": 1259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715753971094396e-05,
+      "loss": 0.9197,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713380855055035e-05,
+      "loss": 0.802,
+      "step": 1261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711005699177903e-05,
+      "loss": 0.8208,
+      "step": 1262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870862850401904e-05,
+      "loss": 0.7601,
+      "step": 1263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706249270134956e-05,
+      "loss": 0.7932,
+      "step": 1264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703867998082637e-05,
+      "loss": 0.8028,
+      "step": 1265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870148468841955e-05,
+      "loss": 0.8007,
+      "step": 1266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699099341703634e-05,
+      "loss": 0.7394,
+      "step": 1267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696711958493315e-05,
+      "loss": 0.8165,
+      "step": 1268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694322539347486e-05,
+      "loss": 0.7601,
+      "step": 1269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691931084825516e-05,
+      "loss": 0.8422,
+      "step": 1270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868953759548726e-05,
+      "loss": 0.8967,
+      "step": 1271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868714207189304e-05,
+      "loss": 0.7795,
+      "step": 1272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868474451460366e-05,
+      "loss": 0.8313,
+      "step": 1273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682344924180396e-05,
+      "loss": 0.8526,
+      "step": 1274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679943301185002e-05,
+      "loss": 0.8643,
+      "step": 1275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677539646179706e-05,
+      "loss": 0.8003,
+      "step": 1276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675133959727222e-05,
+      "loss": 0.7753,
+      "step": 1277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867272624239072e-05,
+      "loss": 0.7887,
+      "step": 1278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867031649473386e-05,
+      "loss": 0.7343,
+      "step": 1279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667904717320774e-05,
+      "loss": 0.8914,
+      "step": 1280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665490910716075e-05,
+      "loss": 0.8214,
+      "step": 1281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663075075484834e-05,
+      "loss": 0.8833,
+      "step": 1282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8660657212192617e-05,
+      "loss": 0.8621,
+      "step": 1283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658237321405447e-05,
+      "loss": 0.8276,
+      "step": 1284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655815403689837e-05,
+      "loss": 0.8729,
+      "step": 1285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653391459612768e-05,
+      "loss": 0.8129,
+      "step": 1286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650965489741693e-05,
+      "loss": 0.3774,
+      "step": 1287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8648537494644538e-05,
+      "loss": 0.7873,
+      "step": 1288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646107474889713e-05,
+      "loss": 0.8614,
+      "step": 1289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864367543104609e-05,
+      "loss": 0.8138,
+      "step": 1290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864124136368302e-05,
+      "loss": 0.8812,
+      "step": 1291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638805273370338e-05,
+      "loss": 0.7995,
+      "step": 1292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636367160678328e-05,
+      "loss": 0.892,
+      "step": 1293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633927026177772e-05,
+      "loss": 0.7907,
+      "step": 1294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631484870439912e-05,
+      "loss": 0.8144,
+      "step": 1295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862904069403647e-05,
+      "loss": 0.777,
+      "step": 1296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626594497539634e-05,
+      "loss": 0.368,
+      "step": 1297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624146281522065e-05,
+      "loss": 0.7932,
+      "step": 1298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862169604655691e-05,
+      "loss": 0.7844,
+      "step": 1299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861924379321777e-05,
+      "loss": 0.7446,
+      "step": 1300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861678952207873e-05,
+      "loss": 0.821,
+      "step": 1301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861433323371435e-05,
+      "loss": 0.8284,
+      "step": 1302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611874928699652e-05,
+      "loss": 0.858,
+      "step": 1303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609414607610133e-05,
+      "loss": 0.8034,
+      "step": 1304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8606952271021766e-05,
+      "loss": 0.7571,
+      "step": 1305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604487919511e-05,
+      "loss": 0.7921,
+      "step": 1306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860202155365474e-05,
+      "loss": 0.8462,
+      "step": 1307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599553174030382e-05,
+      "loss": 0.8092,
+      "step": 1308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597082781215776e-05,
+      "loss": 0.8126,
+      "step": 1309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594610375789254e-05,
+      "loss": 0.8161,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592135958329617e-05,
+      "loss": 0.8811,
+      "step": 1311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589659529416135e-05,
+      "loss": 0.8341,
+      "step": 1312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587181089628546e-05,
+      "loss": 0.7127,
+      "step": 1313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584700639547072e-05,
+      "loss": 0.8095,
+      "step": 1314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582218179752388e-05,
+      "loss": 0.3482,
+      "step": 1315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579733710825653e-05,
+      "loss": 0.7657,
+      "step": 1316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577247233348487e-05,
+      "loss": 0.8292,
+      "step": 1317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574758747902986e-05,
+      "loss": 0.8334,
+      "step": 1318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572268255071718e-05,
+      "loss": 0.8087,
+      "step": 1319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569775755437716e-05,
+      "loss": 0.8322,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856728124958448e-05,
+      "loss": 0.8171,
+      "step": 1321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564784738095987e-05,
+      "loss": 0.7688,
+      "step": 1322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8562286221556682e-05,
+      "loss": 0.7768,
+      "step": 1323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559785700551474e-05,
+      "loss": 0.8386,
+      "step": 1324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855728317566575e-05,
+      "loss": 0.8012,
+      "step": 1325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855477864748536e-05,
+      "loss": 0.8405,
+      "step": 1326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855227211659662e-05,
+      "loss": 0.8754,
+      "step": 1327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549763583586323e-05,
+      "loss": 0.7708,
+      "step": 1328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547253049041726e-05,
+      "loss": 0.8334,
+      "step": 1329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854474051355055e-05,
+      "loss": 0.8259,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542225977701002e-05,
+      "loss": 0.805,
+      "step": 1331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539709442081734e-05,
+      "loss": 0.7797,
+      "step": 1332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853719090728188e-05,
+      "loss": 0.7883,
+      "step": 1333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853467037389104e-05,
+      "loss": 0.8406,
+      "step": 1334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532147842499285e-05,
+      "loss": 0.7588,
+      "step": 1335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529623313697143e-05,
+      "loss": 0.8192,
+      "step": 1336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852709678807562e-05,
+      "loss": 0.8416,
+      "step": 1337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524568266226184e-05,
+      "loss": 0.836,
+      "step": 1338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522037748740772e-05,
+      "loss": 0.9068,
+      "step": 1339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8519505236211795e-05,
+      "loss": 0.7714,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516970729232115e-05,
+      "loss": 0.8024,
+      "step": 1341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851443422839508e-05,
+      "loss": 0.8846,
+      "step": 1342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511895734294483e-05,
+      "loss": 0.8334,
+      "step": 1343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509355247524606e-05,
+      "loss": 0.7735,
+      "step": 1344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506812768680184e-05,
+      "loss": 0.8227,
+      "step": 1345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850426829835642e-05,
+      "loss": 0.8203,
+      "step": 1346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501721837148988e-05,
+      "loss": 0.7576,
+      "step": 1347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849917338565402e-05,
+      "loss": 0.8735,
+      "step": 1348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496622944468125e-05,
+      "loss": 0.7509,
+      "step": 1349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8494070514188367e-05,
+      "loss": 0.889,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491516095412283e-05,
+      "loss": 0.7803,
+      "step": 1351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488959688737868e-05,
+      "loss": 0.8177,
+      "step": 1352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486401294763596e-05,
+      "loss": 0.8929,
+      "step": 1353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848384091408839e-05,
+      "loss": 0.7896,
+      "step": 1354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481278547311645e-05,
+      "loss": 0.855,
+      "step": 1355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478714195033227e-05,
+      "loss": 0.8501,
+      "step": 1356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847614785785346e-05,
+      "loss": 0.7415,
+      "step": 1357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847357953637313e-05,
+      "loss": 0.8279,
+      "step": 1358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471009231193494e-05,
+      "loss": 0.7959,
+      "step": 1359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468436942916266e-05,
+      "loss": 0.8186,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846586267214364e-05,
+      "loss": 0.7862,
+      "step": 1361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463286419478256e-05,
+      "loss": 0.7878,
+      "step": 1362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460708185523227e-05,
+      "loss": 0.799,
+      "step": 1363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458127970882126e-05,
+      "loss": 0.8126,
+      "step": 1364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845554577615899e-05,
+      "loss": 0.7641,
+      "step": 1365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452961601958327e-05,
+      "loss": 0.8433,
+      "step": 1366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450375448885097e-05,
+      "loss": 0.8751,
+      "step": 1367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447787317544735e-05,
+      "loss": 0.8512,
+      "step": 1368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844519720854313e-05,
+      "loss": 0.8677,
+      "step": 1369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8442605122486635e-05,
+      "loss": 0.8509,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844001105998207e-05,
+      "loss": 0.3731,
+      "step": 1371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8437415021636713e-05,
+      "loss": 0.8115,
+      "step": 1372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843481700805831e-05,
+      "loss": 0.7887,
+      "step": 1373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432217019855068e-05,
+      "loss": 0.3494,
+      "step": 1374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842961505763565e-05,
+      "loss": 0.8093,
+      "step": 1375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842701112200919e-05,
+      "loss": 0.3428,
+      "step": 1376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842440521358528e-05,
+      "loss": 0.7935,
+      "step": 1377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421797332973966e-05,
+      "loss": 0.8763,
+      "step": 1378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419187480785773e-05,
+      "loss": 0.8319,
+      "step": 1379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416575657631678e-05,
+      "loss": 0.7852,
+      "step": 1380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841396186412311e-05,
+      "loss": 0.7936,
+      "step": 1381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411346100871975e-05,
+      "loss": 0.8139,
+      "step": 1382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408728368490637e-05,
+      "loss": 0.7646,
+      "step": 1383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406108667591913e-05,
+      "loss": 0.7788,
+      "step": 1384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403486998789084e-05,
+      "loss": 0.8265,
+      "step": 1385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400863362695897e-05,
+      "loss": 0.8112,
+      "step": 1386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8398237759926553e-05,
+      "loss": 0.7643,
+      "step": 1387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839561019109572e-05,
+      "loss": 0.7716,
+      "step": 1388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839298065681852e-05,
+      "loss": 0.8465,
+      "step": 1389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8390349157710538e-05,
+      "loss": 0.8241,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838771569438782e-05,
+      "loss": 0.8965,
+      "step": 1391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838508026746687e-05,
+      "loss": 0.7635,
+      "step": 1392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382442877564647e-05,
+      "loss": 0.7318,
+      "step": 1393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837980352529858e-05,
+      "loss": 0.8143,
+      "step": 1394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837716221128655e-05,
+      "loss": 0.8659,
+      "step": 1395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374518936146906e-05,
+      "loss": 0.8452,
+      "step": 1396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837187370049844e-05,
+      "loss": 0.7031,
+      "step": 1397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369226504960414e-05,
+      "loss": 0.8007,
+      "step": 1398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366577350152553e-05,
+      "loss": 0.7723,
+      "step": 1399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836392623669503e-05,
+      "loss": 0.8869,
+      "step": 1400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836127316520848e-05,
+      "loss": 0.8204,
+      "step": 1401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358618136314e-05,
+      "loss": 0.8456,
+      "step": 1402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835596115063315e-05,
+      "loss": 0.8053,
+      "step": 1403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353302208787927e-05,
+      "loss": 0.8376,
+      "step": 1404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350641311400813e-05,
+      "loss": 0.7472,
+      "step": 1405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347978459094725e-05,
+      "loss": 0.8256,
+      "step": 1406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834531365249305e-05,
+      "loss": 0.822,
+      "step": 1407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8342646892219636e-05,
+      "loss": 0.8198,
+      "step": 1408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833997817889878e-05,
+      "loss": 0.8041,
+      "step": 1409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337307513155237e-05,
+      "loss": 0.8318,
+      "step": 1410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833463489561422e-05,
+      "loss": 0.8335,
+      "step": 1411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331960326901396e-05,
+      "loss": 0.8635,
+      "step": 1412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83292838076429e-05,
+      "loss": 0.8467,
+      "step": 1413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832660533846531e-05,
+      "loss": 0.7968,
+      "step": 1414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832392491999567e-05,
+      "loss": 0.779,
+      "step": 1415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832124255286148e-05,
+      "loss": 0.8116,
+      "step": 1416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318558237690685e-05,
+      "loss": 0.7617,
+      "step": 1417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315871975111697e-05,
+      "loss": 0.8093,
+      "step": 1418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313183765753383e-05,
+      "loss": 0.7729,
+      "step": 1419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310493610245068e-05,
+      "loss": 0.7372,
+      "step": 1420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307801509216513e-05,
+      "loss": 0.8045,
+      "step": 1421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305107463297967e-05,
+      "loss": 0.8188,
+      "step": 1422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830241147312011e-05,
+      "loss": 0.3798,
+      "step": 1423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829971353931408e-05,
+      "loss": 0.7689,
+      "step": 1424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829701366251148e-05,
+      "loss": 0.8328,
+      "step": 1425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294311843344358e-05,
+      "loss": 0.7929,
+      "step": 1426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291608082445225e-05,
+      "loss": 0.8284,
+      "step": 1427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828890238044704e-05,
+      "loss": 0.8543,
+      "step": 1428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828619473798322e-05,
+      "loss": 0.8154,
+      "step": 1429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283485155687635e-05,
+      "loss": 0.7877,
+      "step": 1430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280773634194607e-05,
+      "loss": 0.8701,
+      "step": 1431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8278060174138916e-05,
+      "loss": 0.8196,
+      "step": 1432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275344776155793e-05,
+      "loss": 0.8798,
+      "step": 1433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827262744088093e-05,
+      "loss": 0.8165,
+      "step": 1434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269908168950453e-05,
+      "loss": 0.7508,
+      "step": 1435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267186961000965e-05,
+      "loss": 0.8535,
+      "step": 1436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826446381766951e-05,
+      "loss": 0.8574,
+      "step": 1437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261738739593583e-05,
+      "loss": 0.8486,
+      "step": 1438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8259011727411142e-05,
+      "loss": 0.8547,
+      "step": 1439
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825628278176059e-05,
+      "loss": 0.8443,
+      "step": 1440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253551903280777e-05,
+      "loss": 0.7917,
+      "step": 1441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825081909261102e-05,
+      "loss": 0.8074,
+      "step": 1442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824808435039108e-05,
+      "loss": 0.818,
+      "step": 1443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245347677261173e-05,
+      "loss": 0.8439,
+      "step": 1444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824260907386196e-05,
+      "loss": 0.3858,
+      "step": 1445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239868540834562e-05,
+      "loss": 0.8303,
+      "step": 1446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823712607882055e-05,
+      "loss": 0.754,
+      "step": 1447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234381688461943e-05,
+      "loss": 0.7977,
+      "step": 1448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823163537040122e-05,
+      "loss": 0.8141,
+      "step": 1449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228887125281294e-05,
+      "loss": 0.3514,
+      "step": 1450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822613695374555e-05,
+      "loss": 0.7631,
+      "step": 1451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223384856437807e-05,
+      "loss": 0.8134,
+      "step": 1452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822063083400235e-05,
+      "loss": 0.8508,
+      "step": 1453
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217874887083903e-05,
+      "loss": 0.8144,
+      "step": 1454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215117016327645e-05,
+      "loss": 0.7777,
+      "step": 1455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82123572223792e-05,
+      "loss": 0.741,
+      "step": 1456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209595505884653e-05,
+      "loss": 0.8505,
+      "step": 1457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206831867490536e-05,
+      "loss": 0.8655,
+      "step": 1458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204066307843816e-05,
+      "loss": 0.9254,
+      "step": 1459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820129882759193e-05,
+      "loss": 0.7585,
+      "step": 1460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198529427382757e-05,
+      "loss": 0.8389,
+      "step": 1461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195758107864626e-05,
+      "loss": 0.8177,
+      "step": 1462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192984869686304e-05,
+      "loss": 0.8426,
+      "step": 1463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190209713497026e-05,
+      "loss": 0.7913,
+      "step": 1464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818743263994647e-05,
+      "loss": 0.8,
+      "step": 1465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818465364968475e-05,
+      "loss": 0.7836,
+      "step": 1466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818187274336245e-05,
+      "loss": 0.8414,
+      "step": 1467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817908992163058e-05,
+      "loss": 0.8869,
+      "step": 1468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817630518514062e-05,
+      "loss": 0.3751,
+      "step": 1469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173518534544486e-05,
+      "loss": 0.8004,
+      "step": 1470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817072997049454e-05,
+      "loss": 0.8574,
+      "step": 1471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167939493643594e-05,
+      "loss": 0.819,
+      "step": 1472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816514710464492e-05,
+      "loss": 0.3389,
+      "step": 1473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816235280415222e-05,
+      "loss": 0.8565,
+      "step": 1474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159556592819648e-05,
+      "loss": 0.7719,
+      "step": 1475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815675847130182e-05,
+      "loss": 0.8377,
+      "step": 1476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153958440253778e-05,
+      "loss": 0.7824,
+      "step": 1477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151156500331025e-05,
+      "loss": 0.813,
+      "step": 1478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148352652189504e-05,
+      "loss": 0.7387,
+      "step": 1479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814554689648561e-05,
+      "loss": 0.809,
+      "step": 1480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814273923387618e-05,
+      "loss": 0.8452,
+      "step": 1481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139929665018498e-05,
+      "loss": 0.703,
+      "step": 1482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137118190570296e-05,
+      "loss": 0.7613,
+      "step": 1483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8134304811189753e-05,
+      "loss": 0.811,
+      "step": 1484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813148952753549e-05,
+      "loss": 0.8138,
+      "step": 1485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812867234026658e-05,
+      "loss": 0.7679,
+      "step": 1486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125853250042537e-05,
+      "loss": 0.351,
+      "step": 1487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123032257523315e-05,
+      "loss": 0.8925,
+      "step": 1488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120209363369333e-05,
+      "loss": 0.7317,
+      "step": 1489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811738456824143e-05,
+      "loss": 0.8634,
+      "step": 1490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114557872800906e-05,
+      "loss": 0.7872,
+      "step": 1491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111729277709504e-05,
+      "loss": 0.8221,
+      "step": 1492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108898783629404e-05,
+      "loss": 0.8574,
+      "step": 1493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106066391223244e-05,
+      "loss": 0.7988,
+      "step": 1494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810323210115409e-05,
+      "loss": 0.8134,
+      "step": 1495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100395914085468e-05,
+      "loss": 0.8263,
+      "step": 1496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097557830681335e-05,
+      "loss": 0.8296,
+      "step": 1497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094717851606103e-05,
+      "loss": 0.8336,
+      "step": 1498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809187597752462e-05,
+      "loss": 0.8411,
+      "step": 1499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089032209102182e-05,
+      "loss": 0.7787,
+      "step": 1500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086186547004524e-05,
+      "loss": 0.8288,
+      "step": 1501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808333899189783e-05,
+      "loss": 0.8675,
+      "step": 1502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808048954444872e-05,
+      "loss": 0.7835,
+      "step": 1503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077638205324264e-05,
+      "loss": 0.8121,
+      "step": 1504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074784975191974e-05,
+      "loss": 0.8325,
+      "step": 1505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80719298547198e-05,
+      "loss": 0.7818,
+      "step": 1506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069072844576138e-05,
+      "loss": 0.7963,
+      "step": 1507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066213945429823e-05,
+      "loss": 0.8885,
+      "step": 1508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806335315795014e-05,
+      "loss": 0.7039,
+      "step": 1509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8060490482806805e-05,
+      "loss": 0.8047,
+      "step": 1510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057625920669987e-05,
+      "loss": 0.7969,
+      "step": 1511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805475947221029e-05,
+      "loss": 0.8088,
+      "step": 1512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051891138098762e-05,
+      "loss": 0.8047,
+      "step": 1513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049020919006886e-05,
+      "loss": 0.7992,
+      "step": 1514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80461488156066e-05,
+      "loss": 0.8069,
+      "step": 1515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804327482857027e-05,
+      "loss": 0.8038,
+      "step": 1516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040398958570712e-05,
+      "loss": 0.7617,
+      "step": 1517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037521206281178e-05,
+      "loss": 0.8386,
+      "step": 1518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034641572375356e-05,
+      "loss": 0.7927,
+      "step": 1519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031760057527384e-05,
+      "loss": 0.7648,
+      "step": 1520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028876662411837e-05,
+      "loss": 0.7877,
+      "step": 1521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802599138770373e-05,
+      "loss": 0.8858,
+      "step": 1522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8023104234078515e-05,
+      "loss": 0.7183,
+      "step": 1523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020215202212086e-05,
+      "loss": 0.8865,
+      "step": 1524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801732429278078e-05,
+      "loss": 0.8801,
+      "step": 1525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014431506461368e-05,
+      "loss": 0.729,
+      "step": 1526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011536843931066e-05,
+      "loss": 0.8017,
+      "step": 1527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008640305867522e-05,
+      "loss": 0.7812,
+      "step": 1528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005741892948833e-05,
+      "loss": 0.8051,
+      "step": 1529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800284160585352e-05,
+      "loss": 0.8389,
+      "step": 1530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999939445260562e-05,
+      "loss": 0.8154,
+      "step": 1531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799703541184936e-05,
+      "loss": 0.9053,
+      "step": 1532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994129506299765e-05,
+      "loss": 0.8107,
+      "step": 1533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799122172929206e-05,
+      "loss": 0.8892,
+      "step": 1534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988312081506963e-05,
+      "loss": 0.8786,
+      "step": 1535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985400563625636e-05,
+      "loss": 0.7931,
+      "step": 1536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798248717632968e-05,
+      "loss": 0.8247,
+      "step": 1537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797957192030113e-05,
+      "loss": 0.799,
+      "step": 1538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797665479622246e-05,
+      "loss": 0.8261,
+      "step": 1539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7973735804776577e-05,
+      "loss": 0.7796,
+      "step": 1540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797081494664683e-05,
+      "loss": 0.7876,
+      "step": 1541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967892222517008e-05,
+      "loss": 0.7968,
+      "step": 1542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964967633071325e-05,
+      "loss": 0.7708,
+      "step": 1543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962041178994444e-05,
+      "loss": 0.787,
+      "step": 1544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959112860971462e-05,
+      "loss": 0.7498,
+      "step": 1545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956182679687907e-05,
+      "loss": 0.8034,
+      "step": 1546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953250635829744e-05,
+      "loss": 0.8323,
+      "step": 1547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795031673008338e-05,
+      "loss": 0.875,
+      "step": 1548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794738096313565e-05,
+      "loss": 0.777,
+      "step": 1549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944443335673836e-05,
+      "loss": 0.7738,
+      "step": 1550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794150384838564e-05,
+      "loss": 0.8382,
+      "step": 1551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7938562501959216e-05,
+      "loss": 0.7539,
+      "step": 1552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935619297083142e-05,
+      "loss": 0.7372,
+      "step": 1553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932674234446435e-05,
+      "loss": 0.7192,
+      "step": 1554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792972731473854e-05,
+      "loss": 0.4095,
+      "step": 1555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792677853864935e-05,
+      "loss": 0.8337,
+      "step": 1556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923827906869184e-05,
+      "loss": 0.7461,
+      "step": 1557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920875420088797e-05,
+      "loss": 0.7927,
+      "step": 1558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917921078999376e-05,
+      "loss": 0.8276,
+      "step": 1559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914964884292543e-05,
+      "loss": 0.8612,
+      "step": 1560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791200683666036e-05,
+      "loss": 0.8282,
+      "step": 1561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909046936795312e-05,
+      "loss": 0.7694,
+      "step": 1562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790608518539033e-05,
+      "loss": 0.8271,
+      "step": 1563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790312158313877e-05,
+      "loss": 0.8161,
+      "step": 1564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900156130734417e-05,
+      "loss": 0.8457,
+      "step": 1565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897188828871506e-05,
+      "loss": 0.8296,
+      "step": 1566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894219678244686e-05,
+      "loss": 0.7568,
+      "step": 1567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7891248679549052e-05,
+      "loss": 0.8286,
+      "step": 1568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7888275833480123e-05,
+      "loss": 0.8347,
+      "step": 1569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885301140733856e-05,
+      "loss": 0.8477,
+      "step": 1570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7882324602006643e-05,
+      "loss": 0.7892,
+      "step": 1571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879346217995295e-05,
+      "loss": 0.7818,
+      "step": 1572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7876365989397073e-05,
+      "loss": 0.8049,
+      "step": 1573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7873383916909656e-05,
+      "loss": 0.7331,
+      "step": 1574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870400001231156e-05,
+      "loss": 0.8118,
+      "step": 1575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867414243060125e-05,
+      "loss": 0.7735,
+      "step": 1576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7864426643095537e-05,
+      "loss": 0.7557,
+      "step": 1577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786143720203681e-05,
+      "loss": 0.8556,
+      "step": 1578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785844592058377e-05,
+      "loss": 0.7815,
+      "step": 1579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78554527994367e-05,
+      "loss": 0.8377,
+      "step": 1580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852457839296295e-05,
+      "loss": 0.7701,
+      "step": 1581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784946104086369e-05,
+      "loss": 0.3633,
+      "step": 1582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846462404840453e-05,
+      "loss": 0.7817,
+      "step": 1583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784346193192857e-05,
+      "loss": 0.8639,
+      "step": 1584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840459622830462e-05,
+      "loss": 0.8002,
+      "step": 1585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837455478248985e-05,
+      "loss": 0.8074,
+      "step": 1586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783444949888742e-05,
+      "loss": 0.7934,
+      "step": 1587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831441685449485e-05,
+      "loss": 0.8457,
+      "step": 1588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828432038639313e-05,
+      "loss": 0.7996,
+      "step": 1589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782542055916148e-05,
+      "loss": 0.7642,
+      "step": 1590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822407247720982e-05,
+      "loss": 0.8537,
+      "step": 1591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819392105023255e-05,
+      "loss": 0.8008,
+      "step": 1592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816375131774145e-05,
+      "loss": 0.7535,
+      "step": 1593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7813356328679945e-05,
+      "loss": 0.8387,
+      "step": 1594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7810335696447362e-05,
+      "loss": 0.7363,
+      "step": 1595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780731323578355e-05,
+      "loss": 0.7972,
+      "step": 1596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780428894739607e-05,
+      "loss": 0.8463,
+      "step": 1597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801262831992923e-05,
+      "loss": 0.7749,
+      "step": 1598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798234890282536e-05,
+      "loss": 0.7461,
+      "step": 1599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795205122973757e-05,
+      "loss": 0.7776,
+      "step": 1600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792173530775875e-05,
+      "loss": 0.8554,
+      "step": 1601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778914011439859e-05,
+      "loss": 0.8662,
+      "step": 1602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786104874552045e-05,
+      "loss": 0.8619,
+      "step": 1603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7783067811946795e-05,
+      "loss": 0.7956,
+      "step": 1604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780028927293834e-05,
+      "loss": 0.7172,
+      "step": 1605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7776988221304572e-05,
+      "loss": 0.8665,
+      "step": 1606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773945694690855e-05,
+      "loss": 0.8518,
+      "step": 1607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770901348164947e-05,
+      "loss": 0.8229,
+      "step": 1608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767855182439547e-05,
+      "loss": 0.7781,
+      "step": 1609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776480719822777e-05,
+      "loss": 0.8589,
+      "step": 1610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761757396243164e-05,
+      "loss": 0.7727,
+      "step": 1611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77587057771997e-05,
+      "loss": 0.8191,
+      "step": 1612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7755652341811773e-05,
+      "loss": 0.7847,
+      "step": 1613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7752597090794202e-05,
+      "loss": 0.8215,
+      "step": 1614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774954002486224e-05,
+      "loss": 0.7509,
+      "step": 1615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746481144731556e-05,
+      "loss": 0.7755,
+      "step": 1616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743420451118247e-05,
+      "loss": 0.8711,
+      "step": 1617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740357944738827e-05,
+      "loss": 0.8118,
+      "step": 1618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737293626310247e-05,
+      "loss": 0.8188,
+      "step": 1619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773422749654988e-05,
+      "loss": 0.7965,
+      "step": 1620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731159556175516e-05,
+      "loss": 0.794,
+      "step": 1621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728089805905368e-05,
+      "loss": 0.8016,
+      "step": 1622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772501824645808e-05,
+      "loss": 0.8488,
+      "step": 1623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772194487855272e-05,
+      "loss": 0.7712,
+      "step": 1624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771886970290877e-05,
+      "loss": 0.7769,
+      "step": 1625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715792720246144e-05,
+      "loss": 0.8455,
+      "step": 1626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771271393128518e-05,
+      "loss": 0.8048,
+      "step": 1627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770963333674663e-05,
+      "loss": 0.8078,
+      "step": 1628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7706550937351675e-05,
+      "loss": 0.8106,
+      "step": 1629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703466733821916e-05,
+      "loss": 0.8558,
+      "step": 1630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700380726879375e-05,
+      "loss": 0.7715,
+      "step": 1631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697292917246506e-05,
+      "loss": 0.3676,
+      "step": 1632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7694203305646173e-05,
+      "loss": 0.8021,
+      "step": 1633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769111189280167e-05,
+      "loss": 0.8192,
+      "step": 1634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76880186794367e-05,
+      "loss": 0.3507,
+      "step": 1635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684923666275412e-05,
+      "loss": 0.3783,
+      "step": 1636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7681826854042347e-05,
+      "loss": 0.7976,
+      "step": 1637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7678728243462488e-05,
+      "loss": 0.7801,
+      "step": 1638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675627835261237e-05,
+      "loss": 0.8563,
+      "step": 1639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76725256301644e-05,
+      "loss": 0.7426,
+      "step": 1640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766942162889822e-05,
+      "loss": 0.7852,
+      "step": 1641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7666315832189364e-05,
+      "loss": 0.8008,
+      "step": 1642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663208240764906e-05,
+      "loss": 0.6877,
+      "step": 1643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7660098855352344e-05,
+      "loss": 0.7704,
+      "step": 1644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76569876766796e-05,
+      "loss": 0.742,
+      "step": 1645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653874705475015e-05,
+      "loss": 0.7965,
+      "step": 1646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650759942467345e-05,
+      "loss": 0.738,
+      "step": 1647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647643388385766e-05,
+      "loss": 0.8274,
+      "step": 1648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644525043959884e-05,
+      "loss": 0.8083,
+      "step": 1649
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764140490991971e-05,
+      "loss": 0.8183,
+      "step": 1650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638282986995686e-05,
+      "loss": 0.831,
+      "step": 1651
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763515927591866e-05,
+      "loss": 0.7834,
+      "step": 1652
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632033777419907e-05,
+      "loss": 0.7827,
+      "step": 1653
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762890649223112e-05,
+      "loss": 0.6993,
+      "step": 1654
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762577742108441e-05,
+      "loss": 0.763,
+      "step": 1655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622646564712304e-05,
+      "loss": 0.8645,
+      "step": 1656
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619513923847748e-05,
+      "loss": 0.7675,
+      "step": 1657
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761637949922411e-05,
+      "loss": 0.8495,
+      "step": 1658
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613243291575165e-05,
+      "loss": 0.7612,
+      "step": 1659
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610105301635117e-05,
+      "loss": 0.8581,
+      "step": 1660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760696553013858e-05,
+      "loss": 0.7078,
+      "step": 1661
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603823977820586e-05,
+      "loss": 0.7697,
+      "step": 1662
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7600680645416583e-05,
+      "loss": 0.8479,
+      "step": 1663
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597535533662447e-05,
+      "loss": 0.7732,
+      "step": 1664
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594388643294455e-05,
+      "loss": 0.8002,
+      "step": 1665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591239975049304e-05,
+      "loss": 0.8283,
+      "step": 1666
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588089529664113e-05,
+      "loss": 0.7879,
+      "step": 1667
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584937307876418e-05,
+      "loss": 0.7568,
+      "step": 1668
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581783310424163e-05,
+      "loss": 0.7953,
+      "step": 1669
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757862753804571e-05,
+      "loss": 0.7833,
+      "step": 1670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575469991479844e-05,
+      "loss": 0.3666,
+      "step": 1671
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757231067146575e-05,
+      "loss": 0.7333,
+      "step": 1672
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756914957874305e-05,
+      "loss": 0.8145,
+      "step": 1673
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756598671405176e-05,
+      "loss": 0.7029,
+      "step": 1674
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756282207813232e-05,
+      "loss": 0.8145,
+      "step": 1675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755965567172558e-05,
+      "loss": 0.872,
+      "step": 1676
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755648749557282e-05,
+      "loss": 0.775,
+      "step": 1677
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553317550415714e-05,
+      "loss": 0.8306,
+      "step": 1678
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550145836996365e-05,
+      "loss": 0.7559,
+      "step": 1679
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7546972356057275e-05,
+      "loss": 0.8101,
+      "step": 1680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754379710834138e-05,
+      "loss": 0.8472,
+      "step": 1681
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754062009459201e-05,
+      "loss": 0.7483,
+      "step": 1682
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753744131555292e-05,
+      "loss": 0.8357,
+      "step": 1683
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534260771968274e-05,
+      "loss": 0.7789,
+      "step": 1684
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531078464582652e-05,
+      "loss": 0.7624,
+      "step": 1685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7527894394141048e-05,
+      "loss": 0.8282,
+      "step": 1686
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524708561388853e-05,
+      "loss": 0.7839,
+      "step": 1687
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521520967071898e-05,
+      "loss": 0.7821,
+      "step": 1688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518331611936404e-05,
+      "loss": 0.8304,
+      "step": 1689
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515140496729015e-05,
+      "loss": 0.7378,
+      "step": 1690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7511947622196782e-05,
+      "loss": 0.8381,
+      "step": 1691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750875298908717e-05,
+      "loss": 0.8117,
+      "step": 1692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505556598148057e-05,
+      "loss": 0.711,
+      "step": 1693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7502358450127727e-05,
+      "loss": 0.8617,
+      "step": 1694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499158545774887e-05,
+      "loss": 0.8597,
+      "step": 1695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749595688583864e-05,
+      "loss": 0.8359,
+      "step": 1696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492753471068515e-05,
+      "loss": 0.8098,
+      "step": 1697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7489548302214438e-05,
+      "loss": 0.7494,
+      "step": 1698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486341380026754e-05,
+      "loss": 0.8134,
+      "step": 1699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483132705256216e-05,
+      "loss": 0.8475,
+      "step": 1700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479922278653987e-05,
+      "loss": 0.3875,
+      "step": 1701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7476710100971646e-05,
+      "loss": 0.859,
+      "step": 1702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747349617296117e-05,
+      "loss": 0.7983,
+      "step": 1703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470280495374957e-05,
+      "loss": 0.735,
+      "step": 1704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746706306896581e-05,
+      "loss": 0.8301,
+      "step": 1705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746384389448694e-05,
+      "loss": 0.7917,
+      "step": 1706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7460622972691966e-05,
+      "loss": 0.8899,
+      "step": 1707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457400304334923e-05,
+      "loss": 0.8002,
+      "step": 1708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454175890170253e-05,
+      "loss": 0.8206,
+      "step": 1709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74509497309528e-05,
+      "loss": 0.7756,
+      "step": 1710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744772182743782e-05,
+      "loss": 0.7591,
+      "step": 1711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7444492180380982e-05,
+      "loss": 0.8221,
+      "step": 1712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441260790538356e-05,
+      "loss": 0.7963,
+      "step": 1713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743802765866643e-05,
+      "loss": 0.7687,
+      "step": 1714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434792785522085e-05,
+      "loss": 0.8062,
+      "step": 1715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743155617186262e-05,
+      "loss": 0.8242,
+      "step": 1716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428317818445745e-05,
+      "loss": 0.787,
+      "step": 1717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7425077726029565e-05,
+      "loss": 0.8119,
+      "step": 1718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74218358953726e-05,
+      "loss": 0.8487,
+      "step": 1719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741859232723378e-05,
+      "loss": 0.7564,
+      "step": 1720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741534702237243e-05,
+      "loss": 0.8122,
+      "step": 1721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74120999815483e-05,
+      "loss": 0.8392,
+      "step": 1722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408851205521524e-05,
+      "loss": 0.8419,
+      "step": 1723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740560069505266e-05,
+      "loss": 0.7409,
+      "step": 1724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402348450902664e-05,
+      "loss": 0.7624,
+      "step": 1725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73990944738329e-05,
+      "loss": 0.7724,
+      "step": 1726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395838764605136e-05,
+      "loss": 0.8372,
+      "step": 1727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739258132398155e-05,
+      "loss": 0.8047,
+      "step": 1728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7389322152724716e-05,
+      "loss": 0.7467,
+      "step": 1729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386061251597625e-05,
+      "loss": 0.8605,
+      "step": 1730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738279862136367e-05,
+      "loss": 0.8432,
+      "step": 1731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7379534262786634e-05,
+      "loss": 0.8297,
+      "step": 1732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7376268176630728e-05,
+      "loss": 0.7482,
+      "step": 1733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7373000363660546e-05,
+      "loss": 0.8602,
+      "step": 1734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369730824641106e-05,
+      "loss": 0.8287,
+      "step": 1735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366459560337816e-05,
+      "loss": 0.8199,
+      "step": 1736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7363186571516492e-05,
+      "loss": 0.8087,
+      "step": 1737
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359911858943355e-05,
+      "loss": 0.7782,
+      "step": 1738
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7356635423385026e-05,
+      "loss": 0.8365,
+      "step": 1739
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7353357265608534e-05,
+      "loss": 0.8036,
+      "step": 1740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735007738638131e-05,
+      "loss": 0.8716,
+      "step": 1741
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346795786471187e-05,
+      "loss": 0.7626,
+      "step": 1742
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343512466646396e-05,
+      "loss": 0.8248,
+      "step": 1743
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340227427675583e-05,
+      "loss": 0.873,
+      "step": 1744
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336940670327783e-05,
+      "loss": 0.7834,
+      "step": 1745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733365219537244e-05,
+      "loss": 0.8816,
+      "step": 1746
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73303620035794e-05,
+      "loss": 0.7771,
+      "step": 1747
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7327070095718914e-05,
+      "loss": 0.8663,
+      "step": 1748
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7323776472561625e-05,
+      "loss": 0.7303,
+      "step": 1749
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320481134878588e-05,
+      "loss": 0.8553,
+      "step": 1750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317184083441253e-05,
+      "loss": 0.7937,
+      "step": 1751
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313885319021474e-05,
+      "loss": 0.7709,
+      "step": 1752
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73105848423915e-05,
+      "loss": 0.8595,
+      "step": 1753
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7307282654323994e-05,
+      "loss": 0.7786,
+      "step": 1754
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730397875559201e-05,
+      "loss": 0.7819,
+      "step": 1755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7300673146969e-05,
+      "loss": 0.7924,
+      "step": 1756
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729736582922882e-05,
+      "loss": 0.6961,
+      "step": 1757
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294056803145728e-05,
+      "loss": 0.8835,
+      "step": 1758
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7290746069494387e-05,
+      "loss": 0.7294,
+      "step": 1759
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287433629049842e-05,
+      "loss": 0.8701,
+      "step": 1760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7284119482587557e-05,
+      "loss": 0.7366,
+      "step": 1761
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280803630883384e-05,
+      "loss": 0.7956,
+      "step": 1762
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277486074713578e-05,
+      "loss": 0.3829,
+      "step": 1763
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727416681485479e-05,
+      "loss": 0.7749,
+      "step": 1764
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270845852084076e-05,
+      "loss": 0.8002,
+      "step": 1765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267523187178882e-05,
+      "loss": 0.8296,
+      "step": 1766
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264198820917062e-05,
+      "loss": 0.813,
+      "step": 1767
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260872754076864e-05,
+      "loss": 0.8032,
+      "step": 1768
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725754498743693e-05,
+      "loss": 0.8695,
+      "step": 1769
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7254215521776306e-05,
+      "loss": 0.3898,
+      "step": 1770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725088435787443e-05,
+      "loss": 0.745,
+      "step": 1771
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247551496511147e-05,
+      "loss": 0.7368,
+      "step": 1772
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724421693846669e-05,
+      "loss": 0.8546,
+      "step": 1773
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.724088068452169e-05,
+      "loss": 0.7742,
+      "step": 1774
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7237542735457184e-05,
+      "loss": 0.8443,
+      "step": 1775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234203092054592e-05,
+      "loss": 0.7772,
+      "step": 1776
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230861755095742e-05,
+      "loss": 0.8096,
+      "step": 1777
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227518725362854e-05,
+      "loss": 0.7727,
+      "step": 1778
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7224174003638548e-05,
+      "loss": 0.6945,
+      "step": 1779
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7220827590705832e-05,
+      "loss": 0.834,
+      "step": 1780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7217479487348116e-05,
+      "loss": 0.8061,
+      "step": 1781
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7214129694349205e-05,
+      "loss": 0.7824,
+      "step": 1782
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72107782124933e-05,
+      "loss": 0.7441,
+      "step": 1783
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207425042564994e-05,
+      "loss": 0.8052,
+      "step": 1784
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7204070185349282e-05,
+      "loss": 0.7962,
+      "step": 1785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200713641631545e-05,
+      "loss": 0.708,
+      "step": 1786
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197355412197568e-05,
+      "loss": 0.3927,
+      "step": 1787
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193995497833523e-05,
+      "loss": 0.74,
+      "step": 1788
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190633899325978e-05,
+      "loss": 0.7972,
+      "step": 1789
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71872706174619e-05,
+      "loss": 0.8762,
+      "step": 1790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718390565302865e-05,
+      "loss": 0.7563,
+      "step": 1791
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180539006813973e-05,
+      "loss": 0.868,
+      "step": 1792
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177170679606016e-05,
+      "loss": 0.7485,
+      "step": 1793
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717380067219332e-05,
+      "loss": 0.8558,
+      "step": 1794
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717042898536482e-05,
+      "loss": 0.7396,
+      "step": 1795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167055619909835e-05,
+      "loss": 0.8462,
+      "step": 1796
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7163680576618087e-05,
+      "loss": 0.7554,
+      "step": 1797
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160303856279688e-05,
+      "loss": 0.774,
+      "step": 1798
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156925459685143e-05,
+      "loss": 0.8211,
+      "step": 1799
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153545387625344e-05,
+      "loss": 0.8074,
+      "step": 1800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150163640891582e-05,
+      "loss": 0.7484,
+      "step": 1801
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7146780220275538e-05,
+      "loss": 0.7485,
+      "step": 1802
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7143395126569283e-05,
+      "loss": 0.7879,
+      "step": 1803
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140008360565287e-05,
+      "loss": 0.7628,
+      "step": 1804
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136619923056397e-05,
+      "loss": 0.8157,
+      "step": 1805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133229814835863e-05,
+      "loss": 0.728,
+      "step": 1806
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129838036697327e-05,
+      "loss": 0.8425,
+      "step": 1807
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712644458943481e-05,
+      "loss": 0.8458,
+      "step": 1808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7123049473842742e-05,
+      "loss": 0.8203,
+      "step": 1809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7119652690715927e-05,
+      "loss": 0.8081,
+      "step": 1810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116254240849568e-05,
+      "loss": 0.8219,
+      "step": 1811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112854125039252e-05,
+      "loss": 0.764,
+      "step": 1812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710945234408096e-05,
+      "loss": 0.3761,
+      "step": 1813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710604889877107e-05,
+      "loss": 0.7646,
+      "step": 1814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102643789906337e-05,
+      "loss": 0.8451,
+      "step": 1815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709923701828391e-05,
+      "loss": 0.8699,
+      "step": 1816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7095828584701327e-05,
+      "loss": 0.9261,
+      "step": 1817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709241848995652e-05,
+      "loss": 0.7922,
+      "step": 1818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708900673484781e-05,
+      "loss": 0.3904,
+      "step": 1819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708559332017389e-05,
+      "loss": 0.7571,
+      "step": 1820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082178246733864e-05,
+      "loss": 0.7501,
+      "step": 1821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7078761515327215e-05,
+      "loss": 0.7966,
+      "step": 1822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075343126753805e-05,
+      "loss": 0.8299,
+      "step": 1823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70719230818139e-05,
+      "loss": 0.8465,
+      "step": 1824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068501381308148e-05,
+      "loss": 0.7678,
+      "step": 1825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065078026037576e-05,
+      "loss": 0.7855,
+      "step": 1826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061653016803607e-05,
+      "loss": 0.8555,
+      "step": 1827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7058226354408055e-05,
+      "loss": 0.8195,
+      "step": 1828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7054798039653115e-05,
+      "loss": 0.7768,
+      "step": 1829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051368073341357e-05,
+      "loss": 0.7716,
+      "step": 1830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7047936456275766e-05,
+      "loss": 0.7735,
+      "step": 1831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704450318925969e-05,
+      "loss": 0.7868,
+      "step": 1832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704106827309687e-05,
+      "loss": 0.729,
+      "step": 1833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703763170859144e-05,
+      "loss": 0.8168,
+      "step": 1834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034193496547903e-05,
+      "loss": 0.8258,
+      "step": 1835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703075363777117e-05,
+      "loss": 0.793,
+      "step": 1836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027312133066516e-05,
+      "loss": 0.7421,
+      "step": 1837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7023868983239613e-05,
+      "loss": 0.7843,
+      "step": 1838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7020424189096524e-05,
+      "loss": 0.8005,
+      "step": 1839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701697775144368e-05,
+      "loss": 0.3567,
+      "step": 1840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7013529671087912e-05,
+      "loss": 0.7835,
+      "step": 1841
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010079948836432e-05,
+      "loss": 0.7931,
+      "step": 1842
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006628585496824e-05,
+      "loss": 0.8063,
+      "step": 1843
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7003175581877074e-05,
+      "loss": 0.6482,
+      "step": 1844
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699972093878554e-05,
+      "loss": 0.7958,
+      "step": 1845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996264657030972e-05,
+      "loss": 0.8286,
+      "step": 1846
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992806737422497e-05,
+      "loss": 0.8391,
+      "step": 1847
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989347180769627e-05,
+      "loss": 0.8299,
+      "step": 1848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698588598788226e-05,
+      "loss": 0.8226,
+      "step": 1849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698242315957068e-05,
+      "loss": 0.822,
+      "step": 1850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978958696645537e-05,
+      "loss": 0.8083,
+      "step": 1851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975492599917883e-05,
+      "loss": 0.7788,
+      "step": 1852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6972024870199148e-05,
+      "loss": 0.7351,
+      "step": 1853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696855550830114e-05,
+      "loss": 0.7617,
+      "step": 1854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965084515036047e-05,
+      "loss": 0.7992,
+      "step": 1855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6961611891216443e-05,
+      "loss": 0.8707,
+      "step": 1856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695813763765529e-05,
+      "loss": 0.3661,
+      "step": 1857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954661755165918e-05,
+      "loss": 0.7221,
+      "step": 1858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695118424456205e-05,
+      "loss": 0.8196,
+      "step": 1859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6947705106657776e-05,
+      "loss": 0.8372,
+      "step": 1860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944224342267584e-05,
+      "loss": 0.8267,
+      "step": 1861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940741952206342e-05,
+      "loss": 0.8041,
+      "step": 1862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693725793728928e-05,
+      "loss": 0.8013,
+      "step": 1863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6933772298332018e-05,
+      "loss": 0.7578,
+      "step": 1864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693028503615057e-05,
+      "loss": 0.8429,
+      "step": 1865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692679615156131e-05,
+      "loss": 0.7876,
+      "step": 1866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923305645381e-05,
+      "loss": 0.7684,
+      "step": 1867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919813518426785e-05,
+      "loss": 0.7878,
+      "step": 1868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916319771516187e-05,
+      "loss": 0.7402,
+      "step": 1869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69128244054671e-05,
+      "loss": 0.8386,
+      "step": 1870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690932742109781e-05,
+      "loss": 0.8052,
+      "step": 1871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905828819226973e-05,
+      "loss": 0.7097,
+      "step": 1872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902328600673627e-05,
+      "loss": 0.7901,
+      "step": 1873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689882676625718e-05,
+      "loss": 0.7744,
+      "step": 1874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6895323316797438e-05,
+      "loss": 0.7852,
+      "step": 1875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6891818253114562e-05,
+      "loss": 0.7709,
+      "step": 1876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688831157602911e-05,
+      "loss": 0.3776,
+      "step": 1877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6884803286362e-05,
+      "loss": 0.8196,
+      "step": 1878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881293384934544e-05,
+      "loss": 0.8636,
+      "step": 1879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877781872568423e-05,
+      "loss": 0.8177,
+      "step": 1880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874268750085694e-05,
+      "loss": 0.7996,
+      "step": 1881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870754018308795e-05,
+      "loss": 0.8052,
+      "step": 1882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867237678060538e-05,
+      "loss": 0.8439,
+      "step": 1883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6863719730164113e-05,
+      "loss": 0.3841,
+      "step": 1884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686020017544309e-05,
+      "loss": 0.7998,
+      "step": 1885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6856679014721402e-05,
+      "loss": 0.7195,
+      "step": 1886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853156248823375e-05,
+      "loss": 0.8408,
+      "step": 1887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849631878573696e-05,
+      "loss": 0.7875,
+      "step": 1888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684610590479744e-05,
+      "loss": 0.7448,
+      "step": 1889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684257832832005e-05,
+      "loss": 0.7938,
+      "step": 1890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683904914996735e-05,
+      "loss": 0.7808,
+      "step": 1891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683551837056552e-05,
+      "loss": 0.7571,
+      "step": 1892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6831985990941148e-05,
+      "loss": 0.8165,
+      "step": 1893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828452011921174e-05,
+      "loss": 0.7278,
+      "step": 1894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682491643433291e-05,
+      "loss": 0.8457,
+      "step": 1895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6821379259004048e-05,
+      "loss": 0.7864,
+      "step": 1896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817840486762665e-05,
+      "loss": 0.8518,
+      "step": 1897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814300118437194e-05,
+      "loss": 0.7841,
+      "step": 1898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810758154856454e-05,
+      "loss": 0.361,
+      "step": 1899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807214596849632e-05,
+      "loss": 0.7468,
+      "step": 1900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803669445246286e-05,
+      "loss": 0.754,
+      "step": 1901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800122700876357e-05,
+      "loss": 0.7483,
+      "step": 1902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796574364570143e-05,
+      "loss": 0.8108,
+      "step": 1903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793024437158335e-05,
+      "loss": 0.8702,
+      "step": 1904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789472919471973e-05,
+      "loss": 0.8178,
+      "step": 1905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678591981234249e-05,
+      "loss": 0.8302,
+      "step": 1906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782365116601683e-05,
+      "loss": 0.8104,
+      "step": 1907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677880883308172e-05,
+      "loss": 0.8498,
+      "step": 1908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6775250962615138e-05,
+      "loss": 0.7567,
+      "step": 1909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677169150603485e-05,
+      "loss": 0.7151,
+      "step": 1910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6768130464174138e-05,
+      "loss": 0.7598,
+      "step": 1911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764567837866663e-05,
+      "loss": 0.7242,
+      "step": 1912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676100362794644e-05,
+      "loss": 0.8256,
+      "step": 1913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6757437835247877e-05,
+      "loss": 0.8205,
+      "step": 1914
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675387046060573e-05,
+      "loss": 0.7926,
+      "step": 1915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750301504855144e-05,
+      "loss": 0.8098,
+      "step": 1916
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674673096883162e-05,
+      "loss": 0.8054,
+      "step": 1917
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6743158853371035e-05,
+      "loss": 0.7919,
+      "step": 1918
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673958515930964e-05,
+      "loss": 0.736,
+      "step": 1919
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736009887484047e-05,
+      "loss": 0.8801,
+      "step": 1920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732433038731245e-05,
+      "loss": 0.7709,
+      "step": 1921
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728854613888588e-05,
+      "loss": 0.7868,
+      "step": 1922
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67252746137938e-05,
+      "loss": 0.7365,
+      "step": 1923
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672169303928497e-05,
+      "loss": 0.8096,
+      "step": 1924
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718109891200562e-05,
+      "loss": 0.69,
+      "step": 1925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6714525170379402e-05,
+      "loss": 0.7161,
+      "step": 1926
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6710938877660697e-05,
+      "loss": 0.808,
+      "step": 1927
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707351013884e-05,
+      "loss": 0.7712,
+      "step": 1928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703761579889256e-05,
+      "loss": 0.7961,
+      "step": 1929
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670017057651676e-05,
+      "loss": 0.7403,
+      "step": 1930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6696578004607183e-05,
+      "loss": 0.8106,
+      "step": 1931
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669298386500156e-05,
+      "loss": 0.7347,
+      "step": 1932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6689388158541288e-05,
+      "loss": 0.8191,
+      "step": 1933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685790886068144e-05,
+      "loss": 0.8328,
+      "step": 1934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668219204842426e-05,
+      "loss": 0.8305,
+      "step": 1935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6678591646452142e-05,
+      "loss": 0.7262,
+      "step": 1936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674989680994654e-05,
+      "loss": 0.7773,
+      "step": 1937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671386152895033e-05,
+      "loss": 0.7104,
+      "step": 1938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666778106299688e-05,
+      "loss": 0.7745,
+      "step": 1939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6664174412144166e-05,
+      "loss": 0.3726,
+      "step": 1940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660566201181212e-05,
+      "loss": 0.8108,
+      "step": 1941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665695643095272e-05,
+      "loss": 0.7824,
+      "step": 1942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653345102303757e-05,
+      "loss": 0.8247,
+      "step": 1943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649732216079743e-05,
+      "loss": 0.7594,
+      "step": 1944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6646117773126466e-05,
+      "loss": 0.8583,
+      "step": 1945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664250177429009e-05,
+      "loss": 0.7277,
+      "step": 1946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638884220417133e-05,
+      "loss": 0.8235,
+      "step": 1947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6635265112354476e-05,
+      "loss": 0.8872,
+      "step": 1948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631644450949373e-05,
+      "loss": 0.6719,
+      "step": 1949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6628022237049425e-05,
+      "loss": 0.8473,
+      "step": 1950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6624398471502615e-05,
+      "loss": 0.783,
+      "step": 1951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662077315515728e-05,
+      "loss": 0.8262,
+      "step": 1952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661714628886212e-05,
+      "loss": 0.7412,
+      "step": 1953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6613517873466202e-05,
+      "loss": 0.8241,
+      "step": 1954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660988790981895e-05,
+      "loss": 0.7135,
+      "step": 1955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606256398770152e-05,
+      "loss": 0.8061,
+      "step": 1956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6602623341169962e-05,
+      "loss": 0.8271,
+      "step": 1957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659898873786889e-05,
+      "loss": 0.8479,
+      "step": 1958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659535258971782e-05,
+      "loss": 0.8625,
+      "step": 1959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6591714897567977e-05,
+      "loss": 0.7595,
+      "step": 1960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588075662270968e-05,
+      "loss": 0.8468,
+      "step": 1961
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6584434884678747e-05,
+      "loss": 0.8074,
+      "step": 1962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580792565643642e-05,
+      "loss": 0.7625,
+      "step": 1963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657714870601833e-05,
+      "loss": 0.757,
+      "step": 1964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573503306655852e-05,
+      "loss": 0.7632,
+      "step": 1965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569856368409613e-05,
+      "loss": 0.7911,
+      "step": 1966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566207892133376e-05,
+      "loss": 0.7846,
+      "step": 1967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6562557878681265e-05,
+      "loss": 0.8264,
+      "step": 1968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6558906328907757e-05,
+      "loss": 0.8253,
+      "step": 1969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555253243667705e-05,
+      "loss": 0.836,
+      "step": 1970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6551598623816307e-05,
+      "loss": 0.8136,
+      "step": 1971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6547942470209117e-05,
+      "loss": 0.8582,
+      "step": 1972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654428478370206e-05,
+      "loss": 0.7391,
+      "step": 1973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654062556515142e-05,
+      "loss": 0.8483,
+      "step": 1974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653696481541383e-05,
+      "loss": 0.8132,
+      "step": 1975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653330253534629e-05,
+      "loss": 0.8095,
+      "step": 1976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6529638725806147e-05,
+      "loss": 0.8195,
+      "step": 1977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6525973387651124e-05,
+      "loss": 0.7912,
+      "step": 1978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652230652173928e-05,
+      "loss": 0.8152,
+      "step": 1979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6518638128929055e-05,
+      "loss": 0.7635,
+      "step": 1980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651496821007923e-05,
+      "loss": 0.8251,
+      "step": 1981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651129676604894e-05,
+      "loss": 0.8222,
+      "step": 1982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6507623797697697e-05,
+      "loss": 0.7751,
+      "step": 1983
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503949305885356e-05,
+      "loss": 0.8203,
+      "step": 1984
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500273291472123e-05,
+      "loss": 0.8133,
+      "step": 1985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649659575531857e-05,
+      "loss": 0.7962,
+      "step": 1986
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6492916698285627e-05,
+      "loss": 0.7942,
+      "step": 1987
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489236121234575e-05,
+      "loss": 0.8378,
+      "step": 1988
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648555402502705e-05,
+      "loss": 0.3824,
+      "step": 1989
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648187041052505e-05,
+      "loss": 0.8318,
+      "step": 1990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647818527859092e-05,
+      "loss": 0.7588,
+      "step": 1991
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647449863008736e-05,
+      "loss": 0.7257,
+      "step": 1992
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6470810465877437e-05,
+      "loss": 0.7811,
+      "step": 1993
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467120786824566e-05,
+      "loss": 0.7111,
+      "step": 1994
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646342959379251e-05,
+      "loss": 0.7421,
+      "step": 1995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459736887645395e-05,
+      "loss": 0.7836,
+      "step": 1996
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64560426692477e-05,
+      "loss": 0.8238,
+      "step": 1997
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6452346939464253e-05,
+      "loss": 0.8166,
+      "step": 1998
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644864969916024e-05,
+      "loss": 0.7891,
+      "step": 1999
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6444950949201207e-05,
+      "loss": 0.7654,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6441250690453033e-05,
+      "loss": 0.7772,
+      "step": 2001
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6437548923781977e-05,
+      "loss": 0.73,
+      "step": 2002
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6433845650054627e-05,
+      "loss": 0.3696,
+      "step": 2003
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6430140870137943e-05,
+      "loss": 0.8569,
+      "step": 2004
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6426434584899224e-05,
+      "loss": 0.8152,
+      "step": 2005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422726795206124e-05,
+      "loss": 0.7625,
+      "step": 2006
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641901750192666e-05,
+      "loss": 0.8391,
+      "step": 2007
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415306705929183e-05,
+      "loss": 0.8193,
+      "step": 2008
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411594408082413e-05,
+      "loss": 0.3674,
+      "step": 2009
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407880609255408e-05,
+      "loss": 0.7678,
+      "step": 2010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404165310317587e-05,
+      "loss": 0.8394,
+      "step": 2011
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6400448512138716e-05,
+      "loss": 0.822,
+      "step": 2012
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396730215588913e-05,
+      "loss": 0.7703,
+      "step": 2013
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6393010421538646e-05,
+      "loss": 0.827,
+      "step": 2014
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6389289130858732e-05,
+      "loss": 0.3558,
+      "step": 2015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385566344420345e-05,
+      "loss": 0.7886,
+      "step": 2016
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6381842063094998e-05,
+      "loss": 0.7759,
+      "step": 2017
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6378116287754563e-05,
+      "loss": 0.8169,
+      "step": 2018
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637438901927126e-05,
+      "loss": 0.8626,
+      "step": 2019
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637066025851766e-05,
+      "loss": 0.775,
+      "step": 2020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366930006366675e-05,
+      "loss": 0.7867,
+      "step": 2021
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636319826369158e-05,
+      "loss": 0.813,
+      "step": 2022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6359465031365978e-05,
+      "loss": 0.822,
+      "step": 2023
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355730310263847e-05,
+      "loss": 0.8265,
+      "step": 2024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6351994101259498e-05,
+      "loss": 0.7856,
+      "step": 2025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348256405227586e-05,
+      "loss": 0.8375,
+      "step": 2026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6344517223043127e-05,
+      "loss": 0.7303,
+      "step": 2027
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634077655558148e-05,
+      "loss": 0.8367,
+      "step": 2028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633703440371834e-05,
+      "loss": 0.8032,
+      "step": 2029
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333290768329774e-05,
+      "loss": 0.803,
+      "step": 2030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632954565029217e-05,
+      "loss": 0.7575,
+      "step": 2031
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6325799050482282e-05,
+      "loss": 0.7374,
+      "step": 2032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6322050969777204e-05,
+      "loss": 0.7394,
+      "step": 2033
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318301409054374e-05,
+      "loss": 0.756,
+      "step": 2034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314550369191583e-05,
+      "loss": 0.801,
+      "step": 2035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631079785106696e-05,
+      "loss": 0.8081,
+      "step": 2036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630704385555899e-05,
+      "loss": 0.7528,
+      "step": 2037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6303288383546494e-05,
+      "loss": 0.7754,
+      "step": 2038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6299531435908647e-05,
+      "loss": 0.7958,
+      "step": 2039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629577301352496e-05,
+      "loss": 0.8061,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.62920131172753e-05,
+      "loss": 0.8275,
+      "step": 2041
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628825174803988e-05,
+      "loss": 0.7462,
+      "step": 2042
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628448890669924e-05,
+      "loss": 0.8185,
+      "step": 2043
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280724594134277e-05,
+      "loss": 0.7937,
+      "step": 2044
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276958811226234e-05,
+      "loss": 0.7394,
+      "step": 2045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273191558856703e-05,
+      "loss": 0.705,
+      "step": 2046
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269422837907608e-05,
+      "loss": 0.8199,
+      "step": 2047
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6265652649261218e-05,
+      "loss": 0.8007,
+      "step": 2048
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261880993800155e-05,
+      "loss": 0.6746,
+      "step": 2049
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258107872407376e-05,
+      "loss": 0.8019,
+      "step": 2050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6254333285966185e-05,
+      "loss": 0.8848,
+      "step": 2051
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250557235360233e-05,
+      "loss": 0.6854,
+      "step": 2052
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6246779721473497e-05,
+      "loss": 0.7256,
+      "step": 2053
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6243000745190317e-05,
+      "loss": 0.8057,
+      "step": 2054
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623922030739536e-05,
+      "loss": 0.7946,
+      "step": 2055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6235438408973658e-05,
+      "loss": 0.7932,
+      "step": 2056
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6231655050810548e-05,
+      "loss": 0.7076,
+      "step": 2057
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622787023379174e-05,
+      "loss": 0.3917,
+      "step": 2058
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224083958803274e-05,
+      "loss": 0.7754,
+      "step": 2059
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622029622673153e-05,
+      "loss": 0.8212,
+      "step": 2060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621650703846324e-05,
+      "loss": 0.8604,
+      "step": 2061
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6212716394885454e-05,
+      "loss": 0.782,
+      "step": 2062
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208924296885586e-05,
+      "loss": 0.7641,
+      "step": 2063
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6205130745351382e-05,
+      "loss": 0.7755,
+      "step": 2064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6201335741170928e-05,
+      "loss": 0.8633,
+      "step": 2065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6197539285232645e-05,
+      "loss": 0.3323,
+      "step": 2066
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193741378425304e-05,
+      "loss": 0.8042,
+      "step": 2067
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189942021638006e-05,
+      "loss": 0.804,
+      "step": 2068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6186141215760196e-05,
+      "loss": 0.8199,
+      "step": 2069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182338961681662e-05,
+      "loss": 0.7119,
+      "step": 2070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6178535260292523e-05,
+      "loss": 0.7985,
+      "step": 2071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174730112483246e-05,
+      "loss": 0.7963,
+      "step": 2072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6170923519144624e-05,
+      "loss": 0.7566,
+      "step": 2073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167115481167807e-05,
+      "loss": 0.4072,
+      "step": 2074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616330599944426e-05,
+      "loss": 0.8687,
+      "step": 2075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159495074865806e-05,
+      "loss": 0.8242,
+      "step": 2076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155682708324593e-05,
+      "loss": 0.7711,
+      "step": 2077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151868900713114e-05,
+      "loss": 0.7496,
+      "step": 2078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61480536529242e-05,
+      "loss": 0.6862,
+      "step": 2079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6144236965851008e-05,
+      "loss": 0.7776,
+      "step": 2080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614041884038705e-05,
+      "loss": 0.7466,
+      "step": 2081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613659927742615e-05,
+      "loss": 0.7892,
+      "step": 2082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6132778277862503e-05,
+      "loss": 0.8421,
+      "step": 2083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128955842590603e-05,
+      "loss": 0.8636,
+      "step": 2084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6125131972505307e-05,
+      "loss": 0.8931,
+      "step": 2085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6121306668501794e-05,
+      "loss": 0.7617,
+      "step": 2086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6117479931475588e-05,
+      "loss": 0.7059,
+      "step": 2087
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611365176232254e-05,
+      "loss": 0.7839,
+      "step": 2088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610982216193884e-05,
+      "loss": 0.7776,
+      "step": 2089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610599113122101e-05,
+      "loss": 0.8181,
+      "step": 2090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6102158671065916e-05,
+      "loss": 0.738,
+      "step": 2091
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098324782370754e-05,
+      "loss": 0.7808,
+      "step": 2092
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609448946603304e-05,
+      "loss": 0.7696,
+      "step": 2093
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090652722950654e-05,
+      "loss": 0.7896,
+      "step": 2094
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086814554021782e-05,
+      "loss": 0.8145,
+      "step": 2095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608297496014496e-05,
+      "loss": 0.7573,
+      "step": 2096
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079133942219044e-05,
+      "loss": 0.8457,
+      "step": 2097
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6075291501143242e-05,
+      "loss": 0.8238,
+      "step": 2098
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071447637817082e-05,
+      "loss": 0.8324,
+      "step": 2099
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067602353140427e-05,
+      "loss": 0.7722,
+      "step": 2100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6063755648013473e-05,
+      "loss": 0.8263,
+      "step": 2101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605990752333675e-05,
+      "loss": 0.7387,
+      "step": 2102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605605798001112e-05,
+      "loss": 0.8662,
+      "step": 2103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6052207018937774e-05,
+      "loss": 0.7822,
+      "step": 2104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6048354641018242e-05,
+      "loss": 0.8197,
+      "step": 2105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6044500847154383e-05,
+      "loss": 0.3765,
+      "step": 2106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604064563824838e-05,
+      "loss": 0.7422,
+      "step": 2107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603678901520275e-05,
+      "loss": 0.402,
+      "step": 2108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6032930978920354e-05,
+      "loss": 0.7663,
+      "step": 2109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602907153030437e-05,
+      "loss": 0.3726,
+      "step": 2110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025210670258307e-05,
+      "loss": 0.8103,
+      "step": 2111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602134839968601e-05,
+      "loss": 0.7567,
+      "step": 2112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017484719491652e-05,
+      "loss": 0.8194,
+      "step": 2113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013619630579743e-05,
+      "loss": 0.7365,
+      "step": 2114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009753133855107e-05,
+      "loss": 0.7659,
+      "step": 2115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005885230222906e-05,
+      "loss": 0.7995,
+      "step": 2116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600201592058864e-05,
+      "loss": 0.7757,
+      "step": 2117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998145205858122e-05,
+      "loss": 0.7932,
+      "step": 2118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599427308693751e-05,
+      "loss": 0.7815,
+      "step": 2119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990399564733274e-05,
+      "loss": 0.7896,
+      "step": 2120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986524640152227e-05,
+      "loss": 0.7712,
+      "step": 2121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59826483141015e-05,
+      "loss": 0.7803,
+      "step": 2122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978770587488566e-05,
+      "loss": 0.8245,
+      "step": 2123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5974891461221207e-05,
+      "loss": 0.7724,
+      "step": 2124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5971010936207543e-05,
+      "loss": 0.8669,
+      "step": 2125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5967129013356027e-05,
+      "loss": 0.8033,
+      "step": 2126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5963245693575425e-05,
+      "loss": 0.7985,
+      "step": 2127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5959360977774847e-05,
+      "loss": 0.8185,
+      "step": 2128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5955474866863712e-05,
+      "loss": 0.7265,
+      "step": 2129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5951587361751776e-05,
+      "loss": 0.7587,
+      "step": 2130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947698463349124e-05,
+      "loss": 0.7594,
+      "step": 2131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594380817256616e-05,
+      "loss": 0.8656,
+      "step": 2132
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5939916490313617e-05,
+      "loss": 0.7629,
+      "step": 2133
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5936023417502554e-05,
+      "loss": 0.7411,
+      "step": 2134
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5932128955044355e-05,
+      "loss": 0.7959,
+      "step": 2135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592823310385073e-05,
+      "loss": 0.7574,
+      "step": 2136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924335864833712e-05,
+      "loss": 0.7446,
+      "step": 2137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5920437238905662e-05,
+      "loss": 0.7786,
+      "step": 2138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591653722697926e-05,
+      "loss": 0.8763,
+      "step": 2139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912635829967524e-05,
+      "loss": 0.8521,
+      "step": 2140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5908733048783774e-05,
+      "loss": 0.8244,
+      "step": 2141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590482888434168e-05,
+      "loss": 0.3569,
+      "step": 2142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5900923337555215e-05,
+      "loss": 0.388,
+      "step": 2143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589701640933868e-05,
+      "loss": 0.7628,
+      "step": 2144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5893108100606715e-05,
+      "loss": 0.8059,
+      "step": 2145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588919841227426e-05,
+      "loss": 0.7918,
+      "step": 2146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885287345256592e-05,
+      "loss": 0.7707,
+      "step": 2147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881374900469307e-05,
+      "loss": 0.3296,
+      "step": 2148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587746107882833e-05,
+      "loss": 0.8211,
+      "step": 2149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5873545881249893e-05,
+      "loss": 0.8739,
+      "step": 2150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5869629308650563e-05,
+      "loss": 0.8166,
+      "step": 2151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5865711361947234e-05,
+      "loss": 0.7916,
+      "step": 2152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58617920420571e-05,
+      "loss": 0.7804,
+      "step": 2153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5857871349897696e-05,
+      "loss": 0.82,
+      "step": 2154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5853949286386872e-05,
+      "loss": 0.4031,
+      "step": 2155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850025852442802e-05,
+      "loss": 0.8307,
+      "step": 2156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5846101048983973e-05,
+      "loss": 0.7809,
+      "step": 2157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5842174876929196e-05,
+      "loss": 0.791,
+      "step": 2158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583824733719761e-05,
+      "loss": 0.7394,
+      "step": 2159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834318430708662e-05,
+      "loss": 0.791,
+      "step": 2160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830388158382128e-05,
+      "loss": 0.7654,
+      "step": 2161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58264565211381e-05,
+      "loss": 0.7936,
+      "step": 2162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5822523519896988e-05,
+      "loss": 0.7658,
+      "step": 2163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581858915557953e-05,
+      "loss": 0.8218,
+      "step": 2164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581465342910677e-05,
+      "loss": 0.3672,
+      "step": 2165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5810716341400083e-05,
+      "loss": 0.726,
+      "step": 2166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580677789338115e-05,
+      "loss": 0.832,
+      "step": 2167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802838085971985e-05,
+      "loss": 0.7946,
+      "step": 2168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798896920094907e-05,
+      "loss": 0.7512,
+      "step": 2169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5794954396672568e-05,
+      "loss": 0.7397,
+      "step": 2170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791010516627915e-05,
+      "loss": 0.3909,
+      "step": 2171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5787065280884234e-05,
+      "loss": 0.7189,
+      "step": 2172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5783118690365124e-05,
+      "loss": 0.8049,
+      "step": 2173
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577917074599449e-05,
+      "loss": 0.7921,
+      "step": 2174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577522144869657e-05,
+      "loss": 0.8517,
+      "step": 2175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5771270799395906e-05,
+      "loss": 0.7325,
+      "step": 2176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576731879901736e-05,
+      "loss": 0.8063,
+      "step": 2177
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5763365448486114e-05,
+      "loss": 0.3787,
+      "step": 2178
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759410748727663e-05,
+      "loss": 0.7405,
+      "step": 2179
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575545470066781e-05,
+      "loss": 0.7507,
+      "step": 2180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57514973052327e-05,
+      "loss": 0.8455,
+      "step": 2181
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747538563348757e-05,
+      "loss": 0.8153,
+      "step": 2182
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574357847594275e-05,
+      "loss": 0.7159,
+      "step": 2183
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5739617043941742e-05,
+      "loss": 0.8119,
+      "step": 2184
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735654268273132e-05,
+      "loss": 0.7059,
+      "step": 2185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731690149864614e-05,
+      "loss": 0.7872,
+      "step": 2186
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727724689644204e-05,
+      "loss": 0.7575,
+      "step": 2187
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5723757888540237e-05,
+      "loss": 0.6876,
+      "step": 2188
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571978974748135e-05,
+      "loss": 0.7166,
+      "step": 2189
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5715820267396508e-05,
+      "loss": 0.8613,
+      "step": 2190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571184944921498e-05,
+      "loss": 0.8503,
+      "step": 2191
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570787729386635e-05,
+      "loss": 0.7796,
+      "step": 2192
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5703903802280516e-05,
+      "loss": 0.8502,
+      "step": 2193
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5699928975387687e-05,
+      "loss": 0.7815,
+      "step": 2194
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695952814118388e-05,
+      "loss": 0.8229,
+      "step": 2195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691975319403456e-05,
+      "loss": 0.3722,
+      "step": 2196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568799649217404e-05,
+      "loss": 0.8735,
+      "step": 2197
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5684016333361588e-05,
+      "loss": 0.3734,
+      "step": 2198
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5680034843897884e-05,
+      "loss": 0.3735,
+      "step": 2199
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5676052024715004e-05,
+      "loss": 0.7172,
+      "step": 2200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672067876745347e-05,
+      "loss": 0.6967,
+      "step": 2201
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668082400921613e-05,
+      "loss": 0.7752,
+      "step": 2202
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5664095598176823e-05,
+      "loss": 0.7564,
+      "step": 2203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5660107469444303e-05,
+      "loss": 0.3535,
+      "step": 2204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5656118015657685e-05,
+      "loss": 0.8064,
+      "step": 2205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565212723775092e-05,
+      "loss": 0.8041,
+      "step": 2206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5648135136658265e-05,
+      "loss": 0.7329,
+      "step": 2207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564414171331429e-05,
+      "loss": 0.7864,
+      "step": 2208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640146968653863e-05,
+      "loss": 0.7838,
+      "step": 2209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5636150903612178e-05,
+      "loss": 0.7716,
+      "step": 2210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563215351912473e-05,
+      "loss": 0.7797,
+      "step": 2211
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5628154816127315e-05,
+      "loss": 0.7592,
+      "step": 2212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5624154795556058e-05,
+      "loss": 0.8208,
+      "step": 2213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620153458347368e-05,
+      "loss": 0.8316,
+      "step": 2214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5616150805437983e-05,
+      "loss": 0.7694,
+      "step": 2215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5612146837764935e-05,
+      "loss": 0.8542,
+      "step": 2216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608141556265567e-05,
+      "loss": 0.7868,
+      "step": 2217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604134961877542e-05,
+      "loss": 0.7972,
+      "step": 2218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560012705553881e-05,
+      "loss": 0.8024,
+      "step": 2219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596117838187637e-05,
+      "loss": 0.7902,
+      "step": 2220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5592107310762607e-05,
+      "loss": 0.7618,
+      "step": 2221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5588095474202597e-05,
+      "loss": 0.7064,
+      "step": 2222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5584082329446786e-05,
+      "loss": 0.6701,
+      "step": 2223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580067877434677e-05,
+      "loss": 0.8683,
+      "step": 2224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5576052119106067e-05,
+      "loss": 0.8218,
+      "step": 2225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557203505540106e-05,
+      "loss": 0.7496,
+      "step": 2226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5568016687260064e-05,
+      "loss": 0.7595,
+      "step": 2227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55639970156238e-05,
+      "loss": 0.8505,
+      "step": 2228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5559976041433292e-05,
+      "loss": 0.7767,
+      "step": 2229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5555953765629862e-05,
+      "loss": 0.7317,
+      "step": 2230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555193018915514e-05,
+      "loss": 0.837,
+      "step": 2231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5547905312951065e-05,
+      "loss": 0.7838,
+      "step": 2232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554387913795987e-05,
+      "loss": 0.376,
+      "step": 2233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539851665124107e-05,
+      "loss": 0.8251,
+      "step": 2234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553582289538662e-05,
+      "loss": 0.7723,
+      "step": 2235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5531792829690558e-05,
+      "loss": 0.8164,
+      "step": 2236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527761468979385e-05,
+      "loss": 0.6813,
+      "step": 2237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552372881419685e-05,
+      "loss": 0.3828,
+      "step": 2238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5519694866287013e-05,
+      "loss": 0.8077,
+      "step": 2239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5515659626194245e-05,
+      "loss": 0.7389,
+      "step": 2240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511623094863203e-05,
+      "loss": 0.7581,
+      "step": 2241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507585273238866e-05,
+      "loss": 0.844,
+      "step": 2242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5503546162266495e-05,
+      "loss": 0.7408,
+      "step": 2243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5499505762891666e-05,
+      "loss": 0.7491,
+      "step": 2244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495464076060254e-05,
+      "loss": 0.8234,
+      "step": 2245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549142110271843e-05,
+      "loss": 0.7923,
+      "step": 2246
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487376843812675e-05,
+      "loss": 0.7793,
+      "step": 2247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5483331300289764e-05,
+      "loss": 0.7663,
+      "step": 2248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547928447309678e-05,
+      "loss": 0.7751,
+      "step": 2249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475236363181093e-05,
+      "loss": 0.8428,
+      "step": 2250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547118697149039e-05,
+      "loss": 0.7681,
+      "step": 2251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467136298972644e-05,
+      "loss": 0.855,
+      "step": 2252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463084346576138e-05,
+      "loss": 0.7359,
+      "step": 2253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545903111524945e-05,
+      "loss": 0.8306,
+      "step": 2254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5454976605941453e-05,
+      "loss": 0.8232,
+      "step": 2255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545092081960133e-05,
+      "loss": 0.796,
+      "step": 2256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446863757178555e-05,
+      "loss": 0.746,
+      "step": 2257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5442805419622905e-05,
+      "loss": 0.8048,
+      "step": 2258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5438745807884445e-05,
+      "loss": 0.7921,
+      "step": 2259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5434684922913553e-05,
+      "loss": 0.8024,
+      "step": 2260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430622765660902e-05,
+      "loss": 0.787,
+      "step": 2261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426559337077453e-05,
+      "loss": 0.7705,
+      "step": 2262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422494638114477e-05,
+      "loss": 0.8307,
+      "step": 2263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5418428669723528e-05,
+      "loss": 0.8079,
+      "step": 2264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414361432856475e-05,
+      "loss": 0.7834,
+      "step": 2265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5410292928465467e-05,
+      "loss": 0.7735,
+      "step": 2266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5406223157502966e-05,
+      "loss": 0.8187,
+      "step": 2267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540215212092171e-05,
+      "loss": 0.7624,
+      "step": 2268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5398079819674756e-05,
+      "loss": 0.8361,
+      "step": 2269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539400625471544e-05,
+      "loss": 0.7559,
+      "step": 2270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389931426997404e-05,
+      "loss": 0.8368,
+      "step": 2271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5385855337474577e-05,
+      "loss": 0.3779,
+      "step": 2272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538177798710119e-05,
+      "loss": 0.7586,
+      "step": 2273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377699376831772e-05,
+      "loss": 0.8498,
+      "step": 2274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537361950762114e-05,
+      "loss": 0.8664,
+      "step": 2275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5369538380424404e-05,
+      "loss": 0.8367,
+      "step": 2276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536545599619697e-05,
+      "loss": 0.7578,
+      "step": 2277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5361372355894552e-05,
+      "loss": 0.8186,
+      "step": 2278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357287460473135e-05,
+      "loss": 0.675,
+      "step": 2279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535320131088902e-05,
+      "loss": 0.7399,
+      "step": 2280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349113908098782e-05,
+      "loss": 0.749,
+      "step": 2281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53450252530593e-05,
+      "loss": 0.7868,
+      "step": 2282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534093534672775e-05,
+      "loss": 0.7639,
+      "step": 2283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5336844190061592e-05,
+      "loss": 0.8043,
+      "step": 2284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5332751784018584e-05,
+      "loss": 0.7941,
+      "step": 2285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328658129556772e-05,
+      "loss": 0.8089,
+      "step": 2286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53245632276345e-05,
+      "loss": 0.7788,
+      "step": 2287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320467079210404e-05,
+      "loss": 0.8045,
+      "step": 2288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5316369685243403e-05,
+      "loss": 0.8259,
+      "step": 2289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531227104669271e-05,
+      "loss": 0.7004,
+      "step": 2290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308171164517845e-05,
+      "loss": 0.7965,
+      "step": 2291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304070039678597e-05,
+      "loss": 0.8076,
+      "step": 2292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299967673135063e-05,
+      "loss": 0.8557,
+      "step": 2293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295864065847618e-05,
+      "loss": 0.7493,
+      "step": 2294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5291759218776934e-05,
+      "loss": 0.8286,
+      "step": 2295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5287653132883976e-05,
+      "loss": 0.7695,
+      "step": 2296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5283545809129988e-05,
+      "loss": 0.7956,
+      "step": 2297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527943724847652e-05,
+      "loss": 0.7518,
+      "step": 2298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52753274518854e-05,
+      "loss": 0.8025,
+      "step": 2299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5271216420318746e-05,
+      "loss": 0.3851,
+      "step": 2300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5267104154738966e-05,
+      "loss": 0.7555,
+      "step": 2301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5262990656108762e-05,
+      "loss": 0.8419,
+      "step": 2302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258875925391117e-05,
+      "loss": 0.7241,
+      "step": 2303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254759963549308e-05,
+      "loss": 0.7892,
+      "step": 2304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5250642771546898e-05,
+      "loss": 0.7177,
+      "step": 2305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5246524350347738e-05,
+      "loss": 0.7509,
+      "step": 2306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524240470091597e-05,
+      "loss": 0.7892,
+      "step": 2307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5238283824216015e-05,
+      "loss": 0.81,
+      "step": 2308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523416172121259e-05,
+      "loss": 0.8518,
+      "step": 2309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230038392870698e-05,
+      "loss": 0.8207,
+      "step": 2310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225913840155626e-05,
+      "loss": 0.7951,
+      "step": 2311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221788064032946e-05,
+      "loss": 0.7682,
+      "step": 2312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217661065468518e-05,
+      "loss": 0.8478,
+      "step": 2313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5213532845428494e-05,
+      "loss": 0.8041,
+      "step": 2314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209403404879305e-05,
+      "loss": 0.8255,
+      "step": 2315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520527274478767e-05,
+      "loss": 0.7149,
+      "step": 2316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520114086612059e-05,
+      "loss": 0.8,
+      "step": 2317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5197007769845358e-05,
+      "loss": 0.7973,
+      "step": 2318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5192873456929546e-05,
+      "loss": 0.7635,
+      "step": 2319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518873792834102e-05,
+      "loss": 0.8037,
+      "step": 2320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5184601185047916e-05,
+      "loss": 0.8525,
+      "step": 2321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180463228018662e-05,
+      "loss": 0.8592,
+      "step": 2322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517632405822198e-05,
+      "loss": 0.7619,
+      "step": 2323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5172183676626854e-05,
+      "loss": 0.7392,
+      "step": 2324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5168042084202577e-05,
+      "loss": 0.8252,
+      "step": 2325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163899281918704e-05,
+      "loss": 0.825,
+      "step": 2326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159755270745083e-05,
+      "loss": 0.3699,
+      "step": 2327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155610051651844e-05,
+      "loss": 0.772,
+      "step": 2328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5151463625609401e-05,
+      "loss": 0.7602,
+      "step": 2329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147315993588449e-05,
+      "loss": 0.7827,
+      "step": 2330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143167156559963e-05,
+      "loss": 0.8206,
+      "step": 2331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5139017115495207e-05,
+      "loss": 0.8286,
+      "step": 2332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5134865871365718e-05,
+      "loss": 0.8192,
+      "step": 2333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5130713425143322e-05,
+      "loss": 0.3831,
+      "step": 2334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126559777800123e-05,
+      "loss": 0.7888,
+      "step": 2335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5122404930308502e-05,
+      "loss": 0.7707,
+      "step": 2336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118248883641134e-05,
+      "loss": 0.8732,
+      "step": 2337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114091638770965e-05,
+      "loss": 0.7993,
+      "step": 2338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5109933196671214e-05,
+      "loss": 0.7118,
+      "step": 2339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5105773558315396e-05,
+      "loss": 0.7809,
+      "step": 2340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101612724677299e-05,
+      "loss": 0.7969,
+      "step": 2341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509745069673099e-05,
+      "loss": 0.8237,
+      "step": 2342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509328747545082e-05,
+      "loss": 0.7788,
+      "step": 2343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089123061811408e-05,
+      "loss": 0.8143,
+      "step": 2344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5084957456787668e-05,
+      "loss": 0.7761,
+      "step": 2345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080790661354781e-05,
+      "loss": 0.7543,
+      "step": 2346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076622676488212e-05,
+      "loss": 0.8243,
+      "step": 2347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5072453503163705e-05,
+      "loss": 0.8725,
+      "step": 2348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068283142357275e-05,
+      "loss": 0.8335,
+      "step": 2349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064111595045223e-05,
+      "loss": 0.3766,
+      "step": 2350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059938862204126e-05,
+      "loss": 0.7598,
+      "step": 2351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055764944810842e-05,
+      "loss": 0.7953,
+      "step": 2352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051589843842491e-05,
+      "loss": 0.798,
+      "step": 2353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047413560276492e-05,
+      "loss": 0.8948,
+      "step": 2354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043236095090523e-05,
+      "loss": 0.7811,
+      "step": 2355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503905744926255e-05,
+      "loss": 0.8127,
+      "step": 2356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034877623770806e-05,
+      "loss": 0.7861,
+      "step": 2357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503069661959381e-05,
+      "loss": 0.8176,
+      "step": 2358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502651443771035e-05,
+      "loss": 0.815,
+      "step": 2359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502233107909949e-05,
+      "loss": 0.7886,
+      "step": 2360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018146544740573e-05,
+      "loss": 0.7879,
+      "step": 2361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013960835613216e-05,
+      "loss": 0.7772,
+      "step": 2362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500977395269731e-05,
+      "loss": 0.7873,
+      "step": 2363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500558589697302e-05,
+      "loss": 0.7573,
+      "step": 2364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500139666942079e-05,
+      "loss": 0.8548,
+      "step": 2365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499720627102133e-05,
+      "loss": 0.814,
+      "step": 2366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993014702755635e-05,
+      "loss": 0.7243,
+      "step": 2367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4988821965604963e-05,
+      "loss": 0.7561,
+      "step": 2368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4984628060550852e-05,
+      "loss": 0.3494,
+      "step": 2369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980432988575118e-05,
+      "loss": 0.7533,
+      "step": 2370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976236750659837e-05,
+      "loss": 0.8402,
+      "step": 2371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4972039347787366e-05,
+      "loss": 0.7859,
+      "step": 2372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967840780940339e-05,
+      "loss": 0.8556,
+      "step": 2373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963641051101651e-05,
+      "loss": 0.7721,
+      "step": 2374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4959440159254482e-05,
+      "loss": 0.777,
+      "step": 2375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4955238106382272e-05,
+      "loss": 0.3899,
+      "step": 2376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951034893468746e-05,
+      "loss": 0.7986,
+      "step": 2377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4946830521497887e-05,
+      "loss": 0.7738,
+      "step": 2378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942624991453956e-05,
+      "loss": 0.851,
+      "step": 2379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938418304321484e-05,
+      "loss": 0.7801,
+      "step": 2380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493421046108528e-05,
+      "loss": 0.7338,
+      "step": 2381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493000146273041e-05,
+      "loss": 0.8476,
+      "step": 2382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4925791310242221e-05,
+      "loss": 0.7981,
+      "step": 2383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921580004606326e-05,
+      "loss": 0.7261,
+      "step": 2384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491736754680861e-05,
+      "loss": 0.7691,
+      "step": 2385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913153937835221e-05,
+      "loss": 0.7842,
+      "step": 2386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908939178672588e-05,
+      "loss": 0.817,
+      "step": 2387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904723270307405e-05,
+      "loss": 0.7956,
+      "step": 2388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900506213726625e-05,
+      "loss": 0.7136,
+      "step": 2389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489628800991748e-05,
+      "loss": 0.743,
+      "step": 2390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4892068659867475e-05,
+      "loss": 0.7971,
+      "step": 2391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887848164564373e-05,
+      "loss": 0.8177,
+      "step": 2392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883626524996208e-05,
+      "loss": 0.382,
+      "step": 2393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4879403742151283e-05,
+      "loss": 0.7948,
+      "step": 2394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875179817018173e-05,
+      "loss": 0.7734,
+      "step": 2395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4870954750585711e-05,
+      "loss": 0.8473,
+      "step": 2396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4866728543843004e-05,
+      "loss": 0.7845,
+      "step": 2397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486250119777942e-05,
+      "loss": 0.7228,
+      "step": 2398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4858272713384606e-05,
+      "loss": 0.7267,
+      "step": 2399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485404309164846e-05,
+      "loss": 0.7741,
+      "step": 2400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484981233356116e-05,
+      "loss": 0.7208,
+      "step": 2401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845580440113138e-05,
+      "loss": 0.7761,
+      "step": 2402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48413474122951e-05,
+      "loss": 0.7858,
+      "step": 2403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837113251098014e-05,
+      "loss": 0.7276,
+      "step": 2404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4832877957513117e-05,
+      "loss": 0.7919,
+      "step": 2405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828641532531904e-05,
+      "loss": 0.7205,
+      "step": 2406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824403977146142e-05,
+      "loss": 0.7889,
+      "step": 2407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4820165292347862e-05,
+      "loss": 0.8124,
+      "step": 2408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481592547912935e-05,
+      "loss": 0.789,
+      "step": 2409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811684538483173e-05,
+      "loss": 0.7064,
+      "step": 2410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807442471402144e-05,
+      "loss": 0.7661,
+      "step": 2411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4803199278879352e-05,
+      "loss": 0.7867,
+      "step": 2412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4798954961908148e-05,
+      "loss": 0.7926,
+      "step": 2413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4794709521482138e-05,
+      "loss": 0.7837,
+      "step": 2414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790462958595202e-05,
+      "loss": 0.7709,
+      "step": 2415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786215274241474e-05,
+      "loss": 0.8246,
+      "step": 2416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4781966469415355e-05,
+      "loss": 0.7593,
+      "step": 2417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777716545111509e-05,
+      "loss": 0.719,
+      "step": 2418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773465502324859e-05,
+      "loss": 0.7968,
+      "step": 2419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769213342050593e-05,
+      "loss": 0.7466,
+      "step": 2420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764960065284155e-05,
+      "loss": 0.8128,
+      "step": 2421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760705673021258e-05,
+      "loss": 0.7034,
+      "step": 2422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4756450166257875e-05,
+      "loss": 0.8332,
+      "step": 2423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475219354599023e-05,
+      "loss": 0.3642,
+      "step": 2424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474793581321482e-05,
+      "loss": 0.8642,
+      "step": 2425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743676968928395e-05,
+      "loss": 0.8188,
+      "step": 2426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4739417014127972e-05,
+      "loss": 0.3618,
+      "step": 2427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473515594981082e-05,
+      "loss": 0.7547,
+      "step": 2428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473089377697447e-05,
+      "loss": 0.7266,
+      "step": 2429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472663049661672e-05,
+      "loss": 0.764,
+      "step": 2430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722366109735612e-05,
+      "loss": 0.7798,
+      "step": 2431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471810061732947e-05,
+      "loss": 0.7167,
+      "step": 2432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713834020396849e-05,
+      "loss": 0.7817,
+      "step": 2433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709566319936581e-05,
+      "loss": 0.7539,
+      "step": 2434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705297516947757e-05,
+      "loss": 0.3401,
+      "step": 2435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4701027612429714e-05,
+      "loss": 0.753,
+      "step": 2436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469675660738206e-05,
+      "loss": 0.8366,
+      "step": 2437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692484502804651e-05,
+      "loss": 0.7944,
+      "step": 2438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688211299697605e-05,
+      "loss": 0.7651,
+      "step": 2439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683936999061297e-05,
+      "loss": 0.7577,
+      "step": 2440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4679661601896358e-05,
+      "loss": 0.7466,
+      "step": 2441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675385109203673e-05,
+      "loss": 0.7581,
+      "step": 2442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4671107521984387e-05,
+      "loss": 0.7652,
+      "step": 2443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666828841239904e-05,
+      "loss": 0.8,
+      "step": 2444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662549067971876e-05,
+      "loss": 0.7534,
+      "step": 2445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4658268203182222e-05,
+      "loss": 0.8533,
+      "step": 2446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653986247873102e-05,
+      "loss": 0.7434,
+      "step": 2447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4649703203046941e-05,
+      "loss": 0.7704,
+      "step": 2448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4645419069706421e-05,
+      "loss": 0.7715,
+      "step": 2449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4641133848854474e-05,
+      "loss": 0.7338,
+      "step": 2450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636847541494284e-05,
+      "loss": 0.7297,
+      "step": 2451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632560148629293e-05,
+      "loss": 0.7668,
+      "step": 2452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628271671263202e-05,
+      "loss": 0.7801,
+      "step": 2453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623982110399957e-05,
+      "loss": 0.7537,
+      "step": 2454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4619691467043764e-05,
+      "loss": 0.3689,
+      "step": 2455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615399742199077e-05,
+      "loss": 0.7935,
+      "step": 2456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611106936870608e-05,
+      "loss": 0.8587,
+      "step": 2457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460681305206332e-05,
+      "loss": 0.8217,
+      "step": 2458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602518088782426e-05,
+      "loss": 0.826,
+      "step": 2459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45982220480334e-05,
+      "loss": 0.7153,
+      "step": 2460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593924930821963e-05,
+      "loss": 0.7051,
+      "step": 2461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458962673815408e-05,
+      "loss": 0.7988,
+      "step": 2462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4585327471035977e-05,
+      "loss": 0.7483,
+      "step": 2463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581027130474135e-05,
+      "loss": 0.751,
+      "step": 2464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4576725717475283e-05,
+      "loss": 0.7804,
+      "step": 2465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572423233046386e-05,
+      "loss": 0.8042,
+      "step": 2466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568119678194687e-05,
+      "loss": 0.8451,
+      "step": 2467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4563815053927659e-05,
+      "loss": 0.365,
+      "step": 2468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559509361253035e-05,
+      "loss": 0.7533,
+      "step": 2469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4555202601178791e-05,
+      "loss": 0.786,
+      "step": 2470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4550894774713161e-05,
+      "loss": 0.7627,
+      "step": 2471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454658588286462e-05,
+      "loss": 0.7746,
+      "step": 2472
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542275926641904e-05,
+      "loss": 0.7573,
+      "step": 2473
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4537964907053985e-05,
+      "loss": 0.8204,
+      "step": 2474
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4533652825110091e-05,
+      "loss": 0.7313,
+      "step": 2475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529339681819703e-05,
+      "loss": 0.402,
+      "step": 2476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525025478192541e-05,
+      "loss": 0.8168,
+      "step": 2477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4520710215238575e-05,
+      "loss": 0.8098,
+      "step": 2478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4516393893968032e-05,
+      "loss": 0.7884,
+      "step": 2479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4512076515391375e-05,
+      "loss": 0.7873,
+      "step": 2480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507758080519325e-05,
+      "loss": 0.7627,
+      "step": 2481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503438590362838e-05,
+      "loss": 0.7533,
+      "step": 2482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4499118045933129e-05,
+      "loss": 0.8246,
+      "step": 2483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4494796448241655e-05,
+      "loss": 0.6973,
+      "step": 2484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490473798300117e-05,
+      "loss": 0.7201,
+      "step": 2485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486150097120471e-05,
+      "loss": 0.7894,
+      "step": 2486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481825345714904e-05,
+      "loss": 0.7755,
+      "step": 2487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4477499545095863e-05,
+      "loss": 0.7868,
+      "step": 2488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4473172696276035e-05,
+      "loss": 0.7149,
+      "step": 2489
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4468844800268353e-05,
+      "loss": 0.7046,
+      "step": 2490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464515858085995e-05,
+      "loss": 0.8283,
+      "step": 2491
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446018587074238e-05,
+      "loss": 0.3943,
+      "step": 2492
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455854839251182e-05,
+      "loss": 0.7373,
+      "step": 2493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4451522764626308e-05,
+      "loss": 0.8514,
+      "step": 2494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447189647881921e-05,
+      "loss": 0.8291,
+      "step": 2495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442855490032413e-05,
+      "loss": 0.7345,
+      "step": 2496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438520292092431e-05,
+      "loss": 0.3728,
+      "step": 2497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434184055076863e-05,
+      "loss": 0.7523,
+      "step": 2498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442984678000084e-05,
+      "loss": 0.8006,
+      "step": 2499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4425508467879734e-05,
+      "loss": 0.7289,
+      "step": 2500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4421169119729164e-05,
+      "loss": 0.7256,
+      "step": 2501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4416828736564988e-05,
+      "loss": 0.8163,
+      "step": 2502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412487319403306e-05,
+      "loss": 0.8404,
+      "step": 2503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4408144869260463e-05,
+      "loss": 0.8541,
+      "step": 2504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440380138715304e-05,
+      "loss": 0.7261,
+      "step": 2505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439945687409787e-05,
+      "loss": 0.7815,
+      "step": 2506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4395111331112017e-05,
+      "loss": 0.7874,
+      "step": 2507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4390764759212794e-05,
+      "loss": 0.7836,
+      "step": 2508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4386417159417745e-05,
+      "loss": 0.838,
+      "step": 2509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4382068532744667e-05,
+      "loss": 0.7872,
+      "step": 2510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377718880211586e-05,
+      "loss": 0.7582,
+      "step": 2511
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373368202836777e-05,
+      "loss": 0.7699,
+      "step": 2512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436901650163875e-05,
+      "loss": 0.8427,
+      "step": 2513
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364663777636257e-05,
+      "loss": 0.8502,
+      "step": 2514
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4360310031848284e-05,
+      "loss": 0.7331,
+      "step": 2515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4355955265294066e-05,
+      "loss": 0.7166,
+      "step": 2516
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435159947899307e-05,
+      "loss": 0.7022,
+      "step": 2517
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4347242673965e-05,
+      "loss": 0.336,
+      "step": 2518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4342884851229803e-05,
+      "loss": 0.7185,
+      "step": 2519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4338526011807666e-05,
+      "loss": 0.7724,
+      "step": 2520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433416615671901e-05,
+      "loss": 0.7375,
+      "step": 2521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4329805286984493e-05,
+      "loss": 0.7718,
+      "step": 2522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4325443403625012e-05,
+      "loss": 0.7884,
+      "step": 2523
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4321080507661702e-05,
+      "loss": 0.7581,
+      "step": 2524
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316716600115937e-05,
+      "loss": 0.7055,
+      "step": 2525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312351682009324e-05,
+      "loss": 0.7669,
+      "step": 2526
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307985754363707e-05,
+      "loss": 0.7876,
+      "step": 2527
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4303618818201166e-05,
+      "loss": 0.7895,
+      "step": 2528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299250874544022e-05,
+      "loss": 0.839,
+      "step": 2529
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294881924414825e-05,
+      "loss": 0.7902,
+      "step": 2530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429051196883637e-05,
+      "loss": 0.8074,
+      "step": 2531
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286141008831676e-05,
+      "loss": 0.8194,
+      "step": 2532
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281769045424001e-05,
+      "loss": 0.8071,
+      "step": 2533
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4277396079636842e-05,
+      "loss": 0.7567,
+      "step": 2534
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273022112493928e-05,
+      "loss": 0.7154,
+      "step": 2535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4268647145019223e-05,
+      "loss": 0.7893,
+      "step": 2536
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4264271178236924e-05,
+      "loss": 0.8236,
+      "step": 2537
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259894213171461e-05,
+      "loss": 0.8006,
+      "step": 2538
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4255516250847503e-05,
+      "loss": 0.7551,
+      "step": 2539
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251137292289948e-05,
+      "loss": 0.7632,
+      "step": 2540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4246757338523922e-05,
+      "loss": 0.7893,
+      "step": 2541
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4242376390574797e-05,
+      "loss": 0.781,
+      "step": 2542
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237994449468167e-05,
+      "loss": 0.7475,
+      "step": 2543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233611516229861e-05,
+      "loss": 0.8343,
+      "step": 2544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229227591885946e-05,
+      "loss": 0.7944,
+      "step": 2545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224842677462714e-05,
+      "loss": 0.7626,
+      "step": 2546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220456773986691e-05,
+      "loss": 0.7875,
+      "step": 2547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4216069882484633e-05,
+      "loss": 0.8092,
+      "step": 2548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4211682003983536e-05,
+      "loss": 0.7195,
+      "step": 2549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420729313951061e-05,
+      "loss": 0.8003,
+      "step": 2550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202903290093313e-05,
+      "loss": 0.8097,
+      "step": 2551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198512456759326e-05,
+      "loss": 0.6926,
+      "step": 2552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4194120640536557e-05,
+      "loss": 0.8268,
+      "step": 2553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4189727842453151e-05,
+      "loss": 0.8093,
+      "step": 2554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4185334063537482e-05,
+      "loss": 0.7616,
+      "step": 2555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4180939304818147e-05,
+      "loss": 0.7704,
+      "step": 2556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176543567323977e-05,
+      "loss": 0.8431,
+      "step": 2557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172146852084039e-05,
+      "loss": 0.7579,
+      "step": 2558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416774916012761e-05,
+      "loss": 0.7891,
+      "step": 2559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4163350492484218e-05,
+      "loss": 0.7691,
+      "step": 2560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4158950850183604e-05,
+      "loss": 0.8644,
+      "step": 2561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154550234255743e-05,
+      "loss": 0.8063,
+      "step": 2562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150148645730837e-05,
+      "loss": 0.7145,
+      "step": 2563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4145746085639315e-05,
+      "loss": 0.7906,
+      "step": 2564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141342555011835e-05,
+      "loss": 0.7957,
+      "step": 2565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4136938054879284e-05,
+      "loss": 0.6517,
+      "step": 2566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4132532586272767e-05,
+      "loss": 0.7625,
+      "step": 2567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4128126150223627e-05,
+      "loss": 0.8145,
+      "step": 2568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123718747763426e-05,
+      "loss": 0.3518,
+      "step": 2569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4119310379923956e-05,
+      "loss": 0.8061,
+      "step": 2570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4114901047737237e-05,
+      "loss": 0.8065,
+      "step": 2571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110490752235506e-05,
+      "loss": 0.7576,
+      "step": 2572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106079494451232e-05,
+      "loss": 0.7556,
+      "step": 2573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101667275417112e-05,
+      "loss": 0.7989,
+      "step": 2574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4097254096166062e-05,
+      "loss": 0.8204,
+      "step": 2575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092839957731224e-05,
+      "loss": 0.7554,
+      "step": 2576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4088424861145966e-05,
+      "loss": 0.8105,
+      "step": 2577
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408400880744388e-05,
+      "loss": 0.7666,
+      "step": 2578
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079591797658785e-05,
+      "loss": 0.8246,
+      "step": 2579
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407517383282472e-05,
+      "loss": 0.7309,
+      "step": 2580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070754913975947e-05,
+      "loss": 0.7498,
+      "step": 2581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066335042146949e-05,
+      "loss": 0.7665,
+      "step": 2582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061914218372441e-05,
+      "loss": 0.7341,
+      "step": 2583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057492443687359e-05,
+      "loss": 0.811,
+      "step": 2584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405306971912685e-05,
+      "loss": 0.7857,
+      "step": 2585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4048646045726298e-05,
+      "loss": 0.7464,
+      "step": 2586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044221424521301e-05,
+      "loss": 0.7238,
+      "step": 2587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039795856547677e-05,
+      "loss": 0.7173,
+      "step": 2588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4035369342841476e-05,
+      "loss": 0.7887,
+      "step": 2589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403094188443896e-05,
+      "loss": 0.7461,
+      "step": 2590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4026513482376611e-05,
+      "loss": 0.7032,
+      "step": 2591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022084137691144e-05,
+      "loss": 0.7316,
+      "step": 2592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401765385141948e-05,
+      "loss": 0.7706,
+      "step": 2593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4013222624598768e-05,
+      "loss": 0.7596,
+      "step": 2594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400879045826638e-05,
+      "loss": 0.3868,
+      "step": 2595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004357353459896e-05,
+      "loss": 0.7653,
+      "step": 2596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999923311217134e-05,
+      "loss": 0.7603,
+      "step": 2597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3995488332576115e-05,
+      "loss": 0.7271,
+      "step": 2598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991052418575086e-05,
+      "loss": 0.7461,
+      "step": 2599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3986615570252513e-05,
+      "loss": 0.8194,
+      "step": 2600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398217778864708e-05,
+      "loss": 0.7571,
+      "step": 2601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977739074797692e-05,
+      "loss": 0.7766,
+      "step": 2602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3973299429743468e-05,
+      "loss": 0.8098,
+      "step": 2603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3968858854523744e-05,
+      "loss": 0.7833,
+      "step": 2604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396441735017808e-05,
+      "loss": 0.7759,
+      "step": 2605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395997491774625e-05,
+      "loss": 0.7468,
+      "step": 2606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955531558268249e-05,
+      "loss": 0.803,
+      "step": 2607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951087272784278e-05,
+      "loss": 0.7651,
+      "step": 2608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3946642062334765e-05,
+      "loss": 0.8135,
+      "step": 2609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3942195927960354e-05,
+      "loss": 0.8034,
+      "step": 2610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937748870701904e-05,
+      "loss": 0.7626,
+      "step": 2611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3933300891600483e-05,
+      "loss": 0.8155,
+      "step": 2612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3928851991697385e-05,
+      "loss": 0.7053,
+      "step": 2613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3924402172034115e-05,
+      "loss": 0.7472,
+      "step": 2614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919951433652395e-05,
+      "loss": 0.7824,
+      "step": 2615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915499777594162e-05,
+      "loss": 0.8378,
+      "step": 2616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391104720490156e-05,
+      "loss": 0.7643,
+      "step": 2617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390659371661696e-05,
+      "loss": 0.7502,
+      "step": 2618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3902139313782942e-05,
+      "loss": 0.6996,
+      "step": 2619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897683997442301e-05,
+      "loss": 0.7953,
+      "step": 2620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389322776863804e-05,
+      "loss": 0.7512,
+      "step": 2621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888770628413379e-05,
+      "loss": 0.7674,
+      "step": 2622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884312577811761e-05,
+      "loss": 0.8254,
+      "step": 2623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3879853617876824e-05,
+      "loss": 0.851,
+      "step": 2624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875393749652434e-05,
+      "loss": 0.7822,
+      "step": 2625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870932974182664e-05,
+      "loss": 0.7741,
+      "step": 2626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38664712925118e-05,
+      "loss": 0.6953,
+      "step": 2627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3862008705684338e-05,
+      "loss": 0.7768,
+      "step": 2628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3857545214744987e-05,
+      "loss": 0.7809,
+      "step": 2629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853080820738671e-05,
+      "loss": 0.6907,
+      "step": 2630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384861552471052e-05,
+      "loss": 0.7082,
+      "step": 2631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844149327705881e-05,
+      "loss": 0.7849,
+      "step": 2632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3839682230770306e-05,
+      "loss": 0.8495,
+      "step": 2633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835214234949561e-05,
+      "loss": 0.7223,
+      "step": 2634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3830745341289621e-05,
+      "loss": 0.7952,
+      "step": 2635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382627555083668e-05,
+      "loss": 0.8422,
+      "step": 2636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382180486463712e-05,
+      "loss": 0.3834,
+      "step": 2637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381733328373756e-05,
+      "loss": 0.717,
+      "step": 2638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3812860809184806e-05,
+      "loss": 0.7138,
+      "step": 2639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808387442025888e-05,
+      "loss": 0.821,
+      "step": 2640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3803913183308036e-05,
+      "loss": 0.7808,
+      "step": 2641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799438034078691e-05,
+      "loss": 0.7258,
+      "step": 2642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3794961995385512e-05,
+      "loss": 0.7088,
+      "step": 2643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379048506827635e-05,
+      "loss": 0.7574,
+      "step": 2644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786007253799271e-05,
+      "loss": 0.7484,
+      "step": 2645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3781528553002554e-05,
+      "loss": 0.8163,
+      "step": 2646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3777048966934681e-05,
+      "loss": 0.7024,
+      "step": 2647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3772568496644336e-05,
+      "loss": 0.8562,
+      "step": 2648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376808714318042e-05,
+      "loss": 0.7615,
+      "step": 2649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763604907592034e-05,
+      "loss": 0.7324,
+      "step": 2650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759121790928491e-05,
+      "loss": 0.7411,
+      "step": 2651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3754637794239303e-05,
+      "loss": 0.7777,
+      "step": 2652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3750152918574192e-05,
+      "loss": 0.7611,
+      "step": 2653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745667164983083e-05,
+      "loss": 0.8346,
+      "step": 2654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741180534516115e-05,
+      "loss": 0.7524,
+      "step": 2655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736693028223627e-05,
+      "loss": 0.7298,
+      "step": 2656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732204647156155e-05,
+      "loss": 0.8099,
+      "step": 2657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3727715392364454e-05,
+      "loss": 0.7977,
+      "step": 2658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372322526489947e-05,
+      "loss": 0.7765,
+      "step": 2659
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371873426581237e-05,
+      "loss": 0.8087,
+      "step": 2660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714242396154504e-05,
+      "loss": 0.78,
+      "step": 2661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709749656977445e-05,
+      "loss": 0.8717,
+      "step": 2662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3705256049332958e-05,
+      "loss": 0.8358,
+      "step": 2663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3700761574273016e-05,
+      "loss": 0.8487,
+      "step": 2664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696266232849796e-05,
+      "loss": 0.8155,
+      "step": 2665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3691770026115667e-05,
+      "loss": 0.7401,
+      "step": 2666
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3687272955123217e-05,
+      "loss": 0.7439,
+      "step": 2667
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682775020925228e-05,
+      "loss": 0.7909,
+      "step": 2668
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678276224574682e-05,
+      "loss": 0.3755,
+      "step": 2669
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673776567124768e-05,
+      "loss": 0.6268,
+      "step": 2670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366927604962887e-05,
+      "loss": 0.7846,
+      "step": 2671
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3664774673140582e-05,
+      "loss": 0.7719,
+      "step": 2672
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366027243871369e-05,
+      "loss": 0.7891,
+      "step": 2673
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655769347402192e-05,
+      "loss": 0.3876,
+      "step": 2674
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651265400260275e-05,
+      "loss": 0.7943,
+      "step": 2675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3646760598342331e-05,
+      "loss": 0.806,
+      "step": 2676
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642254942702953e-05,
+      "loss": 0.7819,
+      "step": 2677
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3637748434396935e-05,
+      "loss": 0.7649,
+      "step": 2678
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633241074479267e-05,
+      "loss": 0.7448,
+      "step": 2679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362873286400514e-05,
+      "loss": 0.7486,
+      "step": 2680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624223804029944e-05,
+      "loss": 0.7781,
+      "step": 2681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361971389560927e-05,
+      "loss": 0.7418,
+      "step": 2682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3615203139798906e-05,
+      "loss": 0.3501,
+      "step": 2683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361069153765484e-05,
+      "loss": 0.8093,
+      "step": 2684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606179090233249e-05,
+      "loss": 0.8913,
+      "step": 2685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3601665798590521e-05,
+      "loss": 0.7452,
+      "step": 2686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359715166378324e-05,
+      "loss": 0.7349,
+      "step": 2687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3592636686868173e-05,
+      "loss": 0.7072,
+      "step": 2688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588120868902299e-05,
+      "loss": 0.7932,
+      "step": 2689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358360421094279e-05,
+      "loss": 0.8453,
+      "step": 2690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3579086714047016e-05,
+      "loss": 0.3533,
+      "step": 2691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3574568379272539e-05,
+      "loss": 0.7691,
+      "step": 2692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570049207677118e-05,
+      "loss": 0.7524,
+      "step": 2693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356552920031871e-05,
+      "loss": 0.7602,
+      "step": 2694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356100835825547e-05,
+      "loss": 0.6999,
+      "step": 2695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3556486682545744e-05,
+      "loss": 0.8156,
+      "step": 2696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355196417424807e-05,
+      "loss": 0.7485,
+      "step": 2697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3547440834421192e-05,
+      "loss": 0.7477,
+      "step": 2698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542916664124042e-05,
+      "loss": 0.8132,
+      "step": 2699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3538391664415742e-05,
+      "loss": 0.8206,
+      "step": 2700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3533865836355616e-05,
+      "loss": 0.7071,
+      "step": 2701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529339181003175e-05,
+      "loss": 0.7972,
+      "step": 2702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3524811699418132e-05,
+      "loss": 0.7986,
+      "step": 2703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520283392660387e-05,
+      "loss": 0.7135,
+      "step": 2704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515754261790035e-05,
+      "loss": 0.3673,
+      "step": 2705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511224307867363e-05,
+      "loss": 0.7711,
+      "step": 2706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350669353195285e-05,
+      "loss": 0.78,
+      "step": 2707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3502161935107177e-05,
+      "loss": 0.8012,
+      "step": 2708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3497629518391198e-05,
+      "loss": 0.7171,
+      "step": 2709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493096282865979e-05,
+      "loss": 0.7836,
+      "step": 2710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3488562229592766e-05,
+      "loss": 0.7551,
+      "step": 2711
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484027359632993e-05,
+      "loss": 0.8174,
+      "step": 2712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3479491674048303e-05,
+      "loss": 0.8504,
+      "step": 2713
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347495517390051e-05,
+      "loss": 0.3441,
+      "step": 2714
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347041786025163e-05,
+      "loss": 0.7858,
+      "step": 2715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465879734163867e-05,
+      "loss": 0.831,
+      "step": 2716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346134079669961e-05,
+      "loss": 0.7329,
+      "step": 2717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345680104892145e-05,
+      "loss": 0.779,
+      "step": 2718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3452260491892157e-05,
+      "loss": 0.36,
+      "step": 2719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3447719126674692e-05,
+      "loss": 0.8233,
+      "step": 2720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344317695433221e-05,
+      "loss": 0.7102,
+      "step": 2721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438633975928049e-05,
+      "loss": 0.7661,
+      "step": 2722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434090192525739e-05,
+      "loss": 0.7445,
+      "step": 2723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3429545605189e-05,
+      "loss": 0.8821,
+      "step": 2724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3425000214981739e-05,
+      "loss": 0.7623,
+      "step": 2725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3420454022968048e-05,
+      "loss": 0.7351,
+      "step": 2726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3415907030212209e-05,
+      "loss": 0.7996,
+      "step": 2727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411359237778694e-05,
+      "loss": 0.7377,
+      "step": 2728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406810646732158e-05,
+      "loss": 0.7551,
+      "step": 2729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3402261258137442e-05,
+      "loss": 0.7783,
+      "step": 2730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3397711073059579e-05,
+      "loss": 0.7908,
+      "step": 2731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393160092563789e-05,
+      "loss": 0.7981,
+      "step": 2732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388608317715468e-05,
+      "loss": 0.7566,
+      "step": 2733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3384055749580209e-05,
+      "loss": 0.7653,
+      "step": 2734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3379502389223786e-05,
+      "loss": 0.8001,
+      "step": 2735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374948237712162e-05,
+      "loss": 0.8163,
+      "step": 2736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370393296111477e-05,
+      "loss": 0.786,
+      "step": 2737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365837565488065e-05,
+      "loss": 0.7595,
+      "step": 2738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3361281046908436e-05,
+      "loss": 0.7111,
+      "step": 2739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3356723741439298e-05,
+      "loss": 0.365,
+      "step": 2740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352165650147528e-05,
+      "loss": 0.7713,
+      "step": 2741
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347606774100197e-05,
+      "loss": 0.7405,
+      "step": 2742
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343047114364553e-05,
+      "loss": 0.7632,
+      "step": 2743
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3338486672008033e-05,
+      "loss": 0.8368,
+      "step": 2744
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3333925448098257e-05,
+      "loss": 0.8201,
+      "step": 2745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329363443703022e-05,
+      "loss": 0.7755,
+      "step": 2746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324800659890313e-05,
+      "loss": 0.7092,
+      "step": 2747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320237097728296e-05,
+      "loss": 0.7932,
+      "step": 2748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3315672758285323e-05,
+      "loss": 0.7578,
+      "step": 2749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331110764262992e-05,
+      "loss": 0.7436,
+      "step": 2750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306541751830802e-05,
+      "loss": 0.7893,
+      "step": 2751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3301975086956857e-05,
+      "loss": 0.7182,
+      "step": 2752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297407649077168e-05,
+      "loss": 0.7706,
+      "step": 2753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3292839439260986e-05,
+      "loss": 0.3623,
+      "step": 2754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288270458577745e-05,
+      "loss": 0.7392,
+      "step": 2755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283700708097067e-05,
+      "loss": 0.3615,
+      "step": 2756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279130188888747e-05,
+      "loss": 0.8801,
+      "step": 2757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327455890202276e-05,
+      "loss": 0.7671,
+      "step": 2758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326998684856927e-05,
+      "loss": 0.8257,
+      "step": 2759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3265414029598604e-05,
+      "loss": 0.7842,
+      "step": 2760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3260840446181283e-05,
+      "loss": 0.7374,
+      "step": 2761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256266099387997e-05,
+      "loss": 0.7416,
+      "step": 2762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251690990289625e-05,
+      "loss": 0.8072,
+      "step": 2763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3247115119957214e-05,
+      "loss": 0.7027,
+      "step": 2764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242538489461998e-05,
+      "loss": 0.7576,
+      "step": 2765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237961099875379e-05,
+      "loss": 0.7302,
+      "step": 2766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233382952268948e-05,
+      "loss": 0.8257,
+      "step": 2767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228804047714462e-05,
+      "loss": 0.7333,
+      "step": 2768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3224224387283869e-05,
+      "loss": 0.7343,
+      "step": 2769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219643972049278e-05,
+      "loss": 0.8134,
+      "step": 2770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3215062803082987e-05,
+      "loss": 0.8218,
+      "step": 2771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3210480881457471e-05,
+      "loss": 0.7559,
+      "step": 2772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3205898208245365e-05,
+      "loss": 0.7047,
+      "step": 2773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3201314784519498e-05,
+      "loss": 0.8527,
+      "step": 2774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3196730611352871e-05,
+      "loss": 0.8317,
+      "step": 2775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319214568981865e-05,
+      "loss": 0.8192,
+      "step": 2776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318756002099019e-05,
+      "loss": 0.7246,
+      "step": 2777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3182973605941013e-05,
+      "loss": 0.7827,
+      "step": 2778
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3178386445744814e-05,
+      "loss": 0.7793,
+      "step": 2779
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173798541475471e-05,
+      "loss": 0.8349,
+      "step": 2780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316920989420703e-05,
+      "loss": 0.8033,
+      "step": 2781
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3164620505013707e-05,
+      "loss": 0.7271,
+      "step": 2782
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3160030374969901e-05,
+      "loss": 0.7494,
+      "step": 2783
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155439505150178e-05,
+      "loss": 0.8438,
+      "step": 2784
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3150847896629284e-05,
+      "loss": 0.7929,
+      "step": 2785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146255550482128e-05,
+      "loss": 0.8175,
+      "step": 2786
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141662467783795e-05,
+      "loss": 0.8013,
+      "step": 2787
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3137068649609553e-05,
+      "loss": 0.7633,
+      "step": 2788
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132474097034824e-05,
+      "loss": 0.7677,
+      "step": 2789
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3127878811135215e-05,
+      "loss": 0.7581,
+      "step": 2790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123282792986503e-05,
+      "loss": 0.7539,
+      "step": 2791
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118686043664635e-05,
+      "loss": 0.8117,
+      "step": 2792
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3114088564245723e-05,
+      "loss": 0.737,
+      "step": 2793
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310949035580606e-05,
+      "loss": 0.8194,
+      "step": 2794
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104891419422105e-05,
+      "loss": 0.7554,
+      "step": 2795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100291756170489e-05,
+      "loss": 0.7918,
+      "step": 2796
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095691367128007e-05,
+      "loss": 0.7923,
+      "step": 2797
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3091090253371632e-05,
+      "loss": 0.793,
+      "step": 2798
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3086488415978505e-05,
+      "loss": 0.7148,
+      "step": 2799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3081885856025931e-05,
+      "loss": 0.788,
+      "step": 2800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307728257459139e-05,
+      "loss": 0.8,
+      "step": 2801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3072678572752526e-05,
+      "loss": 0.818,
+      "step": 2802
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306807385158716e-05,
+      "loss": 0.7863,
+      "step": 2803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063468412173275e-05,
+      "loss": 0.3786,
+      "step": 2804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3058862255589016e-05,
+      "loss": 0.7606,
+      "step": 2805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305425538291271e-05,
+      "loss": 0.7872,
+      "step": 2806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304964779522284e-05,
+      "loss": 0.7403,
+      "step": 2807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3045039493598068e-05,
+      "loss": 0.7859,
+      "step": 2808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040430479117207e-05,
+      "loss": 0.7345,
+      "step": 2809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3035820752859251e-05,
+      "loss": 0.7666,
+      "step": 2810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3031210315903356e-05,
+      "loss": 0.7213,
+      "step": 2811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3026599169328843e-05,
+      "loss": 0.7533,
+      "step": 2812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021987314215201e-05,
+      "loss": 0.8,
+      "step": 2813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017374751642083e-05,
+      "loss": 0.7383,
+      "step": 2814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3012761482689306e-05,
+      "loss": 0.7713,
+      "step": 2815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300814750843686e-05,
+      "loss": 0.7957,
+      "step": 2816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300353282996489e-05,
+      "loss": 0.7997,
+      "step": 2817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998917448353716e-05,
+      "loss": 0.7326,
+      "step": 2818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2994301364683813e-05,
+      "loss": 0.6994,
+      "step": 2819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2989684580035828e-05,
+      "loss": 0.7669,
+      "step": 2820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985067095490566e-05,
+      "loss": 0.7723,
+      "step": 2821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980448912129006e-05,
+      "loss": 0.74,
+      "step": 2822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2975830031032274e-05,
+      "loss": 0.804,
+      "step": 2823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2971210453281675e-05,
+      "loss": 0.7732,
+      "step": 2824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966590179958667e-05,
+      "loss": 0.7465,
+      "step": 2825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2961969212144881e-05,
+      "loss": 0.8075,
+      "step": 2826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957347550922095e-05,
+      "loss": 0.7721,
+      "step": 2827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952725197372264e-05,
+      "loss": 0.7818,
+      "step": 2828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29481021525775e-05,
+      "loss": 0.7811,
+      "step": 2829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294347841762008e-05,
+      "loss": 0.7446,
+      "step": 2830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293885399358243e-05,
+      "loss": 0.765,
+      "step": 2831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293422888154715e-05,
+      "loss": 0.7821,
+      "step": 2832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2929603082597e-05,
+      "loss": 0.8172,
+      "step": 2833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2924976597814897e-05,
+      "loss": 0.8159,
+      "step": 2834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2920349428283922e-05,
+      "loss": 0.6241,
+      "step": 2835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291572157508731e-05,
+      "loss": 0.782,
+      "step": 2836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911093039308465e-05,
+      "loss": 0.753,
+      "step": 2837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290646382203094e-05,
+      "loss": 0.8104,
+      "step": 2838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2901833924338459e-05,
+      "loss": 0.7626,
+      "step": 2839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897203347314893e-05,
+      "loss": 0.7532,
+      "step": 2840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2892572092044286e-05,
+      "loss": 0.7374,
+      "step": 2841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288794015961083e-05,
+      "loss": 0.75,
+      "step": 2842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2883307551098882e-05,
+      "loss": 0.7802,
+      "step": 2843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2878674267592954e-05,
+      "loss": 0.7438,
+      "step": 2844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.287404031017771e-05,
+      "loss": 0.7721,
+      "step": 2845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869405679937988e-05,
+      "loss": 0.6914,
+      "step": 2846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864770377958765e-05,
+      "loss": 0.7473,
+      "step": 2847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2860134405325191e-05,
+      "loss": 0.7385,
+      "step": 2848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285549776312256e-05,
+      "loss": 0.3863,
+      "step": 2849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2850860452436333e-05,
+      "loss": 0.7419,
+      "step": 2850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284622247435212e-05,
+      "loss": 0.8,
+      "step": 2851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841583829955695e-05,
+      "loss": 0.7595,
+      "step": 2852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836944520332981e-05,
+      "loss": 0.8124,
+      "step": 2853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2832304546570056e-05,
+      "loss": 0.8005,
+      "step": 2854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282766390975316e-05,
+      "loss": 0.7619,
+      "step": 2855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2823022610968686e-05,
+      "loss": 0.7643,
+      "step": 2856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818380651303179e-05,
+      "loss": 0.7688,
+      "step": 2857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813738031843341e-05,
+      "loss": 0.7955,
+      "step": 2858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2809094753676026e-05,
+      "loss": 0.7113,
+      "step": 2859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804450817888247e-05,
+      "loss": 0.7882,
+      "step": 2860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799806225567168e-05,
+      "loss": 0.7189,
+      "step": 2861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795160977800107e-05,
+      "loss": 0.7586,
+      "step": 2862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2790515075674535e-05,
+      "loss": 0.8534,
+      "step": 2863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785868520278073e-05,
+      "loss": 0.7751,
+      "step": 2864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781221312698502e-05,
+      "loss": 0.7239,
+      "step": 2865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2776573454023753e-05,
+      "loss": 0.7382,
+      "step": 2866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2771924945341906e-05,
+      "loss": 0.8673,
+      "step": 2867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767275787741197e-05,
+      "loss": 0.7828,
+      "step": 2868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2762625982310015e-05,
+      "loss": 0.8194,
+      "step": 2869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2757975530136895e-05,
+      "loss": 0.8595,
+      "step": 2870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2753324432310525e-05,
+      "loss": 0.8223,
+      "step": 2871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748672689919747e-05,
+      "loss": 0.746,
+      "step": 2872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2744020304053559e-05,
+      "loss": 0.6948,
+      "step": 2873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2739367275801095e-05,
+      "loss": 0.7959,
+      "step": 2874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2734713606251652e-05,
+      "loss": 0.7662,
+      "step": 2875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730059296494674e-05,
+      "loss": 0.3782,
+      "step": 2876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272540434761975e-05,
+      "loss": 0.7817,
+      "step": 2877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2720748760716624e-05,
+      "loss": 0.7682,
+      "step": 2878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271609253687519e-05,
+      "loss": 0.7904,
+      "step": 2879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711435677185486e-05,
+      "loss": 0.7763,
+      "step": 2880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2706778182737702e-05,
+      "loss": 0.8596,
+      "step": 2881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702120054622176e-05,
+      "loss": 0.7772,
+      "step": 2882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697461293929396e-05,
+      "loss": 0.7881,
+      "step": 2883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692801901750001e-05,
+      "loss": 0.8093,
+      "step": 2884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2688141879174763e-05,
+      "loss": 0.8029,
+      "step": 2885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2683481227294618e-05,
+      "loss": 0.835,
+      "step": 2886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2678819947200646e-05,
+      "loss": 0.7444,
+      "step": 2887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267415803998407e-05,
+      "loss": 0.7603,
+      "step": 2888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669495506736262e-05,
+      "loss": 0.7314,
+      "step": 2889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2664832348548735e-05,
+      "loss": 0.8043,
+      "step": 2890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2660168566513157e-05,
+      "loss": 0.8126,
+      "step": 2891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265550416172134e-05,
+      "loss": 0.6983,
+      "step": 2892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2650839135265237e-05,
+      "loss": 0.7306,
+      "step": 2893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264617348823695e-05,
+      "loss": 0.7825,
+      "step": 2894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641507221728726e-05,
+      "loss": 0.7457,
+      "step": 2895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2636840336832958e-05,
+      "loss": 0.7571,
+      "step": 2896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2632172834642184e-05,
+      "loss": 0.794,
+      "step": 2897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627504716249081e-05,
+      "loss": 0.736,
+      "step": 2898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262283598274648e-05,
+      "loss": 0.7214,
+      "step": 2899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2618166635227342e-05,
+      "loss": 0.71,
+      "step": 2900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613496674784789e-05,
+      "loss": 0.8037,
+      "step": 2901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2608826102512076e-05,
+      "loss": 0.7889,
+      "step": 2902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2604154919502595e-05,
+      "loss": 0.7397,
+      "step": 2903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2599483126849898e-05,
+      "loss": 0.7863,
+      "step": 2904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2594810725647668e-05,
+      "loss": 0.3839,
+      "step": 2905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259013771698973e-05,
+      "loss": 0.7987,
+      "step": 2906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258546410197006e-05,
+      "loss": 0.7531,
+      "step": 2907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2580789881682764e-05,
+      "loss": 0.7563,
+      "step": 2908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576115057222098e-05,
+      "loss": 0.7528,
+      "step": 2909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257143962968246e-05,
+      "loss": 0.7802,
+      "step": 2910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2566763600158386e-05,
+      "loss": 0.7553,
+      "step": 2911
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2562086969744551e-05,
+      "loss": 0.799,
+      "step": 2912
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2557409739535773e-05,
+      "loss": 0.7672,
+      "step": 2913
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552731910627011e-05,
+      "loss": 0.7618,
+      "step": 2914
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2548053484113369e-05,
+      "loss": 0.6618,
+      "step": 2915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2543374461090079e-05,
+      "loss": 0.7723,
+      "step": 2916
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2538694842652519e-05,
+      "loss": 0.7576,
+      "step": 2917
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253401462989621e-05,
+      "loss": 0.7707,
+      "step": 2918
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529333823916807e-05,
+      "loss": 0.8662,
+      "step": 2919
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524652425810111e-05,
+      "loss": 0.8466,
+      "step": 2920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2519970436672049e-05,
+      "loss": 0.8557,
+      "step": 2921
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515287857598692e-05,
+      "loss": 0.7105,
+      "step": 2922
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510604689686262e-05,
+      "loss": 0.77,
+      "step": 2923
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505920934031098e-05,
+      "loss": 0.7858,
+      "step": 2924
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501236591729687e-05,
+      "loss": 0.3608,
+      "step": 2925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2496551663878655e-05,
+      "loss": 0.7862,
+      "step": 2926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249186615157476e-05,
+      "loss": 0.7891,
+      "step": 2927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2487180055914905e-05,
+      "loss": 0.6477,
+      "step": 2928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248249337799612e-05,
+      "loss": 0.7101,
+      "step": 2929
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2477806118915575e-05,
+      "loss": 0.7866,
+      "step": 2930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473118279770575e-05,
+      "loss": 0.8148,
+      "step": 2931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2468429861658567e-05,
+      "loss": 0.7026,
+      "step": 2932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463740865677125e-05,
+      "loss": 0.7981,
+      "step": 2933
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2459051292923963e-05,
+      "loss": 0.7084,
+      "step": 2934
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2454361144496926e-05,
+      "loss": 0.3595,
+      "step": 2935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2449670421494003e-05,
+      "loss": 0.7805,
+      "step": 2936
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2444979125013306e-05,
+      "loss": 0.743,
+      "step": 2937
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244028725615309e-05,
+      "loss": 0.7366,
+      "step": 2938
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435594816011736e-05,
+      "loss": 0.7485,
+      "step": 2939
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2430901805687766e-05,
+      "loss": 0.6926,
+      "step": 2940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2426208226279832e-05,
+      "loss": 0.8016,
+      "step": 2941
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2421514078886726e-05,
+      "loss": 0.838,
+      "step": 2942
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2416819364607358e-05,
+      "loss": 0.7858,
+      "step": 2943
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2412124084540781e-05,
+      "loss": 0.8106,
+      "step": 2944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2407428239786185e-05,
+      "loss": 0.7633,
+      "step": 2945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240273183144288e-05,
+      "loss": 0.7562,
+      "step": 2946
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398034860610322e-05,
+      "loss": 0.8428,
+      "step": 2947
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239333732838808e-05,
+      "loss": 0.7982,
+      "step": 2948
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388639235875872e-05,
+      "loss": 0.7493,
+      "step": 2949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383940584173544e-05,
+      "loss": 0.771,
+      "step": 2950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2379241374381062e-05,
+      "loss": 0.737,
+      "step": 2951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374541607598531e-05,
+      "loss": 0.6973,
+      "step": 2952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236984128492619e-05,
+      "loss": 0.75,
+      "step": 2953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365140407464402e-05,
+      "loss": 0.7718,
+      "step": 2954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360438976313658e-05,
+      "loss": 0.7159,
+      "step": 2955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2355736992574584e-05,
+      "loss": 0.7325,
+      "step": 2956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351034457347939e-05,
+      "loss": 0.7484,
+      "step": 2957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346331371734597e-05,
+      "loss": 0.7782,
+      "step": 2958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2341627736835573e-05,
+      "loss": 0.7763,
+      "step": 2959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2336923553752007e-05,
+      "loss": 0.3506,
+      "step": 2960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2332218823585169e-05,
+      "loss": 0.7032,
+      "step": 2961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327513547436451e-05,
+      "loss": 0.7924,
+      "step": 2962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2322807726407382e-05,
+      "loss": 0.7345,
+      "step": 2963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231810136159961e-05,
+      "loss": 0.8723,
+      "step": 2964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231339445411492e-05,
+      "loss": 0.7402,
+      "step": 2965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230868700505521e-05,
+      "loss": 0.8409,
+      "step": 2966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2303979015522518e-05,
+      "loss": 0.7714,
+      "step": 2967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299270486619002e-05,
+      "loss": 0.7784,
+      "step": 2968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2294561419446949e-05,
+      "loss": 0.8284,
+      "step": 2969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2289851815108768e-05,
+      "loss": 0.7292,
+      "step": 2970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2285141674706997e-05,
+      "loss": 0.7679,
+      "step": 2971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280430999344302e-05,
+      "loss": 0.8119,
+      "step": 2972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2275719790123469e-05,
+      "loss": 0.8229,
+      "step": 2973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2271008048147413e-05,
+      "loss": 0.7897,
+      "step": 2974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266295774519168e-05,
+      "loss": 0.7388,
+      "step": 2975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2261582970341898e-05,
+      "loss": 0.7943,
+      "step": 2976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2256869636718891e-05,
+      "loss": 0.8256,
+      "step": 2977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225215577475356e-05,
+      "loss": 0.6945,
+      "step": 2978
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2247441385549433e-05,
+      "loss": 0.8059,
+      "step": 2979
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224272647021017e-05,
+      "loss": 0.7732,
+      "step": 2980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238011029839552e-05,
+      "loss": 0.6771,
+      "step": 2981
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233295065541483e-05,
+      "loss": 0.7884,
+      "step": 2982
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2228578578419992e-05,
+      "loss": 0.7608,
+      "step": 2983
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2223861569579223e-05,
+      "loss": 0.7768,
+      "step": 2984
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221914404012345e-05,
+      "loss": 0.7952,
+      "step": 2985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2214425991157065e-05,
+      "loss": 0.7913,
+      "step": 2986
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2209707423784582e-05,
+      "loss": 0.6959,
+      "step": 2987
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2204988339110635e-05,
+      "loss": 0.76,
+      "step": 2988
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200268738239985e-05,
+      "loss": 0.7146,
+      "step": 2989
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2195548622277505e-05,
+      "loss": 0.7081,
+      "step": 2990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2190827992328196e-05,
+      "loss": 0.7783,
+      "step": 2991
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218610684949718e-05,
+      "loss": 0.709,
+      "step": 2992
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181385194889688e-05,
+      "loss": 0.7487,
+      "step": 2993
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2176663029611083e-05,
+      "loss": 0.7921,
+      "step": 2994
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2171940354766842e-05,
+      "loss": 0.7835,
+      "step": 2995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167217171462566e-05,
+      "loss": 0.8219,
+      "step": 2996
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2162493480803965e-05,
+      "loss": 0.7933,
+      "step": 2997
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2157769283896878e-05,
+      "loss": 0.8711,
+      "step": 2998
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215304458184726e-05,
+      "loss": 0.8813,
+      "step": 2999
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2148319375761177e-05,
+      "loss": 0.8222,
+      "step": 3000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143593666744826e-05,
+      "loss": 0.7662,
+      "step": 3001
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2138867455904511e-05,
+      "loss": 0.6793,
+      "step": 3002
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2134140744346654e-05,
+      "loss": 0.7601,
+      "step": 3003
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2129413533177803e-05,
+      "loss": 0.8113,
+      "step": 3004
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2124685823504618e-05,
+      "loss": 0.7367,
+      "step": 3005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2119957616433868e-05,
+      "loss": 0.7098,
+      "step": 3006
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115228913072449e-05,
+      "loss": 0.787,
+      "step": 3007
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110499714527373e-05,
+      "loss": 0.7591,
+      "step": 3008
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2105770021905763e-05,
+      "loss": 0.812,
+      "step": 3009
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2101039836314855e-05,
+      "loss": 0.736,
+      "step": 3010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2096309158862004e-05,
+      "loss": 0.7744,
+      "step": 3011
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091577990654688e-05,
+      "loss": 0.7488,
+      "step": 3012
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208684633280049e-05,
+      "loss": 0.7178,
+      "step": 3013
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208211418640711e-05,
+      "loss": 0.7924,
+      "step": 3014
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077381552582359e-05,
+      "loss": 0.743,
+      "step": 3015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072648432434169e-05,
+      "loss": 0.6477,
+      "step": 3016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067914827070581e-05,
+      "loss": 0.7505,
+      "step": 3017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2063180737599755e-05,
+      "loss": 0.6997,
+      "step": 3018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2058446165129956e-05,
+      "loss": 0.7749,
+      "step": 3019
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2053711110769568e-05,
+      "loss": 0.7841,
+      "step": 3020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048975575627088e-05,
+      "loss": 0.7187,
+      "step": 3021
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2044239560811121e-05,
+      "loss": 0.7467,
+      "step": 3022
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2039503067430394e-05,
+      "loss": 0.8235,
+      "step": 3023
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203476609659373e-05,
+      "loss": 0.77,
+      "step": 3024
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203002864941008e-05,
+      "loss": 0.7539,
+      "step": 3025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025290726988495e-05,
+      "loss": 0.8059,
+      "step": 3026
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020552330438146e-05,
+      "loss": 0.7902,
+      "step": 3027
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015813460868306e-05,
+      "loss": 0.7689,
+      "step": 3028
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2011074119388369e-05,
+      "loss": 0.3461,
+      "step": 3029
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2006334307107828e-05,
+      "loss": 0.7795,
+      "step": 3030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001594025136301e-05,
+      "loss": 0.8054,
+      "step": 3031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1996853274583497e-05,
+      "loss": 0.6978,
+      "step": 3032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199211205655925e-05,
+      "loss": 0.7183,
+      "step": 3033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19873703721735e-05,
+      "loss": 0.7992,
+      "step": 3034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982628222536292e-05,
+      "loss": 0.6834,
+      "step": 3035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977885608757785e-05,
+      "loss": 0.7904,
+      "step": 3036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973142531948238e-05,
+      "loss": 0.8248,
+      "step": 3037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1968398993218034e-05,
+      "loss": 0.7193,
+      "step": 3038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1963654993677645e-05,
+      "loss": 0.6916,
+      "step": 3039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1958910534437665e-05,
+      "loss": 0.7866,
+      "step": 3040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195416561660879e-05,
+      "loss": 0.7269,
+      "step": 3041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194942024130183e-05,
+      "loss": 0.7559,
+      "step": 3042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1944674409627688e-05,
+      "loss": 0.7994,
+      "step": 3043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1939928122697388e-05,
+      "loss": 0.8214,
+      "step": 3044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935181381622055e-05,
+      "loss": 0.699,
+      "step": 3045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1930434187512916e-05,
+      "loss": 0.7826,
+      "step": 3046
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1925686541481312e-05,
+      "loss": 0.7568,
+      "step": 3047
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920938444638685e-05,
+      "loss": 0.3729,
+      "step": 3048
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1916189898096584e-05,
+      "loss": 0.752,
+      "step": 3049
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1911440902966667e-05,
+      "loss": 0.7242,
+      "step": 3050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1906691460360685e-05,
+      "loss": 0.7725,
+      "step": 3051
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1901941571390506e-05,
+      "loss": 0.8191,
+      "step": 3052
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1897191237168098e-05,
+      "loss": 0.7543,
+      "step": 3053
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892440458805536e-05,
+      "loss": 0.816,
+      "step": 3054
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188768923741499e-05,
+      "loss": 0.3353,
+      "step": 3055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1882937574108746e-05,
+      "loss": 0.7247,
+      "step": 3056
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878185469999187e-05,
+      "loss": 0.7851,
+      "step": 3057
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1873432926198798e-05,
+      "loss": 0.8607,
+      "step": 3058
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186867994382017e-05,
+      "loss": 0.7702,
+      "step": 3059
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863926523975992e-05,
+      "loss": 0.7228,
+      "step": 3060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1859172667779064e-05,
+      "loss": 0.7826,
+      "step": 3061
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185441837634228e-05,
+      "loss": 0.3364,
+      "step": 3062
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849663650778642e-05,
+      "loss": 0.7657,
+      "step": 3063
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1844908492201246e-05,
+      "loss": 0.8013,
+      "step": 3064
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1840152901723296e-05,
+      "loss": 0.7228,
+      "step": 3065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835396880458095e-05,
+      "loss": 0.7818,
+      "step": 3066
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830640429519046e-05,
+      "loss": 0.7644,
+      "step": 3067
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1825883550019657e-05,
+      "loss": 0.8351,
+      "step": 3068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821126243073528e-05,
+      "loss": 0.7498,
+      "step": 3069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816368509794365e-05,
+      "loss": 0.7378,
+      "step": 3070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811610351295974e-05,
+      "loss": 0.7626,
+      "step": 3071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806851768692262e-05,
+      "loss": 0.8264,
+      "step": 3072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1802092763097225e-05,
+      "loss": 0.8306,
+      "step": 3073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797333335624972e-05,
+      "loss": 0.7436,
+      "step": 3074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1792573487389703e-05,
+      "loss": 0.7577,
+      "step": 3075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1787813219505713e-05,
+      "loss": 0.7165,
+      "step": 3076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783052533087407e-05,
+      "loss": 0.748,
+      "step": 3077
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778291429249276e-05,
+      "loss": 0.7543,
+      "step": 3078
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1773529909105916e-05,
+      "loss": 0.7832,
+      "step": 3079
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1768767973772017e-05,
+      "loss": 0.82,
+      "step": 3080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176400562436237e-05,
+      "loss": 0.7904,
+      "step": 3081
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759242861991855e-05,
+      "loss": 0.749,
+      "step": 3082
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1754479687775459e-05,
+      "loss": 0.74,
+      "step": 3083
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749716102828259e-05,
+      "loss": 0.741,
+      "step": 3084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1744952108265426e-05,
+      "loss": 0.7508,
+      "step": 3085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174018770520224e-05,
+      "loss": 0.7582,
+      "step": 3086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1735422894754053e-05,
+      "loss": 0.6746,
+      "step": 3087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730657678036339e-05,
+      "loss": 0.7421,
+      "step": 3088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725892056164648e-05,
+      "loss": 0.7748,
+      "step": 3089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1721126030254634e-05,
+      "loss": 0.8103,
+      "step": 3090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171635960142204e-05,
+      "loss": 0.3868,
+      "step": 3091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1711592770782709e-05,
+      "loss": 0.6721,
+      "step": 3092
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706825539452575e-05,
+      "loss": 0.6744,
+      "step": 3093
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1702057908547666e-05,
+      "loss": 0.7101,
+      "step": 3094
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1697289879184104e-05,
+      "loss": 0.7932,
+      "step": 3095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16925214524781e-05,
+      "loss": 0.7465,
+      "step": 3096
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168775262954597e-05,
+      "loss": 0.7777,
+      "step": 3097
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1682983411504108e-05,
+      "loss": 0.7694,
+      "step": 3098
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1678213799469011e-05,
+      "loss": 0.7333,
+      "step": 3099
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673443794557262e-05,
+      "loss": 0.7474,
+      "step": 3100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166867339788554e-05,
+      "loss": 0.7449,
+      "step": 3101
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663902610570612e-05,
+      "loss": 0.7182,
+      "step": 3102
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659131433729343e-05,
+      "loss": 0.6792,
+      "step": 3103
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1654359868478679e-05,
+      "loss": 0.7723,
+      "step": 3104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1649587915935668e-05,
+      "loss": 0.7155,
+      "step": 3105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1644815577217441e-05,
+      "loss": 0.7376,
+      "step": 3106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640042853441223e-05,
+      "loss": 0.7349,
+      "step": 3107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1635269745724328e-05,
+      "loss": 0.7759,
+      "step": 3108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630496255184156e-05,
+      "loss": 0.7662,
+      "step": 3109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1625722382938205e-05,
+      "loss": 0.7735,
+      "step": 3110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620948130104058e-05,
+      "loss": 0.8381,
+      "step": 3111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616173497799383e-05,
+      "loss": 0.7722,
+      "step": 3112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1611398487141945e-05,
+      "loss": 0.7828,
+      "step": 3113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160662309924959e-05,
+      "loss": 0.7358,
+      "step": 3114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1601847335240257e-05,
+      "loss": 0.7514,
+      "step": 3115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597071196231972e-05,
+      "loss": 0.7759,
+      "step": 3116
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1592294683342849e-05,
+      "loss": 0.6742,
+      "step": 3117
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1587517797691088e-05,
+      "loss": 0.7194,
+      "step": 3118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582740540394978e-05,
+      "loss": 0.7874,
+      "step": 3119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577962912572893e-05,
+      "loss": 0.7523,
+      "step": 3120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573184915343296e-05,
+      "loss": 0.737,
+      "step": 3121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1568406549824735e-05,
+      "loss": 0.7699,
+      "step": 3122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1563627817135848e-05,
+      "loss": 0.7637,
+      "step": 3123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155884871839535e-05,
+      "loss": 0.3879,
+      "step": 3124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155406925472205e-05,
+      "loss": 0.7816,
+      "step": 3125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1549289427234843e-05,
+      "loss": 0.3602,
+      "step": 3126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1544509237052701e-05,
+      "loss": 0.7064,
+      "step": 3127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1539728685294687e-05,
+      "loss": 0.7589,
+      "step": 3128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153494777307995e-05,
+      "loss": 0.7672,
+      "step": 3129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1530166501527719e-05,
+      "loss": 0.8293,
+      "step": 3130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152538487175731e-05,
+      "loss": 0.8271,
+      "step": 3131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1520602884888118e-05,
+      "loss": 0.6499,
+      "step": 3132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1515820542039633e-05,
+      "loss": 0.6819,
+      "step": 3133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1511037844331414e-05,
+      "loss": 0.8162,
+      "step": 3134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1506254792883116e-05,
+      "loss": 0.73,
+      "step": 3135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501471388814463e-05,
+      "loss": 0.6684,
+      "step": 3136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1496687633245276e-05,
+      "loss": 0.7547,
+      "step": 3137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491903527295447e-05,
+      "loss": 0.729,
+      "step": 3138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148711907208496e-05,
+      "loss": 0.7965,
+      "step": 3139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148233426873387e-05,
+      "loss": 0.8187,
+      "step": 3140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477549118362321e-05,
+      "loss": 0.7105,
+      "step": 3141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1472763622090538e-05,
+      "loss": 0.8649,
+      "step": 3142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146797778103882e-05,
+      "loss": 0.7029,
+      "step": 3143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1463191596327559e-05,
+      "loss": 0.78,
+      "step": 3144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1458405069077211e-05,
+      "loss": 0.7981,
+      "step": 3145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145361820040833e-05,
+      "loss": 0.7168,
+      "step": 3146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1448830991441534e-05,
+      "loss": 0.7832,
+      "step": 3147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1444043443297539e-05,
+      "loss": 0.7091,
+      "step": 3148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439255557097118e-05,
+      "loss": 0.7706,
+      "step": 3149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434467333961139e-05,
+      "loss": 0.7507,
+      "step": 3150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429678775010544e-05,
+      "loss": 0.775,
+      "step": 3151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1424889881366355e-05,
+      "loss": 0.7619,
+      "step": 3152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142010065414967e-05,
+      "loss": 0.7271,
+      "step": 3153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1415311094481669e-05,
+      "loss": 0.3531,
+      "step": 3154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410521203483605e-05,
+      "loss": 0.7562,
+      "step": 3155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405730982276811e-05,
+      "loss": 0.7712,
+      "step": 3156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1400940431982699e-05,
+      "loss": 0.7347,
+      "step": 3157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396149553722753e-05,
+      "loss": 0.7535,
+      "step": 3158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.139135834861854e-05,
+      "loss": 0.331,
+      "step": 3159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386566817791697e-05,
+      "loss": 0.7472,
+      "step": 3160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381774962363948e-05,
+      "loss": 0.8079,
+      "step": 3161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137698278345708e-05,
+      "loss": 0.7618,
+      "step": 3162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137219028219296e-05,
+      "loss": 0.7029,
+      "step": 3163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367397459693534e-05,
+      "loss": 0.7871,
+      "step": 3164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1362604317080825e-05,
+      "loss": 0.7757,
+      "step": 3165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1357810855476923e-05,
+      "loss": 0.7355,
+      "step": 3166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1353017076003998e-05,
+      "loss": 0.7871,
+      "step": 3167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1348222979784289e-05,
+      "loss": 0.743,
+      "step": 3168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134342856794012e-05,
+      "loss": 0.7581,
+      "step": 3169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133863384159388e-05,
+      "loss": 0.7628,
+      "step": 3170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1333838801868033e-05,
+      "loss": 0.8234,
+      "step": 3171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329043449885118e-05,
+      "loss": 0.7916,
+      "step": 3172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324247786767743e-05,
+      "loss": 0.812,
+      "step": 3173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1319451813638595e-05,
+      "loss": 0.744,
+      "step": 3174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1314655531620435e-05,
+      "loss": 0.7435,
+      "step": 3175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1309858941836085e-05,
+      "loss": 0.7917,
+      "step": 3176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1305062045408448e-05,
+      "loss": 0.7232,
+      "step": 3177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13002648434605e-05,
+      "loss": 0.353,
+      "step": 3178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129546733711528e-05,
+      "loss": 0.7044,
+      "step": 3179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290669527495913e-05,
+      "loss": 0.7359,
+      "step": 3180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285871415725573e-05,
+      "loss": 0.7494,
+      "step": 3181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281073002927528e-05,
+      "loss": 0.7577,
+      "step": 3182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1276274290225097e-05,
+      "loss": 0.714,
+      "step": 3183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271475278741687e-05,
+      "loss": 0.7473,
+      "step": 3184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266675969600763e-05,
+      "loss": 0.7688,
+      "step": 3185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261876363925858e-05,
+      "loss": 0.8179,
+      "step": 3186
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257076462840587e-05,
+      "loss": 0.7902,
+      "step": 3187
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125227626746862e-05,
+      "loss": 0.6971,
+      "step": 3188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247475778933703e-05,
+      "loss": 0.3469,
+      "step": 3189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242674998359653e-05,
+      "loss": 0.7083,
+      "step": 3190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237873926870348e-05,
+      "loss": 0.7564,
+      "step": 3191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1233072565589743e-05,
+      "loss": 0.8234,
+      "step": 3192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228270915641852e-05,
+      "loss": 0.7302,
+      "step": 3193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1223468978150765e-05,
+      "loss": 0.7891,
+      "step": 3194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1218666754240629e-05,
+      "loss": 0.8258,
+      "step": 3195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121386424503567e-05,
+      "loss": 0.7519,
+      "step": 3196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120906145166017e-05,
+      "loss": 0.7863,
+      "step": 3197
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1204258375238486e-05,
+      "loss": 0.7593,
+      "step": 3198
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1199455016895038e-05,
+      "loss": 0.7362,
+      "step": 3199
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1194651377754311e-05,
+      "loss": 0.7714,
+      "step": 3200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1189847458940853e-05,
+      "loss": 0.6982,
+      "step": 3201
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1185043261579283e-05,
+      "loss": 0.3635,
+      "step": 3202
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1180238786794282e-05,
+      "loss": 0.7305,
+      "step": 3203
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175434035710602e-05,
+      "loss": 0.6903,
+      "step": 3204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170629009453049e-05,
+      "loss": 0.7372,
+      "step": 3205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1165823709146502e-05,
+      "loss": 0.7907,
+      "step": 3206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161018135915902e-05,
+      "loss": 0.7728,
+      "step": 3207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115621229088625e-05,
+      "loss": 0.7458,
+      "step": 3208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151406175182615e-05,
+      "loss": 0.8139,
+      "step": 3209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1146599789930128e-05,
+      "loss": 0.739,
+      "step": 3210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1141793136253987e-05,
+      "loss": 0.8698,
+      "step": 3211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1136986215279445e-05,
+      "loss": 0.3325,
+      "step": 3212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113217902813182e-05,
+      "loss": 0.8116,
+      "step": 3213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127371575936499e-05,
+      "loss": 0.7092,
+      "step": 3214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122563859818924e-05,
+      "loss": 0.8174,
+      "step": 3215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11177558809046e-05,
+      "loss": 0.8102,
+      "step": 3216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1112947640319093e-05,
+      "loss": 0.7427,
+      "step": 3217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108139139188035e-05,
+      "loss": 0.8164,
+      "step": 3218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103330378637114e-05,
+      "loss": 0.7219,
+      "step": 3219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1098521359792082e-05,
+      "loss": 0.7197,
+      "step": 3220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1093712083778748e-05,
+      "loss": 0.7281,
+      "step": 3221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108890255172298e-05,
+      "loss": 0.6689,
+      "step": 3222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1084092764750714e-05,
+      "loss": 0.7315,
+      "step": 3223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107928272398794e-05,
+      "loss": 0.8248,
+      "step": 3224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074472430560705e-05,
+      "loss": 0.6943,
+      "step": 3225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106966188559512e-05,
+      "loss": 0.7826,
+      "step": 3226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1064851090217355e-05,
+      "loss": 0.7411,
+      "step": 3227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1060040045553632e-05,
+      "loss": 0.7954,
+      "step": 3228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055228752730242e-05,
+      "loss": 0.7738,
+      "step": 3229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050417212873523e-05,
+      "loss": 0.7537,
+      "step": 3230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104560542710988e-05,
+      "loss": 0.8009,
+      "step": 3231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1040793396565766e-05,
+      "loss": 0.7631,
+      "step": 3232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1035981122367704e-05,
+      "loss": 0.7873,
+      "step": 3233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031168605642262e-05,
+      "loss": 0.6915,
+      "step": 3234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.102635584751607e-05,
+      "loss": 0.7893,
+      "step": 3235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1021542849115814e-05,
+      "loss": 0.7147,
+      "step": 3236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101672961156824e-05,
+      "loss": 0.764,
+      "step": 3237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011916136000144e-05,
+      "loss": 0.8242,
+      "step": 3238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007102423538378e-05,
+      "loss": 0.752,
+      "step": 3239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1002288475309855e-05,
+      "loss": 0.7549,
+      "step": 3240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997474292441537e-05,
+      "loss": 0.868,
+      "step": 3241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0992659876060448e-05,
+      "loss": 0.6778,
+      "step": 3242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987845227293656e-05,
+      "loss": 0.7566,
+      "step": 3243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0983030347268296e-05,
+      "loss": 0.7994,
+      "step": 3244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0978215237111546e-05,
+      "loss": 0.81,
+      "step": 3245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0973399897950647e-05,
+      "loss": 0.8237,
+      "step": 3246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0968584330912889e-05,
+      "loss": 0.8309,
+      "step": 3247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096376853712561e-05,
+      "loss": 0.8032,
+      "step": 3248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958952517716212e-05,
+      "loss": 0.754,
+      "step": 3249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0954136273812146e-05,
+      "loss": 0.6353,
+      "step": 3250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0949319806540912e-05,
+      "loss": 0.8056,
+      "step": 3251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944503117030063e-05,
+      "loss": 0.7601,
+      "step": 3252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939686206407207e-05,
+      "loss": 0.6687,
+      "step": 3253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09348690758e-05,
+      "loss": 0.7229,
+      "step": 3254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0930051726336158e-05,
+      "loss": 0.7413,
+      "step": 3255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925234159143438e-05,
+      "loss": 0.7324,
+      "step": 3256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0920416375349649e-05,
+      "loss": 0.7467,
+      "step": 3257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091559837608266e-05,
+      "loss": 0.7903,
+      "step": 3258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0910780162470378e-05,
+      "loss": 0.639,
+      "step": 3259
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0905961735640773e-05,
+      "loss": 0.8157,
+      "step": 3260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901143096721851e-05,
+      "loss": 0.7038,
+      "step": 3261
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089632424684168e-05,
+      "loss": 0.8038,
+      "step": 3262
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0891505187128371e-05,
+      "loss": 0.7538,
+      "step": 3263
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886685918710082e-05,
+      "loss": 0.7638,
+      "step": 3264
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881866442715033e-05,
+      "loss": 0.7493,
+      "step": 3265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087704676027147e-05,
+      "loss": 0.7667,
+      "step": 3266
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872226872507707e-05,
+      "loss": 0.7032,
+      "step": 3267
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0867406780552101e-05,
+      "loss": 0.7028,
+      "step": 3268
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862586485533053e-05,
+      "loss": 0.832,
+      "step": 3269
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0857765988579013e-05,
+      "loss": 0.7681,
+      "step": 3270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852945290818477e-05,
+      "loss": 0.715,
+      "step": 3271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0848124393379994e-05,
+      "loss": 0.7571,
+      "step": 3272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0843303297392156e-05,
+      "loss": 0.7087,
+      "step": 3273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0838482003983595e-05,
+      "loss": 0.7713,
+      "step": 3274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833660514283e-05,
+      "loss": 0.7153,
+      "step": 3275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828838829419101e-05,
+      "loss": 0.4006,
+      "step": 3276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0824016950520675e-05,
+      "loss": 0.7395,
+      "step": 3277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081919487871654e-05,
+      "loss": 0.7256,
+      "step": 3278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0814372615135565e-05,
+      "loss": 0.681,
+      "step": 3279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809550160906658e-05,
+      "loss": 0.8154,
+      "step": 3280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0804727517158782e-05,
+      "loss": 0.7474,
+      "step": 3281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079990468502093e-05,
+      "loss": 0.7723,
+      "step": 3282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0795081665622152e-05,
+      "loss": 0.7217,
+      "step": 3283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0790258460091533e-05,
+      "loss": 0.7526,
+      "step": 3284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785435069558208e-05,
+      "loss": 0.7885,
+      "step": 3285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780611495151347e-05,
+      "loss": 0.7101,
+      "step": 3286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0775787738000175e-05,
+      "loss": 0.7748,
+      "step": 3287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770963799233951e-05,
+      "loss": 0.7573,
+      "step": 3288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0766139679981974e-05,
+      "loss": 0.6873,
+      "step": 3289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761315381373594e-05,
+      "loss": 0.73,
+      "step": 3290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0756490904538197e-05,
+      "loss": 0.7333,
+      "step": 3291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751666250605219e-05,
+      "loss": 0.7434,
+      "step": 3292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746841420704122e-05,
+      "loss": 0.8142,
+      "step": 3293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074201641596442e-05,
+      "loss": 0.7616,
+      "step": 3294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0737191237515672e-05,
+      "loss": 0.7869,
+      "step": 3295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0732365886487468e-05,
+      "loss": 0.7607,
+      "step": 3296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072754036400944e-05,
+      "loss": 0.7479,
+      "step": 3297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722714671211265e-05,
+      "loss": 0.7419,
+      "step": 3298
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0717888809222657e-05,
+      "loss": 0.7625,
+      "step": 3299
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071306277917337e-05,
+      "loss": 0.8422,
+      "step": 3300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0708236582193199e-05,
+      "loss": 0.6888,
+      "step": 3301
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0703410219411971e-05,
+      "loss": 0.7604,
+      "step": 3302
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698583691959561e-05,
+      "loss": 0.7301,
+      "step": 3303
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0693757000965878e-05,
+      "loss": 0.7636,
+      "step": 3304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688930147560872e-05,
+      "loss": 0.7605,
+      "step": 3305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0684103132874526e-05,
+      "loss": 0.8147,
+      "step": 3306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679275958036862e-05,
+      "loss": 0.8003,
+      "step": 3307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674448624177944e-05,
+      "loss": 0.7599,
+      "step": 3308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0669621132427875e-05,
+      "loss": 0.3745,
+      "step": 3309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066479348391678e-05,
+      "loss": 0.6856,
+      "step": 3310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659965679774837e-05,
+      "loss": 0.7588,
+      "step": 3311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0655137721132256e-05,
+      "loss": 0.7786,
+      "step": 3312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0650309609119278e-05,
+      "loss": 0.7691,
+      "step": 3313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0645481344866186e-05,
+      "loss": 0.7513,
+      "step": 3314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0640652929503297e-05,
+      "loss": 0.7802,
+      "step": 3315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635824364160958e-05,
+      "loss": 0.7494,
+      "step": 3316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0630995649969561e-05,
+      "loss": 0.75,
+      "step": 3317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626166788059527e-05,
+      "loss": 0.7303,
+      "step": 3318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621337779561309e-05,
+      "loss": 0.7233,
+      "step": 3319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616508625605397e-05,
+      "loss": 0.6795,
+      "step": 3320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061167932732232e-05,
+      "loss": 0.3826,
+      "step": 3321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606849885842632e-05,
+      "loss": 0.8418,
+      "step": 3322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602020302296929e-05,
+      "loss": 0.7831,
+      "step": 3323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0597190577815832e-05,
+      "loss": 0.6888,
+      "step": 3324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592360713530002e-05,
+      "loss": 0.811,
+      "step": 3325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0587530710570126e-05,
+      "loss": 0.7008,
+      "step": 3326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0582700570066932e-05,
+      "loss": 0.7449,
+      "step": 3327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577870293151173e-05,
+      "loss": 0.6883,
+      "step": 3328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573039880953637e-05,
+      "loss": 0.7282,
+      "step": 3329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0568209334605142e-05,
+      "loss": 0.7087,
+      "step": 3330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056337865523654e-05,
+      "loss": 0.7025,
+      "step": 3331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0558547843978713e-05,
+      "loss": 0.7734,
+      "step": 3332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055371690196257e-05,
+      "loss": 0.741,
+      "step": 3333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548885830319058e-05,
+      "loss": 0.7341,
+      "step": 3334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0544054630179149e-05,
+      "loss": 0.7506,
+      "step": 3335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0539223302673846e-05,
+      "loss": 0.7095,
+      "step": 3336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0534391848934183e-05,
+      "loss": 0.749,
+      "step": 3337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529560270091221e-05,
+      "loss": 0.7423,
+      "step": 3338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0524728567276055e-05,
+      "loss": 0.6223,
+      "step": 3339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519896741619803e-05,
+      "loss": 0.7319,
+      "step": 3340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0515064794253621e-05,
+      "loss": 0.7755,
+      "step": 3341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0510232726308683e-05,
+      "loss": 0.7679,
+      "step": 3342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0505400538916193e-05,
+      "loss": 0.8632,
+      "step": 3343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0500568233207391e-05,
+      "loss": 0.7814,
+      "step": 3344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0495735810313538e-05,
+      "loss": 0.7593,
+      "step": 3345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490903271365922e-05,
+      "loss": 0.8097,
+      "step": 3346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048607061749586e-05,
+      "loss": 0.3492,
+      "step": 3347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0481237849834699e-05,
+      "loss": 0.7803,
+      "step": 3348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476404969513804e-05,
+      "loss": 0.692,
+      "step": 3349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0471571977664578e-05,
+      "loss": 0.7533,
+      "step": 3350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046673887541844e-05,
+      "loss": 0.7644,
+      "step": 3351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0461905663906842e-05,
+      "loss": 0.7466,
+      "step": 3352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457072344261253e-05,
+      "loss": 0.7805,
+      "step": 3353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045223891761318e-05,
+      "loss": 0.794,
+      "step": 3354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044740538509414e-05,
+      "loss": 0.7558,
+      "step": 3355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442571747835689e-05,
+      "loss": 0.7836,
+      "step": 3356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437738006969396e-05,
+      "loss": 0.7401,
+      "step": 3357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0432904163626864e-05,
+      "loss": 0.727,
+      "step": 3358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0428070218939711e-05,
+      "loss": 0.7454,
+      "step": 3359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423236174039586e-05,
+      "loss": 0.7561,
+      "step": 3360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0418402030058157e-05,
+      "loss": 0.7062,
+      "step": 3361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041356778812712e-05,
+      "loss": 0.7383,
+      "step": 3362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408733449378186e-05,
+      "loss": 0.3553,
+      "step": 3363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403899014943095e-05,
+      "loss": 0.7861,
+      "step": 3364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039906448595361e-05,
+      "loss": 0.7785,
+      "step": 3365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039422986354151e-05,
+      "loss": 0.7215,
+      "step": 3366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0389395148838599e-05,
+      "loss": 0.8053,
+      "step": 3367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0384560342976708e-05,
+      "loss": 0.393,
+      "step": 3368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0379725447087684e-05,
+      "loss": 0.7267,
+      "step": 3369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374890462303394e-05,
+      "loss": 0.7857,
+      "step": 3370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0370055389755724e-05,
+      "loss": 0.7428,
+      "step": 3371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0365220230576592e-05,
+      "loss": 0.7499,
+      "step": 3372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036038498589792e-05,
+      "loss": 0.7265,
+      "step": 3373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355549656851665e-05,
+      "loss": 0.3777,
+      "step": 3374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350714244569793e-05,
+      "loss": 0.6909,
+      "step": 3375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.03458787501843e-05,
+      "loss": 0.7953,
+      "step": 3376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341043174827183e-05,
+      "loss": 0.8314,
+      "step": 3377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0336207519630481e-05,
+      "loss": 0.8081,
+      "step": 3378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0331371785726234e-05,
+      "loss": 0.754,
+      "step": 3379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032653597424651e-05,
+      "loss": 0.6339,
+      "step": 3380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321700086323394e-05,
+      "loss": 0.7383,
+      "step": 3381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031686412308898e-05,
+      "loss": 0.7473,
+      "step": 3382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0312028085675393e-05,
+      "loss": 0.694,
+      "step": 3383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0307191975214766e-05,
+      "loss": 0.7449,
+      "step": 3384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302355792839254e-05,
+      "loss": 0.7528,
+      "step": 3385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0297519539681025e-05,
+      "loss": 0.7343,
+      "step": 3386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0292683216872265e-05,
+      "loss": 0.7231,
+      "step": 3387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0287846825545177e-05,
+      "loss": 0.7486,
+      "step": 3388
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283010366831984e-05,
+      "loss": 0.7324,
+      "step": 3389
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278173841864918e-05,
+      "loss": 0.7765,
+      "step": 3390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0273337251776225e-05,
+      "loss": 0.7516,
+      "step": 3391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268500597698174e-05,
+      "loss": 0.7353,
+      "step": 3392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0263663880763046e-05,
+      "loss": 0.7598,
+      "step": 3393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0258827102103137e-05,
+      "loss": 0.773,
+      "step": 3394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253990262850752e-05,
+      "loss": 0.7086,
+      "step": 3395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249153364138219e-05,
+      "loss": 0.7585,
+      "step": 3396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0244316407097872e-05,
+      "loss": 0.7408,
+      "step": 3397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0239479392862063e-05,
+      "loss": 0.7783,
+      "step": 3398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0234642322563162e-05,
+      "loss": 0.7159,
+      "step": 3399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022980519733354e-05,
+      "loss": 0.6948,
+      "step": 3400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022496801830559e-05,
+      "loss": 0.7369,
+      "step": 3401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0220130786611715e-05,
+      "loss": 0.8474,
+      "step": 3402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0215293503384335e-05,
+      "loss": 0.7606,
+      "step": 3403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0210456169755866e-05,
+      "loss": 0.8079,
+      "step": 3404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020561878685876e-05,
+      "loss": 0.7149,
+      "step": 3405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020078135582546e-05,
+      "loss": 0.6737,
+      "step": 3406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0195943877788432e-05,
+      "loss": 0.7682,
+      "step": 3407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019110635388015e-05,
+      "loss": 0.8117,
+      "step": 3408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0186268785233093e-05,
+      "loss": 0.8121,
+      "step": 3409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0181431172979761e-05,
+      "loss": 0.7875,
+      "step": 3410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0176593518252656e-05,
+      "loss": 0.682,
+      "step": 3411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0171755822184292e-05,
+      "loss": 0.3542,
+      "step": 3412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166918085907196e-05,
+      "loss": 0.6938,
+      "step": 3413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0162080310553897e-05,
+      "loss": 0.7723,
+      "step": 3414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0157242497256941e-05,
+      "loss": 0.776,
+      "step": 3415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015240464714888e-05,
+      "loss": 0.7027,
+      "step": 3416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0147566761362273e-05,
+      "loss": 0.8102,
+      "step": 3417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014272884102969e-05,
+      "loss": 0.748,
+      "step": 3418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137890887283705e-05,
+      "loss": 0.7746,
+      "step": 3419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0133052901256905e-05,
+      "loss": 0.6999,
+      "step": 3420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0128214884081886e-05,
+      "loss": 0.3514,
+      "step": 3421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123376836891239e-05,
+      "loss": 0.741,
+      "step": 3422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0118538760817574e-05,
+      "loss": 0.7527,
+      "step": 3423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0113700656993505e-05,
+      "loss": 0.7893,
+      "step": 3424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0108862526551652e-05,
+      "loss": 0.809,
+      "step": 3425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0104024370624644e-05,
+      "loss": 0.8209,
+      "step": 3426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099186190345105e-05,
+      "loss": 0.7108,
+      "step": 3427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094347986845679e-05,
+      "loss": 0.6866,
+      "step": 3428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0089509761259005e-05,
+      "loss": 0.7488,
+      "step": 3429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0084671514717735e-05,
+      "loss": 0.828,
+      "step": 3430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0079833248354519e-05,
+      "loss": 0.7655,
+      "step": 3431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074994963302017e-05,
+      "loss": 0.7324,
+      "step": 3432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0070156660692889e-05,
+      "loss": 0.7288,
+      "step": 3433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065318341659802e-05,
+      "loss": 0.7404,
+      "step": 3434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060480007335429e-05,
+      "loss": 0.7974,
+      "step": 3435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005564165885244e-05,
+      "loss": 0.746,
+      "step": 3436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050803297343512e-05,
+      "loss": 0.7577,
+      "step": 3437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045964923941326e-05,
+      "loss": 0.7518,
+      "step": 3438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004112653977857e-05,
+      "loss": 0.7263,
+      "step": 3439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003628814598792e-05,
+      "loss": 0.752,
+      "step": 3440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031449743702069e-05,
+      "loss": 0.6433,
+      "step": 3441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026611334053704e-05,
+      "loss": 0.7857,
+      "step": 3442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002177291817552e-05,
+      "loss": 0.7465,
+      "step": 3443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0016934497200205e-05,
+      "loss": 0.7114,
+      "step": 3444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0012096072260458e-05,
+      "loss": 0.7816,
+      "step": 3445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0007257644488973e-05,
+      "loss": 0.8055,
+      "step": 3446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002419215018443e-05,
+      "loss": 0.729,
+      "step": 3447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997580784981562e-06,
+      "loss": 0.7354,
+      "step": 3448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99274235551103e-06,
+      "loss": 0.8112,
+      "step": 3449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987903927739544e-06,
+      "loss": 0.7888,
+      "step": 3450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983065502799796e-06,
+      "loss": 0.6726,
+      "step": 3451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.978227081824483e-06,
+      "loss": 0.8434,
+      "step": 3452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973388665946298e-06,
+      "loss": 0.7902,
+      "step": 3453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968550256297936e-06,
+      "loss": 0.7845,
+      "step": 3454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963711854012083e-06,
+      "loss": 0.794,
+      "step": 3455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.958873460221435e-06,
+      "loss": 0.3548,
+      "step": 3456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954035076058675e-06,
+      "loss": 0.6682,
+      "step": 3457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949196702656488e-06,
+      "loss": 0.8433,
+      "step": 3458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.944358341147564e-06,
+      "loss": 0.7909,
+      "step": 3459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.939519992664576e-06,
+      "loss": 0.7428,
+      "step": 3460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934681658340198e-06,
+      "loss": 0.7355,
+      "step": 3461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.929843339307115e-06,
+      "loss": 0.7503,
+      "step": 3462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925005036697985e-06,
+      "loss": 0.7532,
+      "step": 3463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920166751645483e-06,
+      "loss": 0.7762,
+      "step": 3464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91532848528227e-06,
+      "loss": 0.7639,
+      "step": 3465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910490238740997e-06,
+      "loss": 0.7965,
+      "step": 3466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.905652013154326e-06,
+      "loss": 0.6699,
+      "step": 3467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9008138096549e-06,
+      "loss": 0.7825,
+      "step": 3468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89597562937536e-06,
+      "loss": 0.8279,
+      "step": 3469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89113747344835e-06,
+      "loss": 0.6862,
+      "step": 3470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.886299343006495e-06,
+      "loss": 0.7298,
+      "step": 3471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881461239182428e-06,
+      "loss": 0.7606,
+      "step": 3472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876623163108767e-06,
+      "loss": 0.7729,
+      "step": 3473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.871785115918117e-06,
+      "loss": 0.7622,
+      "step": 3474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.866947098743097e-06,
+      "loss": 0.3284,
+      "step": 3475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862109112716295e-06,
+      "loss": 0.7591,
+      "step": 3476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857271158970314e-06,
+      "loss": 0.677,
+      "step": 3477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85243323863773e-06,
+      "loss": 0.7837,
+      "step": 3478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847595352851122e-06,
+      "loss": 0.7333,
+      "step": 3479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842757502743062e-06,
+      "loss": 0.7601,
+      "step": 3480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837919689446104e-06,
+      "loss": 0.6811,
+      "step": 3481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833081914092809e-06,
+      "loss": 0.7806,
+      "step": 3482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828244177815713e-06,
+      "loss": 0.776,
+      "step": 3483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.823406481747346e-06,
+      "loss": 0.6617,
+      "step": 3484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.818568827020242e-06,
+      "loss": 0.761,
+      "step": 3485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81373121476691e-06,
+      "loss": 0.7382,
+      "step": 3486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808893646119852e-06,
+      "loss": 0.7364,
+      "step": 3487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.804056122211572e-06,
+      "loss": 0.6811,
+      "step": 3488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79921864417454e-06,
+      "loss": 0.6981,
+      "step": 3489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794381213141244e-06,
+      "loss": 0.758,
+      "step": 3490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789543830244139e-06,
+      "loss": 0.7272,
+      "step": 3491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78470649661567e-06,
+      "loss": 0.7705,
+      "step": 3492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.779869213388289e-06,
+      "loss": 0.7892,
+      "step": 3493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77503198169441e-06,
+      "loss": 0.7616,
+      "step": 3494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770194802666463e-06,
+      "loss": 0.7307,
+      "step": 3495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.765357677436843e-06,
+      "loss": 0.7851,
+      "step": 3496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.760520607137937e-06,
+      "loss": 0.7053,
+      "step": 3497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75568359290213e-06,
+      "loss": 0.8054,
+      "step": 3498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750846635861786e-06,
+      "loss": 0.8213,
+      "step": 3499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74600973714925e-06,
+      "loss": 0.7752,
+      "step": 3500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.741172897896868e-06,
+      "loss": 0.6533,
+      "step": 3501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.736336119236954e-06,
+      "loss": 0.7796,
+      "step": 3502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731499402301828e-06,
+      "loss": 0.7423,
+      "step": 3503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72666274822378e-06,
+      "loss": 0.741,
+      "step": 3504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721826158135085e-06,
+      "loss": 0.7966,
+      "step": 3505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71698963316802e-06,
+      "loss": 0.8036,
+      "step": 3506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71215317445482e-06,
+      "loss": 0.6692,
+      "step": 3507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.707316783127737e-06,
+      "loss": 0.7445,
+      "step": 3508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70248046031898e-06,
+      "loss": 0.752,
+      "step": 3509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697644207160747e-06,
+      "loss": 0.7375,
+      "step": 3510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692808024785238e-06,
+      "loss": 0.7593,
+      "step": 3511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.687971914324607e-06,
+      "loss": 0.8052,
+      "step": 3512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.683135876911021e-06,
+      "loss": 0.7803,
+      "step": 3513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678299913676613e-06,
+      "loss": 0.6837,
+      "step": 3514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.673464025753491e-06,
+      "loss": 0.7733,
+      "step": 3515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668628214273768e-06,
+      "loss": 0.7759,
+      "step": 3516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663792480369525e-06,
+      "loss": 0.7019,
+      "step": 3517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658956825172819e-06,
+      "loss": 0.7195,
+      "step": 3518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.654121249815707e-06,
+      "loss": 0.6762,
+      "step": 3519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649285755430208e-06,
+      "loss": 0.6835,
+      "step": 3520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644450343148338e-06,
+      "loss": 0.8102,
+      "step": 3521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.639615014102083e-06,
+      "loss": 0.7204,
+      "step": 3522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634779769423412e-06,
+      "loss": 0.8097,
+      "step": 3523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62994461024428e-06,
+      "loss": 0.7061,
+      "step": 3524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625109537696611e-06,
+      "loss": 0.7722,
+      "step": 3525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.620274552912319e-06,
+      "loss": 0.7379,
+      "step": 3526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.615439657023294e-06,
+      "loss": 0.8169,
+      "step": 3527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610604851161403e-06,
+      "loss": 0.6745,
+      "step": 3528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.605770136458494e-06,
+      "loss": 0.7263,
+      "step": 3529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.600935514046393e-06,
+      "loss": 0.7016,
+      "step": 3530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596100985056908e-06,
+      "loss": 0.8008,
+      "step": 3531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591266550621817e-06,
+      "loss": 0.7552,
+      "step": 3532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586432211872882e-06,
+      "loss": 0.7103,
+      "step": 3533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581597969941846e-06,
+      "loss": 0.7668,
+      "step": 3534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576763825960418e-06,
+      "loss": 0.7819,
+      "step": 3535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571929781060292e-06,
+      "loss": 0.7974,
+      "step": 3536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56709583637314e-06,
+      "loss": 0.7487,
+      "step": 3537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562261993030605e-06,
+      "loss": 0.7698,
+      "step": 3538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557428252164317e-06,
+      "loss": 0.7801,
+      "step": 3539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.552594614905863e-06,
+      "loss": 0.7819,
+      "step": 3540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547761082386824e-06,
+      "loss": 0.7695,
+      "step": 3541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542927655738749e-06,
+      "loss": 0.6537,
+      "step": 3542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53809433609316e-06,
+      "loss": 0.7988,
+      "step": 3543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.533261124581563e-06,
+      "loss": 0.7104,
+      "step": 3544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.528428022335425e-06,
+      "loss": 0.3763,
+      "step": 3545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523595030486196e-06,
+      "loss": 0.7456,
+      "step": 3546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.518762150165305e-06,
+      "loss": 0.7377,
+      "step": 3547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51392938250414e-06,
+      "loss": 0.7295,
+      "step": 3548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509096728634081e-06,
+      "loss": 0.7774,
+      "step": 3549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.504264189686467e-06,
+      "loss": 0.7081,
+      "step": 3550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499431766792609e-06,
+      "loss": 0.7565,
+      "step": 3551
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494599461083809e-06,
+      "loss": 0.7683,
+      "step": 3552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489767273691322e-06,
+      "loss": 0.7965,
+      "step": 3553
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48493520574638e-06,
+      "loss": 0.8086,
+      "step": 3554
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.480103258380198e-06,
+      "loss": 0.7331,
+      "step": 3555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.475271432723947e-06,
+      "loss": 0.7764,
+      "step": 3556
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470439729908782e-06,
+      "loss": 0.7689,
+      "step": 3557
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465608151065824e-06,
+      "loss": 0.8121,
+      "step": 3558
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460776697326157e-06,
+      "loss": 0.7742,
+      "step": 3559
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.455945369820856e-06,
+      "loss": 0.7991,
+      "step": 3560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451114169680943e-06,
+      "loss": 0.7356,
+      "step": 3561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.446283098037431e-06,
+      "loss": 0.7599,
+      "step": 3562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.441452156021292e-06,
+      "loss": 0.7134,
+      "step": 3563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436621344763461e-06,
+      "loss": 0.3493,
+      "step": 3564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.431790665394861e-06,
+      "loss": 0.7637,
+      "step": 3565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426960119046363e-06,
+      "loss": 0.7588,
+      "step": 3566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422129706848829e-06,
+      "loss": 0.6965,
+      "step": 3567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.417299429933071e-06,
+      "loss": 0.7339,
+      "step": 3568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.412469289429874e-06,
+      "loss": 0.7395,
+      "step": 3569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407639286470002e-06,
+      "loss": 0.7737,
+      "step": 3570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402809422184173e-06,
+      "loss": 0.7068,
+      "step": 3571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397979697703073e-06,
+      "loss": 0.7593,
+      "step": 3572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393150114157371e-06,
+      "loss": 0.7652,
+      "step": 3573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38832067267768e-06,
+      "loss": 0.8324,
+      "step": 3574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383491374394605e-06,
+      "loss": 0.3738,
+      "step": 3575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378662220438696e-06,
+      "loss": 0.6777,
+      "step": 3576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373833211940474e-06,
+      "loss": 0.7782,
+      "step": 3577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.369004350030442e-06,
+      "loss": 0.7581,
+      "step": 3578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.364175635839042e-06,
+      "loss": 0.7318,
+      "step": 3579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359347070496706e-06,
+      "loss": 0.7318,
+      "step": 3580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.354518655133817e-06,
+      "loss": 0.7453,
+      "step": 3581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.349690390880723e-06,
+      "loss": 0.6859,
+      "step": 3582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.344862278867748e-06,
+      "loss": 0.7304,
+      "step": 3583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340034320225168e-06,
+      "loss": 0.7516,
+      "step": 3584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.335206516083222e-06,
+      "loss": 0.827,
+      "step": 3585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330378867572132e-06,
+      "loss": 0.7524,
+      "step": 3586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325551375822056e-06,
+      "loss": 0.6872,
+      "step": 3587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32072404196314e-06,
+      "loss": 0.7321,
+      "step": 3588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.315896867125481e-06,
+      "loss": 0.797,
+      "step": 3589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311069852439132e-06,
+      "loss": 0.7129,
+      "step": 3590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306242999034123e-06,
+      "loss": 0.7634,
+      "step": 3591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.301416308040439e-06,
+      "loss": 0.7967,
+      "step": 3592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29658978058803e-06,
+      "loss": 0.7688,
+      "step": 3593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291763417806806e-06,
+      "loss": 0.7858,
+      "step": 3594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286937220826631e-06,
+      "loss": 0.7876,
+      "step": 3595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.282111190777345e-06,
+      "loss": 0.7426,
+      "step": 3596
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277285328788738e-06,
+      "loss": 0.7716,
+      "step": 3597
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.272459635990563e-06,
+      "loss": 0.7265,
+      "step": 3598
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.267634113512537e-06,
+      "loss": 0.7585,
+      "step": 3599
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.262808762484328e-06,
+      "loss": 0.8141,
+      "step": 3600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257983584035582e-06,
+      "loss": 0.7417,
+      "step": 3601
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253158579295883e-06,
+      "loss": 0.7593,
+      "step": 3602
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248333749394785e-06,
+      "loss": 0.7718,
+      "step": 3603
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243509095461804e-06,
+      "loss": 0.7039,
+      "step": 3604
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23868461862641e-06,
+      "loss": 0.8212,
+      "step": 3605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233860320018028e-06,
+      "loss": 0.6849,
+      "step": 3606
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229036200766056e-06,
+      "loss": 0.7249,
+      "step": 3607
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224212261999828e-06,
+      "loss": 0.7843,
+      "step": 3608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219388504848654e-06,
+      "loss": 0.7379,
+      "step": 3609
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214564930441796e-06,
+      "loss": 0.767,
+      "step": 3610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209741539908468e-06,
+      "loss": 0.6966,
+      "step": 3611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.204918334377852e-06,
+      "loss": 0.7897,
+      "step": 3612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.200095314979071e-06,
+      "loss": 0.75,
+      "step": 3613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195272482841221e-06,
+      "loss": 0.7514,
+      "step": 3614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190449839093343e-06,
+      "loss": 0.7394,
+      "step": 3615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18562738486444e-06,
+      "loss": 0.8512,
+      "step": 3616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.180805121283463e-06,
+      "loss": 0.7816,
+      "step": 3617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.175983049479328e-06,
+      "loss": 0.7411,
+      "step": 3618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171161170580902e-06,
+      "loss": 0.7253,
+      "step": 3619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166339485717002e-06,
+      "loss": 0.707,
+      "step": 3620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161517996016407e-06,
+      "loss": 0.745,
+      "step": 3621
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156696702607849e-06,
+      "loss": 0.7691,
+      "step": 3622
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.151875606620007e-06,
+      "loss": 0.3336,
+      "step": 3623
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147054709181525e-06,
+      "loss": 0.7967,
+      "step": 3624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.142234011420989e-06,
+      "loss": 0.7535,
+      "step": 3625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137413514466949e-06,
+      "loss": 0.3516,
+      "step": 3626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132593219447902e-06,
+      "loss": 0.705,
+      "step": 3627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127773127492293e-06,
+      "loss": 0.7175,
+      "step": 3628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122953239728532e-06,
+      "loss": 0.7688,
+      "step": 3629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118133557284974e-06,
+      "loss": 0.7982,
+      "step": 3630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113314081289918e-06,
+      "loss": 0.7711,
+      "step": 3631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.108494812871634e-06,
+      "loss": 0.739,
+      "step": 3632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10367575315832e-06,
+      "loss": 0.3916,
+      "step": 3633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098856903278152e-06,
+      "loss": 0.7476,
+      "step": 3634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094038264359232e-06,
+      "loss": 0.3601,
+      "step": 3635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.089219837529622e-06,
+      "loss": 0.8482,
+      "step": 3636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084401623917344e-06,
+      "loss": 0.7612,
+      "step": 3637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.079583624650355e-06,
+      "loss": 0.7628,
+      "step": 3638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074765840856565e-06,
+      "loss": 0.8165,
+      "step": 3639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069948273663845e-06,
+      "loss": 0.6613,
+      "step": 3640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065130924199998e-06,
+      "loss": 0.7661,
+      "step": 3641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060313793592796e-06,
+      "loss": 0.7348,
+      "step": 3642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055496882969942e-06,
+      "loss": 0.7382,
+      "step": 3643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050680193459091e-06,
+      "loss": 0.7265,
+      "step": 3644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.045863726187858e-06,
+      "loss": 0.7278,
+      "step": 3645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041047482283786e-06,
+      "loss": 0.7215,
+      "step": 3646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.036231462874391e-06,
+      "loss": 0.3559,
+      "step": 3647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.031415669087116e-06,
+      "loss": 0.773,
+      "step": 3648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.026600102049355e-06,
+      "loss": 0.7684,
+      "step": 3649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.021784762888456e-06,
+      "loss": 0.6918,
+      "step": 3650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016969652731706e-06,
+      "loss": 0.7906,
+      "step": 3651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012154772706346e-06,
+      "loss": 0.7973,
+      "step": 3652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007340123939557e-06,
+      "loss": 0.7081,
+      "step": 3653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002525707558463e-06,
+      "loss": 0.7984,
+      "step": 3654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99771152469015e-06,
+      "loss": 0.7634,
+      "step": 3655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992897576461627e-06,
+      "loss": 0.7491,
+      "step": 3656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988083863999859e-06,
+      "loss": 0.3697,
+      "step": 3657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.983270388431765e-06,
+      "loss": 0.744,
+      "step": 3658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978457150884187e-06,
+      "loss": 0.69,
+      "step": 3659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973644152483934e-06,
+      "loss": 0.7277,
+      "step": 3660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.968831394357743e-06,
+      "loss": 0.727,
+      "step": 3661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964018877632299e-06,
+      "loss": 0.7807,
+      "step": 3662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959206603434236e-06,
+      "loss": 0.7311,
+      "step": 3663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954394572890121e-06,
+      "loss": 0.7389,
+      "step": 3664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949582787126478e-06,
+      "loss": 0.7047,
+      "step": 3665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944771247269763e-06,
+      "loss": 0.7733,
+      "step": 3666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939959954446368e-06,
+      "loss": 0.6838,
+      "step": 3667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935148909782649e-06,
+      "loss": 0.7176,
+      "step": 3668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93033811440488e-06,
+      "loss": 0.7464,
+      "step": 3669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925527569439297e-06,
+      "loss": 0.746,
+      "step": 3670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.920717276012065e-06,
+      "loss": 0.7582,
+      "step": 3671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.915907235249287e-06,
+      "loss": 0.7237,
+      "step": 3672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911097448277022e-06,
+      "loss": 0.7641,
+      "step": 3673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.906287916221259e-06,
+      "loss": 0.7582,
+      "step": 3674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901478640207921e-06,
+      "loss": 0.7767,
+      "step": 3675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896669621362888e-06,
+      "loss": 0.7943,
+      "step": 3676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891860860811967e-06,
+      "loss": 0.7437,
+      "step": 3677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88705235968091e-06,
+      "loss": 0.713,
+      "step": 3678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882244119095405e-06,
+      "loss": 0.6991,
+      "step": 3679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877436140181079e-06,
+      "loss": 0.7937,
+      "step": 3680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872628424063505e-06,
+      "loss": 0.756,
+      "step": 3681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867820971868183e-06,
+      "loss": 0.7554,
+      "step": 3682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.863013784720559e-06,
+      "loss": 0.3709,
+      "step": 3683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.858206863746018e-06,
+      "loss": 0.7188,
+      "step": 3684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853400210069874e-06,
+      "loss": 0.3668,
+      "step": 3685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848593824817388e-06,
+      "loss": 0.7561,
+      "step": 3686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843787709113757e-06,
+      "loss": 0.7245,
+      "step": 3687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838981864084101e-06,
+      "loss": 0.7805,
+      "step": 3688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.834176290853501e-06,
+      "loss": 0.7615,
+      "step": 3689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829370990546953e-06,
+      "loss": 0.7358,
+      "step": 3690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8245659642894e-06,
+      "loss": 0.7417,
+      "step": 3691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.819761213205721e-06,
+      "loss": 0.7314,
+      "step": 3692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.814956738420722e-06,
+      "loss": 0.7888,
+      "step": 3693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810152541059152e-06,
+      "loss": 0.6555,
+      "step": 3694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805348622245694e-06,
+      "loss": 0.7158,
+      "step": 3695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800544983104965e-06,
+      "loss": 0.7467,
+      "step": 3696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.795741624761515e-06,
+      "loss": 0.7167,
+      "step": 3697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790938548339831e-06,
+      "loss": 0.7573,
+      "step": 3698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.786135754964333e-06,
+      "loss": 0.7875,
+      "step": 3699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.781333245759373e-06,
+      "loss": 0.7364,
+      "step": 3700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77653102184924e-06,
+      "loss": 0.7338,
+      "step": 3701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771729084358151e-06,
+      "loss": 0.8023,
+      "step": 3702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.766927434410259e-06,
+      "loss": 0.746,
+      "step": 3703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.762126073129655e-06,
+      "loss": 0.7337,
+      "step": 3704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75732500164035e-06,
+      "loss": 0.6406,
+      "step": 3705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752524221066299e-06,
+      "loss": 0.7375,
+      "step": 3706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747723732531386e-06,
+      "loss": 0.7308,
+      "step": 3707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.742923537159415e-06,
+      "loss": 0.7506,
+      "step": 3708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738123636074143e-06,
+      "loss": 0.7052,
+      "step": 3709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733324030399242e-06,
+      "loss": 0.8104,
+      "step": 3710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728524721258314e-06,
+      "loss": 0.7098,
+      "step": 3711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723725709774905e-06,
+      "loss": 0.6943,
+      "step": 3712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718926997072474e-06,
+      "loss": 0.7372,
+      "step": 3713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71412858427443e-06,
+      "loss": 0.7441,
+      "step": 3714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709330472504094e-06,
+      "loss": 0.6897,
+      "step": 3715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.704532662884719e-06,
+      "loss": 0.6491,
+      "step": 3716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699735156539503e-06,
+      "loss": 0.7561,
+      "step": 3717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69493795459155e-06,
+      "loss": 0.7003,
+      "step": 3718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.690141058163918e-06,
+      "loss": 0.7585,
+      "step": 3719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68534446837957e-06,
+      "loss": 0.7773,
+      "step": 3720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680548186361405e-06,
+      "loss": 0.7586,
+      "step": 3721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67575221323226e-06,
+      "loss": 0.8386,
+      "step": 3722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670956550114889e-06,
+      "loss": 0.7772,
+      "step": 3723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.666161198131969e-06,
+      "loss": 0.7495,
+      "step": 3724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.661366158406124e-06,
+      "loss": 0.3986,
+      "step": 3725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65657143205988e-06,
+      "loss": 0.7059,
+      "step": 3726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.651777020215713e-06,
+      "loss": 0.7766,
+      "step": 3727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.646982923996009e-06,
+      "loss": 0.7692,
+      "step": 3728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.642189144523079e-06,
+      "loss": 0.755,
+      "step": 3729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637395682919178e-06,
+      "loss": 0.7765,
+      "step": 3730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.632602540306465e-06,
+      "loss": 0.7012,
+      "step": 3731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627809717807043e-06,
+      "loss": 0.7467,
+      "step": 3732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623017216542926e-06,
+      "loss": 0.7902,
+      "step": 3733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618225037636054e-06,
+      "loss": 0.7666,
+      "step": 3734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613433182208305e-06,
+      "loss": 0.7744,
+      "step": 3735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60864165138146e-06,
+      "loss": 0.7101,
+      "step": 3736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.603850446277249e-06,
+      "loss": 0.8314,
+      "step": 3737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599059568017306e-06,
+      "loss": 0.7526,
+      "step": 3738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59426901772319e-06,
+      "loss": 0.8072,
+      "step": 3739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589478796516398e-06,
+      "loss": 0.7134,
+      "step": 3740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.584688905518336e-06,
+      "loss": 0.7125,
+      "step": 3741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579899345850332e-06,
+      "loss": 0.7323,
+      "step": 3742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57511011863365e-06,
+      "loss": 0.7342,
+      "step": 3743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570321224989458e-06,
+      "loss": 0.7274,
+      "step": 3744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565532666038865e-06,
+      "loss": 0.7896,
+      "step": 3745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560744442902887e-06,
+      "loss": 0.7216,
+      "step": 3746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.555956556702464e-06,
+      "loss": 0.7383,
+      "step": 3747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.551169008558467e-06,
+      "loss": 0.8143,
+      "step": 3748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.546381799591672e-06,
+      "loss": 0.6975,
+      "step": 3749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54159493092279e-06,
+      "loss": 0.7643,
+      "step": 3750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.536808403672448e-06,
+      "loss": 0.7281,
+      "step": 3751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53202221896118e-06,
+      "loss": 0.7011,
+      "step": 3752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527236377909467e-06,
+      "loss": 0.7467,
+      "step": 3753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522450881637679e-06,
+      "loss": 0.7134,
+      "step": 3754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.517665731266133e-06,
+      "loss": 0.724,
+      "step": 3755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.512880927915046e-06,
+      "loss": 0.751,
+      "step": 3756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.508096472704554e-06,
+      "loss": 0.6868,
+      "step": 3757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503312366754727e-06,
+      "loss": 0.7272,
+      "step": 3758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49852861118554e-06,
+      "loss": 0.6675,
+      "step": 3759
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.493745207116889e-06,
+      "loss": 0.7222,
+      "step": 3760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48896215566859e-06,
+      "loss": 0.722,
+      "step": 3761
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.484179457960369e-06,
+      "loss": 0.7922,
+      "step": 3762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.479397115111883e-06,
+      "loss": 0.7864,
+      "step": 3763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474615128242695e-06,
+      "loss": 0.381,
+      "step": 3764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.469833498472283e-06,
+      "loss": 0.7986,
+      "step": 3765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465052226920054e-06,
+      "loss": 0.6547,
+      "step": 3766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.460271314705315e-06,
+      "loss": 0.7976,
+      "step": 3767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.455490762947302e-06,
+      "loss": 0.7926,
+      "step": 3768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450710572765162e-06,
+      "loss": 0.7961,
+      "step": 3769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445930745277953e-06,
+      "loss": 0.7609,
+      "step": 3770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441151281604652e-06,
+      "loss": 0.6924,
+      "step": 3771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436372182864157e-06,
+      "loss": 0.7312,
+      "step": 3772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.431593450175267e-06,
+      "loss": 0.8342,
+      "step": 3773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426815084656706e-06,
+      "loss": 0.7361,
+      "step": 3774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42203708742711e-06,
+      "loss": 0.7492,
+      "step": 3775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.417259459605023e-06,
+      "loss": 0.6832,
+      "step": 3776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412482202308915e-06,
+      "loss": 0.801,
+      "step": 3777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.407705316657153e-06,
+      "loss": 0.7143,
+      "step": 3778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40292880376803e-06,
+      "loss": 0.7136,
+      "step": 3779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398152664759745e-06,
+      "loss": 0.3649,
+      "step": 3780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.393376900750414e-06,
+      "loss": 0.7159,
+      "step": 3781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.388601512858059e-06,
+      "loss": 0.6957,
+      "step": 3782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383826502200619e-06,
+      "loss": 0.7641,
+      "step": 3783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.379051869895947e-06,
+      "loss": 0.809,
+      "step": 3784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374277617061795e-06,
+      "loss": 0.7618,
+      "step": 3785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369503744815847e-06,
+      "loss": 0.7746,
+      "step": 3786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364730254275676e-06,
+      "loss": 0.7789,
+      "step": 3787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359957146558779e-06,
+      "loss": 0.6821,
+      "step": 3788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355184422782562e-06,
+      "loss": 0.7025,
+      "step": 3789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350412084064334e-06,
+      "loss": 0.7138,
+      "step": 3790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345640131521323e-06,
+      "loss": 0.7464,
+      "step": 3791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340868566270662e-06,
+      "loss": 0.688,
+      "step": 3792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33609738942939e-06,
+      "loss": 0.7394,
+      "step": 3793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.331326602114464e-06,
+      "loss": 0.6873,
+      "step": 3794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.326556205442744e-06,
+      "loss": 0.7477,
+      "step": 3795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.321786200530992e-06,
+      "loss": 0.8215,
+      "step": 3796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.317016588495895e-06,
+      "loss": 0.7555,
+      "step": 3797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312247370454032e-06,
+      "loss": 0.6879,
+      "step": 3798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.307478547521902e-06,
+      "loss": 0.7304,
+      "step": 3799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302710120815902e-06,
+      "loss": 0.7402,
+      "step": 3800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297942091452335e-06,
+      "loss": 0.8278,
+      "step": 3801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293174460547429e-06,
+      "loss": 0.717,
+      "step": 3802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.288407229217291e-06,
+      "loss": 0.3654,
+      "step": 3803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283640398577963e-06,
+      "loss": 0.7779,
+      "step": 3804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.27887396974537e-06,
+      "loss": 0.8173,
+      "step": 3805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274107943835354e-06,
+      "loss": 0.7049,
+      "step": 3806
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269342321963665e-06,
+      "loss": 0.7537,
+      "step": 3807
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26457710524595e-06,
+      "loss": 0.8015,
+      "step": 3808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.259812294797764e-06,
+      "loss": 0.7534,
+      "step": 3809
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.255047891734576e-06,
+      "loss": 0.7589,
+      "step": 3810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250283897171743e-06,
+      "loss": 0.6942,
+      "step": 3811
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245520312224544e-06,
+      "loss": 0.6841,
+      "step": 3812
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240757138008149e-06,
+      "loss": 0.7195,
+      "step": 3813
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235994375637632e-06,
+      "loss": 0.7492,
+      "step": 3814
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231232026227985e-06,
+      "loss": 0.6986,
+      "step": 3815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226470090894084e-06,
+      "loss": 0.6854,
+      "step": 3816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221708570750725e-06,
+      "loss": 0.7437,
+      "step": 3817
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216947466912597e-06,
+      "loss": 0.8376,
+      "step": 3818
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.212186780494288e-06,
+      "loss": 0.7181,
+      "step": 3819
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2074265126103e-06,
+      "loss": 0.7505,
+      "step": 3820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202666664375028e-06,
+      "loss": 0.791,
+      "step": 3821
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197907236902776e-06,
+      "loss": 0.7885,
+      "step": 3822
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193148231307743e-06,
+      "loss": 0.737,
+      "step": 3823
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188389648704024e-06,
+      "loss": 0.7191,
+      "step": 3824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183631490205636e-06,
+      "loss": 0.7567,
+      "step": 3825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178873756926477e-06,
+      "loss": 0.7683,
+      "step": 3826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174116449980347e-06,
+      "loss": 0.7762,
+      "step": 3827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169359570480957e-06,
+      "loss": 0.7035,
+      "step": 3828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164603119541907e-06,
+      "loss": 0.3886,
+      "step": 3829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159847098276708e-06,
+      "loss": 0.7088,
+      "step": 3830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.155091507798759e-06,
+      "loss": 0.7274,
+      "step": 3831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150336349221361e-06,
+      "loss": 0.7504,
+      "step": 3832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.145581623657723e-06,
+      "loss": 0.7489,
+      "step": 3833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140827332220936e-06,
+      "loss": 0.3669,
+      "step": 3834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13607347602401e-06,
+      "loss": 0.7115,
+      "step": 3835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131320056179836e-06,
+      "loss": 0.8029,
+      "step": 3836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126567073801204e-06,
+      "loss": 0.7475,
+      "step": 3837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121814530000816e-06,
+      "loss": 0.7802,
+      "step": 3838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.117062425891254e-06,
+      "loss": 0.7496,
+      "step": 3839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112310762585012e-06,
+      "loss": 0.3662,
+      "step": 3840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10755954119447e-06,
+      "loss": 0.325,
+      "step": 3841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102808762831902e-06,
+      "loss": 0.7588,
+      "step": 3842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098058428609498e-06,
+      "loss": 0.7647,
+      "step": 3843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09330853963932e-06,
+      "loss": 0.7732,
+      "step": 3844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088559097033337e-06,
+      "loss": 0.6826,
+      "step": 3845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083810101903418e-06,
+      "loss": 0.8424,
+      "step": 3846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.079061555361318e-06,
+      "loss": 0.7886,
+      "step": 3847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07431345851869e-06,
+      "loss": 0.7202,
+      "step": 3848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.069565812487089e-06,
+      "loss": 0.7127,
+      "step": 3849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064818618377947e-06,
+      "loss": 0.6913,
+      "step": 3850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060071877302615e-06,
+      "loss": 0.7311,
+      "step": 3851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055325590372314e-06,
+      "loss": 0.6802,
+      "step": 3852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050579758698173e-06,
+      "loss": 0.7904,
+      "step": 3853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045834383391211e-06,
+      "loss": 0.7448,
+      "step": 3854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041089465562339e-06,
+      "loss": 0.6317,
+      "step": 3855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036345006322358e-06,
+      "loss": 0.7821,
+      "step": 3856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.031601006781973e-06,
+      "loss": 0.7763,
+      "step": 3857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026857468051765e-06,
+      "loss": 0.7175,
+      "step": 3858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02211439124222e-06,
+      "loss": 0.7369,
+      "step": 3859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01737177746371e-06,
+      "loss": 0.7842,
+      "step": 3860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012629627826504e-06,
+      "loss": 0.3644,
+      "step": 3861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007887943440753e-06,
+      "loss": 0.7428,
+      "step": 3862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.003146725416506e-06,
+      "loss": 0.768,
+      "step": 3863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.998405974863704e-06,
+      "loss": 0.7682,
+      "step": 3864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993665692892172e-06,
+      "loss": 0.7161,
+      "step": 3865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.988925880611635e-06,
+      "loss": 0.3693,
+      "step": 3866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984186539131696e-06,
+      "loss": 0.7896,
+      "step": 3867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.979447669561858e-06,
+      "loss": 0.7177,
+      "step": 3868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974709273011509e-06,
+      "loss": 0.7097,
+      "step": 3869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.969971350589922e-06,
+      "loss": 0.7659,
+      "step": 3870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.965233903406273e-06,
+      "loss": 0.7591,
+      "step": 3871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96049693256961e-06,
+      "loss": 0.7317,
+      "step": 3872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.955760439188878e-06,
+      "loss": 0.7819,
+      "step": 3873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951024424372915e-06,
+      "loss": 0.7491,
+      "step": 3874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946288889230434e-06,
+      "loss": 0.7932,
+      "step": 3875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941553834870046e-06,
+      "loss": 0.7405,
+      "step": 3876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93681926240025e-06,
+      "loss": 0.3748,
+      "step": 3877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93208517292942e-06,
+      "loss": 0.738,
+      "step": 3878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927351567565835e-06,
+      "loss": 0.7165,
+      "step": 3879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922618447417646e-06,
+      "loss": 0.7677,
+      "step": 3880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917885813592894e-06,
+      "loss": 0.7243,
+      "step": 3881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913153667199513e-06,
+      "loss": 0.7344,
+      "step": 3882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908422009345312e-06,
+      "loss": 0.694,
+      "step": 3883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903690841137997e-06,
+      "loss": 0.6905,
+      "step": 3884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89896016368515e-06,
+      "loss": 0.6587,
+      "step": 3885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89422997809424e-06,
+      "loss": 0.7022,
+      "step": 3886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88950028547263e-06,
+      "loss": 0.7112,
+      "step": 3887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88477108692755e-06,
+      "loss": 0.7093,
+      "step": 3888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880042383566135e-06,
+      "loss": 0.7404,
+      "step": 3889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.875314176495388e-06,
+      "loss": 0.7156,
+      "step": 3890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.870586466822196e-06,
+      "loss": 0.3436,
+      "step": 3891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865859255653347e-06,
+      "loss": 0.7696,
+      "step": 3892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.861132544095494e-06,
+      "loss": 0.3762,
+      "step": 3893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856406333255175e-06,
+      "loss": 0.7362,
+      "step": 3894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.851680624238825e-06,
+      "loss": 0.7053,
+      "step": 3895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846955418152742e-06,
+      "loss": 0.7767,
+      "step": 3896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.842230716103124e-06,
+      "loss": 0.8035,
+      "step": 3897
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.837506519196039e-06,
+      "loss": 0.7078,
+      "step": 3898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832782828537437e-06,
+      "loss": 0.7068,
+      "step": 3899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82805964523316e-06,
+      "loss": 0.8305,
+      "step": 3900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823336970388917e-06,
+      "loss": 0.7678,
+      "step": 3901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818614805110315e-06,
+      "loss": 0.7313,
+      "step": 3902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.813893150502826e-06,
+      "loss": 0.7475,
+      "step": 3903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.809172007671804e-06,
+      "loss": 0.793,
+      "step": 3904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804451377722498e-06,
+      "loss": 0.3515,
+      "step": 3905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799731261760017e-06,
+      "loss": 0.7174,
+      "step": 3906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795011660889367e-06,
+      "loss": 0.806,
+      "step": 3907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790292576215423e-06,
+      "loss": 0.783,
+      "step": 3908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.785574008842937e-06,
+      "loss": 0.7375,
+      "step": 3909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780855959876552e-06,
+      "loss": 0.7192,
+      "step": 3910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77613843042078e-06,
+      "loss": 0.7514,
+      "step": 3911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77142142158001e-06,
+      "loss": 0.7339,
+      "step": 3912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766704934458518e-06,
+      "loss": 0.6938,
+      "step": 3913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761988970160448e-06,
+      "loss": 0.7172,
+      "step": 3914
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.757273529789832e-06,
+      "loss": 0.7941,
+      "step": 3915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.752558614450573e-06,
+      "loss": 0.7543,
+      "step": 3916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747844225246443e-06,
+      "loss": 0.7746,
+      "step": 3917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74313036328111e-06,
+      "loss": 0.7341,
+      "step": 3918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.738417029658102e-06,
+      "loss": 0.8088,
+      "step": 3919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733704225480834e-06,
+      "loss": 0.7168,
+      "step": 3920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728991951852592e-06,
+      "loss": 0.7405,
+      "step": 3921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.724280209876533e-06,
+      "loss": 0.776,
+      "step": 3922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719569000655702e-06,
+      "loss": 0.7263,
+      "step": 3923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714858325293005e-06,
+      "loss": 0.7561,
+      "step": 3924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710148184891235e-06,
+      "loss": 0.7665,
+      "step": 3925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705438580553056e-06,
+      "loss": 0.7882,
+      "step": 3926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700729513381e-06,
+      "loss": 0.7542,
+      "step": 3927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.696020984477486e-06,
+      "loss": 0.7545,
+      "step": 3928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.691312994944795e-06,
+      "loss": 0.7249,
+      "step": 3929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686605545885083e-06,
+      "loss": 0.77,
+      "step": 3930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681898638400392e-06,
+      "loss": 0.8089,
+      "step": 3931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677192273592621e-06,
+      "loss": 0.7339,
+      "step": 3932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67248645256355e-06,
+      "loss": 0.7028,
+      "step": 3933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667781176414836e-06,
+      "loss": 0.768,
+      "step": 3934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.663076446247996e-06,
+      "loss": 0.7783,
+      "step": 3935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.65837226316443e-06,
+      "loss": 0.7376,
+      "step": 3936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653668628265406e-06,
+      "loss": 0.6694,
+      "step": 3937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648965542652064e-06,
+      "loss": 0.7943,
+      "step": 3938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.644263007425417e-06,
+      "loss": 0.6685,
+      "step": 3939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.639561023686343e-06,
+      "loss": 0.8086,
+      "step": 3940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634859592535601e-06,
+      "loss": 0.8036,
+      "step": 3941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630158715073813e-06,
+      "loss": 0.763,
+      "step": 3942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6254583924014715e-06,
+      "loss": 0.7818,
+      "step": 3943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6207586256189415e-06,
+      "loss": 0.6384,
+      "step": 3944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.616059415826459e-06,
+      "loss": 0.7104,
+      "step": 3945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611360764124129e-06,
+      "loss": 0.7336,
+      "step": 3946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606662671611922e-06,
+      "loss": 0.7281,
+      "step": 3947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601965139389682e-06,
+      "loss": 0.7876,
+      "step": 3948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597268168557121e-06,
+      "loss": 0.7634,
+      "step": 3949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592571760213816e-06,
+      "loss": 0.7165,
+      "step": 3950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5878759154592195e-06,
+      "loss": 0.7792,
+      "step": 3951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.583180635392645e-06,
+      "loss": 0.7502,
+      "step": 3952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.578485921113277e-06,
+      "loss": 0.8182,
+      "step": 3953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573791773720169e-06,
+      "loss": 0.8069,
+      "step": 3954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.569098194312234e-06,
+      "loss": 0.7344,
+      "step": 3955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.564405183988268e-06,
+      "loss": 0.712,
+      "step": 3956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559712743846916e-06,
+      "loss": 0.7116,
+      "step": 3957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.555020874986696e-06,
+      "loss": 0.4126,
+      "step": 3958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550329578506001e-06,
+      "loss": 0.731,
+      "step": 3959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545638855503078e-06,
+      "loss": 0.7231,
+      "step": 3960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54094870707604e-06,
+      "loss": 0.6957,
+      "step": 3961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53625913432288e-06,
+      "loss": 0.7405,
+      "step": 3962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.531570138341434e-06,
+      "loss": 0.768,
+      "step": 3963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.526881720229426e-06,
+      "loss": 0.7847,
+      "step": 3964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52219388108443e-06,
+      "loss": 0.7375,
+      "step": 3965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.517506622003882e-06,
+      "loss": 0.3564,
+      "step": 3966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.512819944085098e-06,
+      "loss": 0.7205,
+      "step": 3967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508133848425239e-06,
+      "loss": 0.726,
+      "step": 3968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5034483361213485e-06,
+      "loss": 0.7527,
+      "step": 3969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498763408270317e-06,
+      "loss": 0.791,
+      "step": 3970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494079065968906e-06,
+      "loss": 0.7431,
+      "step": 3971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489395310313743e-06,
+      "loss": 0.7192,
+      "step": 3972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484712142401307e-06,
+      "loss": 0.7902,
+      "step": 3973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480029563327955e-06,
+      "loss": 0.3641,
+      "step": 3974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.475347574189895e-06,
+      "loss": 0.346,
+      "step": 3975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470666176083193e-06,
+      "loss": 0.6796,
+      "step": 3976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465985370103792e-06,
+      "loss": 0.7487,
+      "step": 3977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.461305157347486e-06,
+      "loss": 0.7165,
+      "step": 3978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.456625538909925e-06,
+      "loss": 0.7134,
+      "step": 3979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.451946515886635e-06,
+      "loss": 0.7867,
+      "step": 3980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447268089372988e-06,
+      "loss": 0.6749,
+      "step": 3981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4425902604642295e-06,
+      "loss": 0.7172,
+      "step": 3982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4379130302554546e-06,
+      "loss": 0.7538,
+      "step": 3983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.433236399841617e-06,
+      "loss": 0.7743,
+      "step": 3984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428560370317542e-06,
+      "loss": 0.6735,
+      "step": 3985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4238849427779015e-06,
+      "loss": 0.6573,
+      "step": 3986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419210118317239e-06,
+      "loss": 0.7328,
+      "step": 3987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4145358980299455e-06,
+      "loss": 0.8357,
+      "step": 3988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409862283010271e-06,
+      "loss": 0.7586,
+      "step": 3989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.405189274352336e-06,
+      "loss": 0.8009,
+      "step": 3990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.400516873150102e-06,
+      "loss": 0.7466,
+      "step": 3991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.395845080497407e-06,
+      "loss": 0.6793,
+      "step": 3992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.391173897487931e-06,
+      "loss": 0.3324,
+      "step": 3993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386503325215212e-06,
+      "loss": 0.7754,
+      "step": 3994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.381833364772659e-06,
+      "loss": 0.7708,
+      "step": 3995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.377164017253527e-06,
+      "loss": 0.7529,
+      "step": 3996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37249528375092e-06,
+      "loss": 0.7832,
+      "step": 3997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.367827165357819e-06,
+      "loss": 0.7926,
+      "step": 3998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363159663167041e-06,
+      "loss": 0.385,
+      "step": 3999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358492778271276e-06,
+      "loss": 0.3591,
+      "step": 4000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.353826511763055e-06,
+      "loss": 0.7876,
+      "step": 4001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.349160864734766e-06,
+      "loss": 0.7301,
+      "step": 4002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344495838278665e-06,
+      "loss": 0.6825,
+      "step": 4003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.339831433486844e-06,
+      "loss": 0.8335,
+      "step": 4004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.335167651451268e-06,
+      "loss": 0.7666,
+      "step": 4005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.330504493263744e-06,
+      "loss": 0.6956,
+      "step": 4006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.325841960015932e-06,
+      "loss": 0.6823,
+      "step": 4007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.321180052799356e-06,
+      "loss": 0.7666,
+      "step": 4008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.316518772705383e-06,
+      "loss": 0.7801,
+      "step": 4009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.31185812082524e-06,
+      "loss": 0.7726,
+      "step": 4010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.307198098250006e-06,
+      "loss": 0.7558,
+      "step": 4011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3025387060706065e-06,
+      "loss": 0.7451,
+      "step": 4012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297879945377826e-06,
+      "loss": 0.7101,
+      "step": 4013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.293221817262304e-06,
+      "loss": 0.7063,
+      "step": 4014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288564322814518e-06,
+      "loss": 0.7585,
+      "step": 4015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.283907463124814e-06,
+      "loss": 0.7222,
+      "step": 4016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.279251239283378e-06,
+      "loss": 0.7208,
+      "step": 4017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274595652380252e-06,
+      "loss": 0.7383,
+      "step": 4018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26994070350533e-06,
+      "loss": 0.7568,
+      "step": 4019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26528639374835e-06,
+      "loss": 0.7167,
+      "step": 4020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.260632724198906e-06,
+      "loss": 0.736,
+      "step": 4021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2559796959464435e-06,
+      "loss": 0.6613,
+      "step": 4022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2513273100802535e-06,
+      "loss": 0.7025,
+      "step": 4023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2466755676894785e-06,
+      "loss": 0.7903,
+      "step": 4024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24202446986311e-06,
+      "loss": 0.7964,
+      "step": 4025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2373740176899876e-06,
+      "loss": 0.6904,
+      "step": 4026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2327242122588045e-06,
+      "loss": 0.7655,
+      "step": 4027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228075054658096e-06,
+      "loss": 0.6964,
+      "step": 4028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223426545976249e-06,
+      "loss": 0.7829,
+      "step": 4029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218778687301499e-06,
+      "loss": 0.7653,
+      "step": 4030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214131479721931e-06,
+      "loss": 0.6886,
+      "step": 4031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.209484924325469e-06,
+      "loss": 0.7929,
+      "step": 4032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204839022199895e-06,
+      "loss": 0.69,
+      "step": 4033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200193774432835e-06,
+      "loss": 0.8072,
+      "step": 4034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195549182111753e-06,
+      "loss": 0.7698,
+      "step": 4035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1909052463239755e-06,
+      "loss": 0.7159,
+      "step": 4036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186261968156665e-06,
+      "loss": 0.6977,
+      "step": 4037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181619348696822e-06,
+      "loss": 0.8163,
+      "step": 4038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.176977389031317e-06,
+      "loss": 0.6896,
+      "step": 4039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17233609024684e-06,
+      "loss": 0.6963,
+      "step": 4040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.167695453429946e-06,
+      "loss": 0.7942,
+      "step": 4041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.163055479667025e-06,
+      "loss": 0.6849,
+      "step": 4042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1584161700443065e-06,
+      "loss": 0.6882,
+      "step": 4043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1537775256478825e-06,
+      "loss": 0.7606,
+      "step": 4044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.149139547563673e-06,
+      "loss": 0.7366,
+      "step": 4045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144502236877442e-06,
+      "loss": 0.7041,
+      "step": 4046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139865594674814e-06,
+      "loss": 0.7417,
+      "step": 4047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135229622041236e-06,
+      "loss": 0.761,
+      "step": 4048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130594320062016e-06,
+      "loss": 0.8222,
+      "step": 4049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1259596898222935e-06,
+      "loss": 0.679,
+      "step": 4050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.12132573240705e-06,
+      "loss": 0.7354,
+      "step": 4051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.116692448901122e-06,
+      "loss": 0.3465,
+      "step": 4052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112059840389169e-06,
+      "loss": 0.7673,
+      "step": 4053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107427907955716e-06,
+      "loss": 0.7749,
+      "step": 4054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102796652685111e-06,
+      "loss": 0.8533,
+      "step": 4055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.098166075661545e-06,
+      "loss": 0.7557,
+      "step": 4056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093536177969065e-06,
+      "loss": 0.3397,
+      "step": 4057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088906960691537e-06,
+      "loss": 0.7745,
+      "step": 4058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.084278424912691e-06,
+      "loss": 0.757,
+      "step": 4059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.079650571716081e-06,
+      "loss": 0.7076,
+      "step": 4060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.075023402185103e-06,
+      "loss": 0.7406,
+      "step": 4061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070396917403002e-06,
+      "loss": 0.7488,
+      "step": 4062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065771118452856e-06,
+      "loss": 0.7908,
+      "step": 4063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061146006417574e-06,
+      "loss": 0.7009,
+      "step": 4064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056521582379927e-06,
+      "loss": 0.8355,
+      "step": 4065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0518978474225e-06,
+      "loss": 0.6925,
+      "step": 4066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047274802627738e-06,
+      "loss": 0.7575,
+      "step": 4067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042652449077909e-06,
+      "loss": 0.7295,
+      "step": 4068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038030787855123e-06,
+      "loss": 0.7726,
+      "step": 4069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033409820041335e-06,
+      "loss": 0.7818,
+      "step": 4070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.028789546718327e-06,
+      "loss": 0.7284,
+      "step": 4071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0241699689677274e-06,
+      "loss": 0.727,
+      "step": 4072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019551087870998e-06,
+      "loss": 0.7888,
+      "step": 4073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014932904509433e-06,
+      "loss": 0.7648,
+      "step": 4074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.010315419964174e-06,
+      "loss": 0.7435,
+      "step": 4075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.005698635316187e-06,
+      "loss": 0.7634,
+      "step": 4076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0010825516462875e-06,
+      "loss": 0.747,
+      "step": 4077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996467170035114e-06,
+      "loss": 0.7424,
+      "step": 4078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991852491563143e-06,
+      "loss": 0.6531,
+      "step": 4079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.987238517310697e-06,
+      "loss": 0.6773,
+      "step": 4080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982625248357924e-06,
+      "loss": 0.7129,
+      "step": 4081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9780126857848015e-06,
+      "loss": 0.7366,
+      "step": 4082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97340083067116e-06,
+      "loss": 0.7127,
+      "step": 4083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968789684096644e-06,
+      "loss": 0.7393,
+      "step": 4084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9641792471407506e-06,
+      "loss": 0.7671,
+      "step": 4085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959569520882799e-06,
+      "loss": 0.6856,
+      "step": 4086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.954960506401936e-06,
+      "loss": 0.3508,
+      "step": 4087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950352204777162e-06,
+      "loss": 0.712,
+      "step": 4088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9457446170872925e-06,
+      "loss": 0.7698,
+      "step": 4089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.941137744410987e-06,
+      "loss": 0.7448,
+      "step": 4090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.936531587826731e-06,
+      "loss": 0.7752,
+      "step": 4091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9319261484128405e-06,
+      "loss": 0.7705,
+      "step": 4092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9273214272474756e-06,
+      "loss": 0.7624,
+      "step": 4093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922717425408614e-06,
+      "loss": 0.6437,
+      "step": 4094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.918114143974072e-06,
+      "loss": 0.7243,
+      "step": 4095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9135115840215e-06,
+      "loss": 0.7501,
+      "step": 4096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908909746628369e-06,
+      "loss": 0.6374,
+      "step": 4097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.904308632871994e-06,
+      "loss": 0.8042,
+      "step": 4098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899708243829515e-06,
+      "loss": 0.7998,
+      "step": 4099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.895108580577897e-06,
+      "loss": 0.7747,
+      "step": 4100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890509644193941e-06,
+      "loss": 0.7831,
+      "step": 4101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885911435754279e-06,
+      "loss": 0.6958,
+      "step": 4102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881313956335368e-06,
+      "loss": 0.656,
+      "step": 4103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876717207013499e-06,
+      "loss": 0.7194,
+      "step": 4104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872121188864787e-06,
+      "loss": 0.7512,
+      "step": 4105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86752590296518e-06,
+      "loss": 0.7921,
+      "step": 4106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.862931350390451e-06,
+      "loss": 0.7402,
+      "step": 4107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858337532216208e-06,
+      "loss": 0.7137,
+      "step": 4108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853744449517876e-06,
+      "loss": 0.7678,
+      "step": 4109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84915210337072e-06,
+      "loss": 0.7066,
+      "step": 4110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8445604948498245e-06,
+      "loss": 0.7277,
+      "step": 4111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.839969625030101e-06,
+      "loss": 0.7686,
+      "step": 4112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835379494986296e-06,
+      "loss": 0.7417,
+      "step": 4113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8307901057929735e-06,
+      "loss": 0.7489,
+      "step": 4114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826201458524531e-06,
+      "loss": 0.7068,
+      "step": 4115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821613554255189e-06,
+      "loss": 0.6638,
+      "step": 4116
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.81702639405899e-06,
+      "loss": 0.7087,
+      "step": 4117
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812439979009812e-06,
+      "loss": 0.7805,
+      "step": 4118
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.807854310181353e-06,
+      "loss": 0.7775,
+      "step": 4119
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8032693886471315e-06,
+      "loss": 0.775,
+      "step": 4120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.798685215480504e-06,
+      "loss": 0.6983,
+      "step": 4121
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7941017917546396e-06,
+      "loss": 0.6989,
+      "step": 4122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789519118542533e-06,
+      "loss": 0.7578,
+      "step": 4123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.784937196917014e-06,
+      "loss": 0.3419,
+      "step": 4124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780356027950722e-06,
+      "loss": 0.6993,
+      "step": 4125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.775775612716135e-06,
+      "loss": 0.6744,
+      "step": 4126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771195952285541e-06,
+      "loss": 0.7253,
+      "step": 4127
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766617047731054e-06,
+      "loss": 0.7295,
+      "step": 4128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.762038900124625e-06,
+      "loss": 0.7286,
+      "step": 4129
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757461510538008e-06,
+      "loss": 0.7831,
+      "step": 4130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752884880042787e-06,
+      "loss": 0.7611,
+      "step": 4131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748309009710379e-06,
+      "loss": 0.7673,
+      "step": 4132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743733900612004e-06,
+      "loss": 0.7064,
+      "step": 4133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.739159553818722e-06,
+      "loss": 0.7965,
+      "step": 4134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734585970401402e-06,
+      "loss": 0.3729,
+      "step": 4135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.730013151430733e-06,
+      "loss": 0.7302,
+      "step": 4136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.725441097977241e-06,
+      "loss": 0.6915,
+      "step": 4137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720869811111254e-06,
+      "loss": 0.6641,
+      "step": 4138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716299291902934e-06,
+      "loss": 0.7746,
+      "step": 4139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711729541422259e-06,
+      "loss": 0.7029,
+      "step": 4140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707160560739017e-06,
+      "loss": 0.7248,
+      "step": 4141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.702592350922835e-06,
+      "loss": 0.705,
+      "step": 4142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.698024913043142e-06,
+      "loss": 0.7332,
+      "step": 4143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693458248169202e-06,
+      "loss": 0.3679,
+      "step": 4144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.688892357370084e-06,
+      "loss": 0.7493,
+      "step": 4145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.684327241714679e-06,
+      "loss": 0.7893,
+      "step": 4146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679762902271705e-06,
+      "loss": 0.7373,
+      "step": 4147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675199340109691e-06,
+      "loss": 0.7693,
+      "step": 4148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670636556296981e-06,
+      "loss": 0.8495,
+      "step": 4149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666074551901746e-06,
+      "loss": 0.8084,
+      "step": 4150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661513327991967e-06,
+      "loss": 0.7568,
+      "step": 4151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65695288563545e-06,
+      "loss": 0.6537,
+      "step": 4152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652393225899808e-06,
+      "loss": 0.6976,
+      "step": 4153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6478343498524735e-06,
+      "loss": 0.8239,
+      "step": 4154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643276258560705e-06,
+      "loss": 0.7588,
+      "step": 4155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638718953091563e-06,
+      "loss": 0.7336,
+      "step": 4156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.634162434511939e-06,
+      "loss": 0.7749,
+      "step": 4157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629606703888527e-06,
+      "loss": 0.7108,
+      "step": 4158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.625051762287841e-06,
+      "loss": 0.6962,
+      "step": 4159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620497610776215e-06,
+      "loss": 0.7498,
+      "step": 4160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.615944250419791e-06,
+      "loss": 0.7504,
+      "step": 4161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611391682284534e-06,
+      "loss": 0.7933,
+      "step": 4162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606839907436215e-06,
+      "loss": 0.7406,
+      "step": 4163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.602288926940421e-06,
+      "loss": 0.7374,
+      "step": 4164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.59773874186256e-06,
+      "loss": 0.7611,
+      "step": 4165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593189353267848e-06,
+      "loss": 0.7466,
+      "step": 4166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.588640762221308e-06,
+      "loss": 0.737,
+      "step": 4167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5840929697877944e-06,
+      "loss": 0.7207,
+      "step": 4168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579545977031954e-06,
+      "loss": 0.7855,
+      "step": 4169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574999785018263e-06,
+      "loss": 0.6769,
+      "step": 4170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.570454394811003e-06,
+      "loss": 0.7833,
+      "step": 4171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.565909807474262e-06,
+      "loss": 0.7077,
+      "step": 4172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561366024071955e-06,
+      "loss": 0.7834,
+      "step": 4173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.556823045667795e-06,
+      "loss": 0.7242,
+      "step": 4174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.552280873325312e-06,
+      "loss": 0.6273,
+      "step": 4175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.547739508107847e-06,
+      "loss": 0.7395,
+      "step": 4176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.543198951078551e-06,
+      "loss": 0.333,
+      "step": 4177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.538659203300392e-06,
+      "loss": 0.8339,
+      "step": 4178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.534120265836137e-06,
+      "loss": 0.763,
+      "step": 4179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529582139748371e-06,
+      "loss": 0.3214,
+      "step": 4180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.525044826099493e-06,
+      "loss": 0.7453,
+      "step": 4181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5205083259517e-06,
+      "loss": 0.7612,
+      "step": 4182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515972640367007e-06,
+      "loss": 0.7766,
+      "step": 4183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.51143777040724e-06,
+      "loss": 0.3298,
+      "step": 4184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5069037171340245e-06,
+      "loss": 0.8096,
+      "step": 4185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502370481608804e-06,
+      "loss": 0.6907,
+      "step": 4186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497838064892826e-06,
+      "loss": 0.7201,
+      "step": 4187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.493306468047151e-06,
+      "loss": 0.3562,
+      "step": 4188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48877569213264e-06,
+      "loss": 0.7667,
+      "step": 4189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484245738209969e-06,
+      "loss": 0.7803,
+      "step": 4190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479716607339616e-06,
+      "loss": 0.6833,
+      "step": 4191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.475188300581869e-06,
+      "loss": 0.7758,
+      "step": 4192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470660818996828e-06,
+      "loss": 0.7299,
+      "step": 4193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.466134163644389e-06,
+      "loss": 0.8371,
+      "step": 4194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46160833558426e-06,
+      "loss": 0.7517,
+      "step": 4195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.457083335875962e-06,
+      "loss": 0.7101,
+      "step": 4196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452559165578807e-06,
+      "loss": 0.7837,
+      "step": 4197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44803582575193e-06,
+      "loss": 0.7237,
+      "step": 4198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4435133174542605e-06,
+      "loss": 0.7329,
+      "step": 4199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438991641744531e-06,
+      "loss": 0.6868,
+      "step": 4200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434470799681292e-06,
+      "loss": 0.7883,
+      "step": 4201
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429950792322884e-06,
+      "loss": 0.6935,
+      "step": 4202
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.425431620727463e-06,
+      "loss": 0.7567,
+      "step": 4203
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.420913285952986e-06,
+      "loss": 0.6899,
+      "step": 4204
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.416395789057209e-06,
+      "loss": 0.7576,
+      "step": 4205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411879131097702e-06,
+      "loss": 0.8307,
+      "step": 4206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4073633131318314e-06,
+      "loss": 0.6702,
+      "step": 4207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402848336216764e-06,
+      "loss": 0.7961,
+      "step": 4208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3983342014094795e-06,
+      "loss": 0.7888,
+      "step": 4209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39382090976675e-06,
+      "loss": 0.8327,
+      "step": 4210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389308462345164e-06,
+      "loss": 0.6788,
+      "step": 4211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.384796860201097e-06,
+      "loss": 0.3638,
+      "step": 4212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3802861043907305e-06,
+      "loss": 0.6417,
+      "step": 4213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375776195970058e-06,
+      "loss": 0.7033,
+      "step": 4214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371267135994867e-06,
+      "loss": 0.7545,
+      "step": 4215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.366758925520738e-06,
+      "loss": 0.6819,
+      "step": 4216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36225156560307e-06,
+      "loss": 0.7521,
+      "step": 4217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357745057297049e-06,
+      "loss": 0.3555,
+      "step": 4218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.353239401657673e-06,
+      "loss": 0.7471,
+      "step": 4219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3487345997397305e-06,
+      "loss": 0.7442,
+      "step": 4220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34423065259781e-06,
+      "loss": 0.7238,
+      "step": 4221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3397275612863115e-06,
+      "loss": 0.6831,
+      "step": 4222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.33522532685942e-06,
+      "loss": 0.3424,
+      "step": 4223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.330723950371133e-06,
+      "loss": 0.7114,
+      "step": 4224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326223432875238e-06,
+      "loss": 0.787,
+      "step": 4225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.32172377542532e-06,
+      "loss": 0.7734,
+      "step": 4226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317224979074776e-06,
+      "loss": 0.7258,
+      "step": 4227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.312727044876783e-06,
+      "loss": 0.7171,
+      "step": 4228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3082299738843365e-06,
+      "loss": 0.6979,
+      "step": 4229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303733767150211e-06,
+      "loss": 0.7215,
+      "step": 4230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.299238425726985e-06,
+      "loss": 0.7397,
+      "step": 4231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.294743950667044e-06,
+      "loss": 0.7165,
+      "step": 4232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2902503430225596e-06,
+      "loss": 0.7618,
+      "step": 4233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285757603845497e-06,
+      "loss": 0.7971,
+      "step": 4234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.281265734187635e-06,
+      "loss": 0.823,
+      "step": 4235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276774735100529e-06,
+      "loss": 0.8087,
+      "step": 4236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.272284607635551e-06,
+      "loss": 0.7389,
+      "step": 4237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267795352843849e-06,
+      "loss": 0.704,
+      "step": 4238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2633069717763775e-06,
+      "loss": 0.7114,
+      "step": 4239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258819465483888e-06,
+      "loss": 0.7432,
+      "step": 4240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254332835016917e-06,
+      "loss": 0.7672,
+      "step": 4241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249847081425813e-06,
+      "loss": 0.7022,
+      "step": 4242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.245362205760703e-06,
+      "loss": 0.7016,
+      "step": 4243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2408782090715115e-06,
+      "loss": 0.758,
+      "step": 4244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236395092407967e-06,
+      "loss": 0.3282,
+      "step": 4245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.231912856819579e-06,
+      "loss": 0.7014,
+      "step": 4246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.227431503355666e-06,
+      "loss": 0.7042,
+      "step": 4247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.222951033065324e-06,
+      "loss": 0.7011,
+      "step": 4248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.218471446997447e-06,
+      "loss": 0.7908,
+      "step": 4249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213992746200732e-06,
+      "loss": 0.7194,
+      "step": 4250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2095149317236555e-06,
+      "loss": 0.7986,
+      "step": 4251
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2050380046144895e-06,
+      "loss": 0.3279,
+      "step": 4252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2005619659213105e-06,
+      "loss": 0.7047,
+      "step": 4253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.196086816691966e-06,
+      "loss": 0.6707,
+      "step": 4254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191612557974116e-06,
+      "loss": 0.7127,
+      "step": 4255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187139190815199e-06,
+      "loss": 0.7829,
+      "step": 4256
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.182666716262443e-06,
+      "loss": 0.7135,
+      "step": 4257
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1781951353628835e-06,
+      "loss": 0.3724,
+      "step": 4258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173724449163326e-06,
+      "loss": 0.746,
+      "step": 4259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1692546587103794e-06,
+      "loss": 0.7646,
+      "step": 4260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164785765050443e-06,
+      "loss": 0.754,
+      "step": 4261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.160317769229698e-06,
+      "loss": 0.6608,
+      "step": 4262
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1558506722941225e-06,
+      "loss": 0.6963,
+      "step": 4263
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151384475289482e-06,
+      "loss": 0.7399,
+      "step": 4264
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146919179261331e-06,
+      "loss": 0.7237,
+      "step": 4265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142454785255017e-06,
+      "loss": 0.7036,
+      "step": 4266
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.137991294315665e-06,
+      "loss": 0.7735,
+      "step": 4267
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1335287074882025e-06,
+      "loss": 0.6699,
+      "step": 4268
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129067025817339e-06,
+      "loss": 0.7834,
+      "step": 4269
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1246062503475675e-06,
+      "loss": 0.7062,
+      "step": 4270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.120146382123178e-06,
+      "loss": 0.7739,
+      "step": 4271
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115687422188242e-06,
+      "loss": 0.7328,
+      "step": 4272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111229371586622e-06,
+      "loss": 0.353,
+      "step": 4273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106772231361964e-06,
+      "loss": 0.7575,
+      "step": 4274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102316002557702e-06,
+      "loss": 0.3704,
+      "step": 4275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0978606862170585e-06,
+      "loss": 0.715,
+      "step": 4276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0934062833830385e-06,
+      "loss": 0.6895,
+      "step": 4277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088952795098442e-06,
+      "loss": 0.7214,
+      "step": 4278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084500222405842e-06,
+      "loss": 0.8096,
+      "step": 4279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080048566347606e-06,
+      "loss": 0.5932,
+      "step": 4280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0755978279658875e-06,
+      "loss": 0.7285,
+      "step": 4281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.071148008302617e-06,
+      "loss": 0.3603,
+      "step": 4282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0666991083995205e-06,
+      "loss": 0.7134,
+      "step": 4283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.062251129298103e-06,
+      "loss": 0.7136,
+      "step": 4284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.057804072039647e-06,
+      "loss": 0.7163,
+      "step": 4285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.053357937665237e-06,
+      "loss": 0.7905,
+      "step": 4286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048912727215727e-06,
+      "loss": 0.706,
+      "step": 4287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.044468441731755e-06,
+      "loss": 0.3354,
+      "step": 4288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040025082253751e-06,
+      "loss": 0.7194,
+      "step": 4289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.03558264982192e-06,
+      "loss": 0.5935,
+      "step": 4290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.031141145476258e-06,
+      "loss": 0.6998,
+      "step": 4291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.026700570256539e-06,
+      "loss": 0.6545,
+      "step": 4292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02226092520231e-06,
+      "loss": 0.7835,
+      "step": 4293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017822211352922e-06,
+      "loss": 0.7625,
+      "step": 4294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.013384429747489e-06,
+      "loss": 0.7853,
+      "step": 4295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008947581424917e-06,
+      "loss": 0.7179,
+      "step": 4296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.004511667423889e-06,
+      "loss": 0.7608,
+      "step": 4297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000076688782867e-06,
+      "loss": 0.8325,
+      "step": 4298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995642646540105e-06,
+      "loss": 0.7505,
+      "step": 4299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991209541733627e-06,
+      "loss": 0.3459,
+      "step": 4300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986777375401233e-06,
+      "loss": 0.8037,
+      "step": 4301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.982346148580525e-06,
+      "loss": 0.7526,
+      "step": 4302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977915862308859e-06,
+      "loss": 0.7229,
+      "step": 4303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.973486517623391e-06,
+      "loss": 0.6845,
+      "step": 4304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.969058115561046e-06,
+      "loss": 0.7344,
+      "step": 4305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.964630657158526e-06,
+      "loss": 0.7015,
+      "step": 4306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960204143452326e-06,
+      "loss": 0.7503,
+      "step": 4307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955778575478701e-06,
+      "loss": 0.7362,
+      "step": 4308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951353954273705e-06,
+      "loss": 0.7915,
+      "step": 4309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946930280873153e-06,
+      "loss": 0.7417,
+      "step": 4310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.942507556312644e-06,
+      "loss": 0.7207,
+      "step": 4311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938085781627561e-06,
+      "loss": 0.763,
+      "step": 4312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.933664957853051e-06,
+      "loss": 0.7591,
+      "step": 4313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929245086024058e-06,
+      "loss": 0.7274,
+      "step": 4314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924826167175285e-06,
+      "loss": 0.7274,
+      "step": 4315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920408202341215e-06,
+      "loss": 0.7547,
+      "step": 4316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.915991192556122e-06,
+      "loss": 0.6496,
+      "step": 4317
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.91157513885404e-06,
+      "loss": 0.8258,
+      "step": 4318
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90716004226878e-06,
+      "loss": 0.7049,
+      "step": 4319
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.902745903833943e-06,
+      "loss": 0.3473,
+      "step": 4320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89833272458289e-06,
+      "loss": 0.7898,
+      "step": 4321
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89392050554877e-06,
+      "loss": 0.7446,
+      "step": 4322
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.889509247764499e-06,
+      "loss": 0.7351,
+      "step": 4323
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885098952262766e-06,
+      "loss": 0.7099,
+      "step": 4324
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.880689620076045e-06,
+      "loss": 0.6646,
+      "step": 4325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876281252236574e-06,
+      "loss": 0.7543,
+      "step": 4326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.871873849776376e-06,
+      "loss": 0.7944,
+      "step": 4327
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.867467413727237e-06,
+      "loss": 0.7457,
+      "step": 4328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863061945120719e-06,
+      "loss": 0.7387,
+      "step": 4329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858657444988167e-06,
+      "loss": 0.7633,
+      "step": 4330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854253914360686e-06,
+      "loss": 0.6533,
+      "step": 4331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849851354269166e-06,
+      "loss": 0.714,
+      "step": 4332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845449765744261e-06,
+      "loss": 0.7026,
+      "step": 4333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.841049149816398e-06,
+      "loss": 0.7349,
+      "step": 4334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836649507515785e-06,
+      "loss": 0.761,
+      "step": 4335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832250839872396e-06,
+      "loss": 0.7516,
+      "step": 4336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827853147915965e-06,
+      "loss": 0.8273,
+      "step": 4337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823456432676027e-06,
+      "loss": 0.7672,
+      "step": 4338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.819060695181854e-06,
+      "loss": 0.7486,
+      "step": 4339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.814665936462522e-06,
+      "loss": 0.6811,
+      "step": 4340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.810272157546853e-06,
+      "loss": 0.7873,
+      "step": 4341
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.805879359463446e-06,
+      "loss": 0.8101,
+      "step": 4342
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801487543240678e-06,
+      "loss": 0.6758,
+      "step": 4343
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797096709906689e-06,
+      "loss": 0.7673,
+      "step": 4344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.792706860489392e-06,
+      "loss": 0.8223,
+      "step": 4345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788317996016471e-06,
+      "loss": 0.6831,
+      "step": 4346
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7839301175153676e-06,
+      "loss": 0.7625,
+      "step": 4347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.779543226013315e-06,
+      "loss": 0.7674,
+      "step": 4348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.775157322537286e-06,
+      "loss": 0.772,
+      "step": 4349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770772408114057e-06,
+      "loss": 0.6436,
+      "step": 4350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.766388483770141e-06,
+      "loss": 0.6635,
+      "step": 4351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762005550531838e-06,
+      "loss": 0.7615,
+      "step": 4352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757623609425206e-06,
+      "loss": 0.6978,
+      "step": 4353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753242661476084e-06,
+      "loss": 0.777,
+      "step": 4354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748862707710055e-06,
+      "loss": 0.7874,
+      "step": 4355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744483749152501e-06,
+      "loss": 0.342,
+      "step": 4356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740105786828536e-06,
+      "loss": 0.7729,
+      "step": 4357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.735728821763078e-06,
+      "loss": 0.7891,
+      "step": 4358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.731352854980779e-06,
+      "loss": 0.7568,
+      "step": 4359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.726977887506074e-06,
+      "loss": 0.7584,
+      "step": 4360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.722603920363159e-06,
+      "loss": 0.7624,
+      "step": 4361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.718230954576001e-06,
+      "loss": 0.7316,
+      "step": 4362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713858991168326e-06,
+      "loss": 0.7188,
+      "step": 4363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7094880311636345e-06,
+      "loss": 0.7849,
+      "step": 4364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705118075585172e-06,
+      "loss": 0.7523,
+      "step": 4365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7007491254559796e-06,
+      "loss": 0.6845,
+      "step": 4366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.696381181798832e-06,
+      "loss": 0.7708,
+      "step": 4367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.692014245636297e-06,
+      "loss": 0.7884,
+      "step": 4368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68764831799068e-06,
+      "loss": 0.7763,
+      "step": 4369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683283399884064e-06,
+      "loss": 0.6935,
+      "step": 4370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.678919492338297e-06,
+      "loss": 0.7796,
+      "step": 4371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.674556596374993e-06,
+      "loss": 0.7357,
+      "step": 4372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.670194713015508e-06,
+      "loss": 0.764,
+      "step": 4373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665833843280995e-06,
+      "loss": 0.7844,
+      "step": 4374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6614739881923315e-06,
+      "loss": 0.7329,
+      "step": 4375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6571151487701985e-06,
+      "loss": 0.6815,
+      "step": 4376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6527573260350035e-06,
+      "loss": 0.7129,
+      "step": 4377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6484005210069335e-06,
+      "loss": 0.7999,
+      "step": 4378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.644044734705935e-06,
+      "loss": 0.6856,
+      "step": 4379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.639689968151716e-06,
+      "loss": 0.7342,
+      "step": 4380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635336222363744e-06,
+      "loss": 0.643,
+      "step": 4381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630983498361254e-06,
+      "loss": 0.6884,
+      "step": 4382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6266317971632255e-06,
+      "loss": 0.7462,
+      "step": 4383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622281119788417e-06,
+      "loss": 0.763,
+      "step": 4384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617931467255336e-06,
+      "loss": 0.7482,
+      "step": 4385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613582840582257e-06,
+      "loss": 0.6666,
+      "step": 4386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60923524078721e-06,
+      "loss": 0.7737,
+      "step": 4387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604888668887983e-06,
+      "loss": 0.7366,
+      "step": 4388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.600543125902131e-06,
+      "loss": 0.733,
+      "step": 4389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.596198612846964e-06,
+      "loss": 0.7721,
+      "step": 4390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.591855130739543e-06,
+      "loss": 0.7763,
+      "step": 4391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587512680596698e-06,
+      "loss": 0.7001,
+      "step": 4392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.583171263435016e-06,
+      "loss": 0.7238,
+      "step": 4393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.578830880270838e-06,
+      "loss": 0.7544,
+      "step": 4394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.574491532120268e-06,
+      "loss": 0.7998,
+      "step": 4395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.570153219999161e-06,
+      "loss": 0.7112,
+      "step": 4396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565815944923138e-06,
+      "loss": 0.7189,
+      "step": 4397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56147970790757e-06,
+      "loss": 0.8012,
+      "step": 4398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557144509967592e-06,
+      "loss": 0.7619,
+      "step": 4399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.552810352118084e-06,
+      "loss": 0.7028,
+      "step": 4400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548477235373694e-06,
+      "loss": 0.7494,
+      "step": 4401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.544145160748821e-06,
+      "loss": 0.7449,
+      "step": 4402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5398141292576215e-06,
+      "loss": 0.7343,
+      "step": 4403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535484141914009e-06,
+      "loss": 0.7718,
+      "step": 4404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.531155199731653e-06,
+      "loss": 0.7706,
+      "step": 4405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526827303723967e-06,
+      "loss": 0.712,
+      "step": 4406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.522500454904143e-06,
+      "loss": 0.7387,
+      "step": 4407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518174654285101e-06,
+      "loss": 0.6861,
+      "step": 4408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.513849902879535e-06,
+      "loss": 0.8058,
+      "step": 4409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5095262016998855e-06,
+      "loss": 0.3549,
+      "step": 4410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.505203551758348e-06,
+      "loss": 0.6721,
+      "step": 4411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500881954066872e-06,
+      "loss": 0.738,
+      "step": 4412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496561409637169e-06,
+      "loss": 0.7967,
+      "step": 4413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.492241919480678e-06,
+      "loss": 0.6522,
+      "step": 4414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.487923484608629e-06,
+      "loss": 0.6975,
+      "step": 4415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.483606106031969e-06,
+      "loss": 0.7554,
+      "step": 4416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.479289784761428e-06,
+      "loss": 0.7376,
+      "step": 4417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474974521807463e-06,
+      "loss": 0.813,
+      "step": 4418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4706603181803005e-06,
+      "loss": 0.6864,
+      "step": 4419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46634717488991e-06,
+      "loss": 0.6564,
+      "step": 4420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.462035092946021e-06,
+      "loss": 0.8386,
+      "step": 4421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457724073358099e-06,
+      "loss": 0.704,
+      "step": 4422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453414117135385e-06,
+      "loss": 0.7829,
+      "step": 4423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.449105225286842e-06,
+      "loss": 0.7959,
+      "step": 4424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.444797398821213e-06,
+      "loss": 0.7667,
+      "step": 4425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44049063874697e-06,
+      "loss": 0.702,
+      "step": 4426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.436184946072345e-06,
+      "loss": 0.6894,
+      "step": 4427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431880321805316e-06,
+      "loss": 0.683,
+      "step": 4428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427576766953615e-06,
+      "loss": 0.7827,
+      "step": 4429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4232742825247216e-06,
+      "loss": 0.7124,
+      "step": 4430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.418972869525869e-06,
+      "loss": 0.6944,
+      "step": 4431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414672528964022e-06,
+      "loss": 0.6878,
+      "step": 4432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410373261845926e-06,
+      "loss": 0.7822,
+      "step": 4433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406075069178039e-06,
+      "loss": 0.7874,
+      "step": 4434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401777951966602e-06,
+      "loss": 0.7402,
+      "step": 4435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3974819112175745e-06,
+      "loss": 0.7231,
+      "step": 4436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3931869479366826e-06,
+      "loss": 0.7987,
+      "step": 4437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3888930631293946e-06,
+      "loss": 0.3367,
+      "step": 4438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3846002578009294e-06,
+      "loss": 0.7836,
+      "step": 4439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3803085329562375e-06,
+      "loss": 0.8189,
+      "step": 4440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376017889600047e-06,
+      "loss": 0.6747,
+      "step": 4441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.371728328736798e-06,
+      "loss": 0.7279,
+      "step": 4442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367439851370709e-06,
+      "loss": 0.7125,
+      "step": 4443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36315245850572e-06,
+      "loss": 0.7326,
+      "step": 4444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35886615114553e-06,
+      "loss": 0.7122,
+      "step": 4445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35458093029358e-06,
+      "loss": 0.7075,
+      "step": 4446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.350296796953059e-06,
+      "loss": 0.7684,
+      "step": 4447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346013752126899e-06,
+      "loss": 0.358,
+      "step": 4448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.341731796817784e-06,
+      "loss": 0.7407,
+      "step": 4449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337450932028122e-06,
+      "loss": 0.7122,
+      "step": 4450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3331711587601e-06,
+      "loss": 0.6467,
+      "step": 4451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328892478015612e-06,
+      "loss": 0.7595,
+      "step": 4452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32461489079633e-06,
+      "loss": 0.7189,
+      "step": 4453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.320338398103646e-06,
+      "loss": 0.6969,
+      "step": 4454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316063000938705e-06,
+      "loss": 0.3643,
+      "step": 4455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311788700302395e-06,
+      "loss": 0.7387,
+      "step": 4456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307515497195352e-06,
+      "loss": 0.7273,
+      "step": 4457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3032433926179395e-06,
+      "loss": 0.7602,
+      "step": 4458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298972387570288e-06,
+      "loss": 0.7697,
+      "step": 4459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294702483052243e-06,
+      "loss": 0.3235,
+      "step": 4460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29043368006342e-06,
+      "loss": 0.7757,
+      "step": 4461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.286165979603155e-06,
+      "loss": 0.6621,
+      "step": 4462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281899382670535e-06,
+      "loss": 0.7573,
+      "step": 4463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277633890264386e-06,
+      "loss": 0.6946,
+      "step": 4464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.273369503383281e-06,
+      "loss": 0.7633,
+      "step": 4465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269106223025528e-06,
+      "loss": 0.716,
+      "step": 4466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.264844050189184e-06,
+      "loss": 0.7509,
+      "step": 4467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.260582985872031e-06,
+      "loss": 0.7203,
+      "step": 4468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256323031071607e-06,
+      "loss": 0.321,
+      "step": 4469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252064186785183e-06,
+      "loss": 0.7583,
+      "step": 4470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.247806454009773e-06,
+      "loss": 0.7242,
+      "step": 4471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2435498337421284e-06,
+      "loss": 0.8351,
+      "step": 4472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239294326978742e-06,
+      "loss": 0.7865,
+      "step": 4473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235039934715846e-06,
+      "loss": 0.3387,
+      "step": 4474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230786657949412e-06,
+      "loss": 0.6778,
+      "step": 4475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226534497675145e-06,
+      "loss": 0.7565,
+      "step": 4476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222283454888495e-06,
+      "loss": 0.8027,
+      "step": 4477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2180335305846485e-06,
+      "loss": 0.3296,
+      "step": 4478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213784725758529e-06,
+      "loss": 0.7499,
+      "step": 4479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209537041404802e-06,
+      "loss": 0.7577,
+      "step": 4480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.205290478517864e-06,
+      "loss": 0.7906,
+      "step": 4481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201045038091859e-06,
+      "loss": 0.6938,
+      "step": 4482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.196800721120648e-06,
+      "loss": 0.7202,
+      "step": 4483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192557528597861e-06,
+      "loss": 0.7653,
+      "step": 4484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188315461516833e-06,
+      "loss": 0.6582,
+      "step": 4485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184074520870652e-06,
+      "loss": 0.8119,
+      "step": 4486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.179834707652143e-06,
+      "loss": 0.7528,
+      "step": 4487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175596022853859e-06,
+      "loss": 0.7598,
+      "step": 4488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171358467468097e-06,
+      "loss": 0.8009,
+      "step": 4489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16712204248689e-06,
+      "loss": 0.7163,
+      "step": 4490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162886748901987e-06,
+      "loss": 0.7672,
+      "step": 4491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1586525877049045e-06,
+      "loss": 0.7614,
+      "step": 4492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154419559886866e-06,
+      "loss": 0.7235,
+      "step": 4493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.150187666438844e-06,
+      "loss": 0.7491,
+      "step": 4494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.145956908351542e-06,
+      "loss": 0.6712,
+      "step": 4495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.141727286615398e-06,
+      "loss": 0.6838,
+      "step": 4496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137498802220582e-06,
+      "loss": 0.7611,
+      "step": 4497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133271456157004e-06,
+      "loss": 0.789,
+      "step": 4498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.129045249414292e-06,
+      "loss": 0.7744,
+      "step": 4499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.124820182981833e-06,
+      "loss": 0.7613,
+      "step": 4500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.120596257848716e-06,
+      "loss": 0.7116,
+      "step": 4501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.116373475003797e-06,
+      "loss": 0.8218,
+      "step": 4502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1121518354356306e-06,
+      "loss": 0.7372,
+      "step": 4503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.107931340132528e-06,
+      "loss": 0.768,
+      "step": 4504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103711990082522e-06,
+      "loss": 0.7467,
+      "step": 4505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099493786273384e-06,
+      "loss": 0.7421,
+      "step": 4506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0952767296926e-06,
+      "loss": 0.6988,
+      "step": 4507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091060821327416e-06,
+      "loss": 0.7691,
+      "step": 4508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08684606216478e-06,
+      "loss": 0.7782,
+      "step": 4509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0826324531913966e-06,
+      "loss": 0.743,
+      "step": 4510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078419995393677e-06,
+      "loss": 0.7079,
+      "step": 4511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.074208689757782e-06,
+      "loss": 0.7802,
+      "step": 4512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069998537269592e-06,
+      "loss": 0.7396,
+      "step": 4513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0657895389147225e-06,
+      "loss": 0.3512,
+      "step": 4514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061581695678516e-06,
+      "loss": 0.7434,
+      "step": 4515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057375008546051e-06,
+      "loss": 0.7588,
+      "step": 4516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.053169478502116e-06,
+      "loss": 0.754,
+      "step": 4517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048965106531259e-06,
+      "loss": 0.6539,
+      "step": 4518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.044761893617727e-06,
+      "loss": 0.7757,
+      "step": 4519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.040559840745522e-06,
+      "loss": 0.7685,
+      "step": 4520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.036358948898351e-06,
+      "loss": 0.7154,
+      "step": 4521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032159219059664e-06,
+      "loss": 0.6621,
+      "step": 4522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027960652212636e-06,
+      "loss": 0.7465,
+      "step": 4523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02376324934017e-06,
+      "loss": 0.7442,
+      "step": 4524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019567011424884e-06,
+      "loss": 0.7593,
+      "step": 4525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.01537193944915e-06,
+      "loss": 0.7423,
+      "step": 4526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.011178034395037e-06,
+      "loss": 0.7631,
+      "step": 4527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.006985297244369e-06,
+      "loss": 0.6972,
+      "step": 4528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.002793728978672e-06,
+      "loss": 0.7415,
+      "step": 4529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998603330579213e-06,
+      "loss": 0.7837,
+      "step": 4530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994414103026981e-06,
+      "loss": 0.7738,
+      "step": 4531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990226047302691e-06,
+      "loss": 0.7066,
+      "step": 4532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9860391643867844e-06,
+      "loss": 0.727,
+      "step": 4533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98185345525943e-06,
+      "loss": 0.7025,
+      "step": 4534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977668920900509e-06,
+      "loss": 0.663,
+      "step": 4535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973485562289654e-06,
+      "loss": 0.774,
+      "step": 4536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969303380406189e-06,
+      "loss": 0.7104,
+      "step": 4537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965122376229195e-06,
+      "loss": 0.731,
+      "step": 4538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.960942550737452e-06,
+      "loss": 0.6973,
+      "step": 4539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956763904909477e-06,
+      "loss": 0.7005,
+      "step": 4540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952586439723509e-06,
+      "loss": 0.7088,
+      "step": 4541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948410156157512e-06,
+      "loss": 0.6482,
+      "step": 4542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944235055189159e-06,
+      "loss": 0.7751,
+      "step": 4543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940061137795876e-06,
+      "loss": 0.683,
+      "step": 4544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935888404954779e-06,
+      "loss": 0.7079,
+      "step": 4545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931716857642729e-06,
+      "loss": 0.7137,
+      "step": 4546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.927546496836299e-06,
+      "loss": 0.7293,
+      "step": 4547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92337732351179e-06,
+      "loss": 0.723,
+      "step": 4548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9192093386452215e-06,
+      "loss": 0.6951,
+      "step": 4549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.915042543212334e-06,
+      "loss": 0.7281,
+      "step": 4550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9108769381885926e-06,
+      "loss": 0.7863,
+      "step": 4551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906712524549187e-06,
+      "loss": 0.768,
+      "step": 4552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.902549303269013e-06,
+      "loss": 0.7413,
+      "step": 4553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898387275322704e-06,
+      "loss": 0.6857,
+      "step": 4554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.894226441684606e-06,
+      "loss": 0.7353,
+      "step": 4555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890066803328789e-06,
+      "loss": 0.7709,
+      "step": 4556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885908361229039e-06,
+      "loss": 0.657,
+      "step": 4557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881751116358866e-06,
+      "loss": 0.7939,
+      "step": 4558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.877595069691497e-06,
+      "loss": 0.7019,
+      "step": 4559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.873440222199882e-06,
+      "loss": 0.7367,
+      "step": 4560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8692865748566806e-06,
+      "loss": 0.638,
+      "step": 4561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865134128634285e-06,
+      "loss": 0.7151,
+      "step": 4562
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860982884504796e-06,
+      "loss": 0.682,
+      "step": 4563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856832843440038e-06,
+      "loss": 0.7738,
+      "step": 4564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852684006411553e-06,
+      "loss": 0.766,
+      "step": 4565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8485363743906e-06,
+      "loss": 0.7412,
+      "step": 4566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84438994834816e-06,
+      "loss": 0.3361,
+      "step": 4567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840244729254919e-06,
+      "loss": 0.6728,
+      "step": 4568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8361007180813e-06,
+      "loss": 0.7623,
+      "step": 4569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831957915797426e-06,
+      "loss": 0.7715,
+      "step": 4570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.827816323373147e-06,
+      "loss": 0.739,
+      "step": 4571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8236759417780234e-06,
+      "loss": 0.7123,
+      "step": 4572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.819536771981339e-06,
+      "loss": 0.6994,
+      "step": 4573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815398814952088e-06,
+      "loss": 0.7471,
+      "step": 4574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8112620716589876e-06,
+      "loss": 0.7679,
+      "step": 4575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807126543070454e-06,
+      "loss": 0.6959,
+      "step": 4576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8029922301546475e-06,
+      "loss": 0.7056,
+      "step": 4577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798859133879414e-06,
+      "loss": 0.7591,
+      "step": 4578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794727255212336e-06,
+      "loss": 0.6901,
+      "step": 4579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790596595120699e-06,
+      "loss": 0.7218,
+      "step": 4580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.786467154571508e-06,
+      "loss": 0.72,
+      "step": 4581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782338934531484e-06,
+      "loss": 0.7594,
+      "step": 4582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778211935967061e-06,
+      "loss": 0.7454,
+      "step": 4583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774086159844377e-06,
+      "loss": 0.7285,
+      "step": 4584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.769961607129307e-06,
+      "loss": 0.7187,
+      "step": 4585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.765838278787411e-06,
+      "loss": 0.7121,
+      "step": 4586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7617161757839895e-06,
+      "loss": 0.6771,
+      "step": 4587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757595299084036e-06,
+      "loss": 0.6865,
+      "step": 4588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.753475649652265e-06,
+      "loss": 0.7202,
+      "step": 4589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.749357228453105e-06,
+      "loss": 0.7304,
+      "step": 4590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.745240036450699e-06,
+      "loss": 0.7424,
+      "step": 4591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.741124074608886e-06,
+      "loss": 0.71,
+      "step": 4592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737009343891245e-06,
+      "loss": 0.7894,
+      "step": 4593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732895845261035e-06,
+      "loss": 0.7515,
+      "step": 4594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728783579681259e-06,
+      "loss": 0.7526,
+      "step": 4595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7246725481146025e-06,
+      "loss": 0.7505,
+      "step": 4596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.720562751523481e-06,
+      "loss": 0.8069,
+      "step": 4597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.716454190870013e-06,
+      "loss": 0.7488,
+      "step": 4598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.712346867116027e-06,
+      "loss": 0.7208,
+      "step": 4599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.708240781223067e-06,
+      "loss": 0.7517,
+      "step": 4600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.704135934152388e-06,
+      "loss": 0.7316,
+      "step": 4601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700032326864938e-06,
+      "loss": 0.3488,
+      "step": 4602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695929960321405e-06,
+      "loss": 0.6881,
+      "step": 4603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691828835482154e-06,
+      "loss": 0.7437,
+      "step": 4604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68772895330729e-06,
+      "loss": 0.3532,
+      "step": 4605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.683630314756601e-06,
+      "loss": 0.6977,
+      "step": 4606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.679532920789598e-06,
+      "loss": 0.7708,
+      "step": 4607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.675436772365498e-06,
+      "loss": 0.762,
+      "step": 4608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6713418704432304e-06,
+      "loss": 0.714,
+      "step": 4609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667248215981416e-06,
+      "loss": 0.7213,
+      "step": 4610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66315580993841e-06,
+      "loss": 0.7048,
+      "step": 4611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659064653272249e-06,
+      "loss": 0.707,
+      "step": 4612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654974746940703e-06,
+      "loss": 0.7383,
+      "step": 4613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.650886091901222e-06,
+      "loss": 0.7437,
+      "step": 4614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.646798689110984e-06,
+      "loss": 0.7398,
+      "step": 4615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642712539526866e-06,
+      "loss": 0.7767,
+      "step": 4616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63862764410545e-06,
+      "loss": 0.349,
+      "step": 4617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634544003803029e-06,
+      "loss": 0.7163,
+      "step": 4618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630461619575601e-06,
+      "loss": 0.8055,
+      "step": 4619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626380492378861e-06,
+      "loss": 0.6737,
+      "step": 4620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622300623168228e-06,
+      "loss": 0.6782,
+      "step": 4621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618222012898805e-06,
+      "loss": 0.7988,
+      "step": 4622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614144662525425e-06,
+      "loss": 0.6994,
+      "step": 4623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610068573002598e-06,
+      "loss": 0.351,
+      "step": 4624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605993745284562e-06,
+      "loss": 0.7433,
+      "step": 4625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.601920180325245e-06,
+      "loss": 0.7064,
+      "step": 4626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.597847879078295e-06,
+      "loss": 0.7305,
+      "step": 4627
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593776842497036e-06,
+      "loss": 0.7317,
+      "step": 4628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.589707071534536e-06,
+      "loss": 0.7123,
+      "step": 4629
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5856385671435285e-06,
+      "loss": 0.8334,
+      "step": 4630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581571330276474e-06,
+      "loss": 0.7523,
+      "step": 4631
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.577505361885527e-06,
+      "loss": 0.6861,
+      "step": 4632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573440662922548e-06,
+      "loss": 0.7512,
+      "step": 4633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5693772343390995e-06,
+      "loss": 0.6543,
+      "step": 4634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565315077086446e-06,
+      "loss": 0.7139,
+      "step": 4635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561254192115556e-06,
+      "loss": 0.7438,
+      "step": 4636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.557194580377102e-06,
+      "loss": 0.752,
+      "step": 4637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.553136242821449e-06,
+      "loss": 0.7784,
+      "step": 4638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5490791803986735e-06,
+      "loss": 0.7358,
+      "step": 4639
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54502339405855e-06,
+      "loss": 0.8266,
+      "step": 4640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540968884750554e-06,
+      "loss": 0.6973,
+      "step": 4641
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536915653423863e-06,
+      "loss": 0.3359,
+      "step": 4642
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532863701027357e-06,
+      "loss": 0.696,
+      "step": 4643
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.528813028509614e-06,
+      "loss": 0.7237,
+      "step": 4644
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52476363681891e-06,
+      "loss": 0.6798,
+      "step": 4645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520715526903225e-06,
+      "loss": 0.7204,
+      "step": 4646
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5166686997102375e-06,
+      "loss": 0.7518,
+      "step": 4647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512623156187327e-06,
+      "loss": 0.7522,
+      "step": 4648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508578897281573e-06,
+      "loss": 0.6785,
+      "step": 4649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.50453592393975e-06,
+      "loss": 0.7049,
+      "step": 4650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500494237108336e-06,
+      "loss": 0.6367,
+      "step": 4651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496453837733511e-06,
+      "loss": 0.7477,
+      "step": 4652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492414726761136e-06,
+      "loss": 0.8248,
+      "step": 4653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4883769051368e-06,
+      "loss": 0.6642,
+      "step": 4654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48434037380576e-06,
+      "loss": 0.6892,
+      "step": 4655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48030513371299e-06,
+      "loss": 0.7331,
+      "step": 4656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.476271185803155e-06,
+      "loss": 0.771,
+      "step": 4657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.472238531020622e-06,
+      "loss": 0.7201,
+      "step": 4658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.468207170309441e-06,
+      "loss": 0.6501,
+      "step": 4659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.464177104613385e-06,
+      "loss": 0.7408,
+      "step": 4660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460148334875895e-06,
+      "loss": 0.3344,
+      "step": 4661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456120862040134e-06,
+      "loss": 0.3638,
+      "step": 4662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4520946870489416e-06,
+      "loss": 0.6894,
+      "step": 4663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.448069810844864e-06,
+      "loss": 0.7367,
+      "step": 4664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444046234370141e-06,
+      "loss": 0.7205,
+      "step": 4665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44002395856671e-06,
+      "loss": 0.7241,
+      "step": 4666
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436002984376199e-06,
+      "loss": 0.744,
+      "step": 4667
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.431983312739939e-06,
+      "loss": 0.364,
+      "step": 4668
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427964944598943e-06,
+      "loss": 0.7964,
+      "step": 4669
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4239478808939376e-06,
+      "loss": 0.7237,
+      "step": 4670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4199321225653226e-06,
+      "loss": 0.8514,
+      "step": 4671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415917670553217e-06,
+      "loss": 0.6804,
+      "step": 4672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.411904525797408e-06,
+      "loss": 0.7325,
+      "step": 4673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.407892689237394e-06,
+      "loss": 0.7362,
+      "step": 4674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4038821618123625e-06,
+      "loss": 0.7496,
+      "step": 4675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399872944461198e-06,
+      "loss": 0.7199,
+      "step": 4676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395865038122461e-06,
+      "loss": 0.6678,
+      "step": 4677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3918584437344355e-06,
+      "loss": 0.751,
+      "step": 4678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387853162235067e-06,
+      "loss": 0.7829,
+      "step": 4679
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383849194562022e-06,
+      "loss": 0.7414,
+      "step": 4680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379846541652635e-06,
+      "loss": 0.739,
+      "step": 4681
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375845204443946e-06,
+      "loss": 0.6987,
+      "step": 4682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371845183872685e-06,
+      "loss": 0.7036,
+      "step": 4683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367846480875273e-06,
+      "loss": 0.6522,
+      "step": 4684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363849096387823e-06,
+      "loss": 0.7685,
+      "step": 4685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359853031346142e-06,
+      "loss": 0.7743,
+      "step": 4686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355858286685714e-06,
+      "loss": 0.7366,
+      "step": 4687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35186486334174e-06,
+      "loss": 0.6772,
+      "step": 4688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3478727622490815e-06,
+      "loss": 0.7701,
+      "step": 4689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.34388198434232e-06,
+      "loss": 0.6966,
+      "step": 4690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339892530555703e-06,
+      "loss": 0.7182,
+      "step": 4691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3359044018231784e-06,
+      "loss": 0.748,
+      "step": 4692
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331917599078388e-06,
+      "loss": 0.7968,
+      "step": 4693
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327932123254658e-06,
+      "loss": 0.7295,
+      "step": 4694
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3239479752849955e-06,
+      "loss": 0.6894,
+      "step": 4695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31996515610212e-06,
+      "loss": 0.775,
+      "step": 4696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315983666638412e-06,
+      "loss": 0.6943,
+      "step": 4697
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312003507825967e-06,
+      "loss": 0.6615,
+      "step": 4698
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308024680596546e-06,
+      "loss": 0.6944,
+      "step": 4699
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3040471858816115e-06,
+      "loss": 0.3135,
+      "step": 4700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300071024612314e-06,
+      "loss": 0.8139,
+      "step": 4701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296096197719485e-06,
+      "loss": 0.7416,
+      "step": 4702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29212270613365e-06,
+      "loss": 0.7019,
+      "step": 4703
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2881505507850245e-06,
+      "loss": 0.7324,
+      "step": 4704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284179732603492e-06,
+      "loss": 0.6851,
+      "step": 4705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280210252518653e-06,
+      "loss": 0.6918,
+      "step": 4706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276242111459768e-06,
+      "loss": 0.7195,
+      "step": 4707
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272275310355799e-06,
+      "loss": 0.694,
+      "step": 4708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268309850135389e-06,
+      "loss": 0.7381,
+      "step": 4709
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.26434573172687e-06,
+      "loss": 0.7551,
+      "step": 4710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2603829560582576e-06,
+      "loss": 0.7104,
+      "step": 4711
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256421524057257e-06,
+      "loss": 0.7408,
+      "step": 4712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252461436651244e-06,
+      "loss": 0.6943,
+      "step": 4713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248502694767306e-06,
+      "loss": 0.7128,
+      "step": 4714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24454529933219e-06,
+      "loss": 0.6736,
+      "step": 4715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240589251272342e-06,
+      "loss": 0.6831,
+      "step": 4716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23663455151389e-06,
+      "loss": 0.7622,
+      "step": 4717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232681200982643e-06,
+      "loss": 0.3474,
+      "step": 4718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.228729200604096e-06,
+      "loss": 0.7235,
+      "step": 4719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.22477855130343e-06,
+      "loss": 0.7853,
+      "step": 4720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2208292540055116e-06,
+      "loss": 0.7512,
+      "step": 4721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.21688130963488e-06,
+      "loss": 0.74,
+      "step": 4722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212934719115768e-06,
+      "loss": 0.7064,
+      "step": 4723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208989483372089e-06,
+      "loss": 0.7131,
+      "step": 4724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2050456033274374e-06,
+      "loss": 0.76,
+      "step": 4725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201103079905093e-06,
+      "loss": 0.7851,
+      "step": 4726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197161914028017e-06,
+      "loss": 0.711,
+      "step": 4727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19322210661885e-06,
+      "loss": 0.7723,
+      "step": 4728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1892836585999234e-06,
+      "loss": 0.7772,
+      "step": 4729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.185346570893233e-06,
+      "loss": 0.7487,
+      "step": 4730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181410844420473e-06,
+      "loss": 0.7512,
+      "step": 4731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177476480103013e-06,
+      "loss": 0.6218,
+      "step": 4732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.173543478861904e-06,
+      "loss": 0.7503,
+      "step": 4733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169611841617876e-06,
+      "loss": 0.7592,
+      "step": 4734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165681569291341e-06,
+      "loss": 0.7102,
+      "step": 4735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.161752662802393e-06,
+      "loss": 0.6648,
+      "step": 4736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1578251230708086e-06,
+      "loss": 0.7095,
+      "step": 4737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153898951016029e-06,
+      "loss": 0.8353,
+      "step": 4738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1499741475572016e-06,
+      "loss": 0.7127,
+      "step": 4739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.146050713613129e-06,
+      "loss": 0.7019,
+      "step": 4740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142128650102306e-06,
+      "loss": 0.8184,
+      "step": 4741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138207957942902e-06,
+      "loss": 0.7671,
+      "step": 4742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1342886380527736e-06,
+      "loss": 0.7289,
+      "step": 4743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130370691349436e-06,
+      "loss": 0.7547,
+      "step": 4744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126454118750112e-06,
+      "loss": 0.6667,
+      "step": 4745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1225389211716715e-06,
+      "loss": 0.7043,
+      "step": 4746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118625099530695e-06,
+      "loss": 0.7392,
+      "step": 4747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114712654743411e-06,
+      "loss": 0.7491,
+      "step": 4748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110801587725743e-06,
+      "loss": 0.748,
+      "step": 4749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.106891899393288e-06,
+      "loss": 0.7071,
+      "step": 4750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102983590661319e-06,
+      "loss": 0.336,
+      "step": 4751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.099076662444788e-06,
+      "loss": 0.7498,
+      "step": 4752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.095171115658325e-06,
+      "loss": 0.8231,
+      "step": 4753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091266951216225e-06,
+      "loss": 0.3539,
+      "step": 4754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0873641700324815e-06,
+      "loss": 0.7381,
+      "step": 4755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083462773020739e-06,
+      "loss": 0.7649,
+      "step": 4756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079562761094342e-06,
+      "loss": 0.7605,
+      "step": 4757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075664135166292e-06,
+      "loss": 0.7595,
+      "step": 4758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0717668961492725e-06,
+      "loss": 0.748,
+      "step": 4759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067871044955647e-06,
+      "loss": 0.8,
+      "step": 4760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063976582497451e-06,
+      "loss": 0.7392,
+      "step": 4761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060083509686384e-06,
+      "loss": 0.7446,
+      "step": 4762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056191827433844e-06,
+      "loss": 0.7216,
+      "step": 4763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.052301536650877e-06,
+      "loss": 0.6971,
+      "step": 4764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048412638248227e-06,
+      "loss": 0.7308,
+      "step": 4765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044525133136292e-06,
+      "loss": 0.7383,
+      "step": 4766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040639022225158e-06,
+      "loss": 0.757,
+      "step": 4767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036754306424576e-06,
+      "loss": 0.753,
+      "step": 4768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032870986643975e-06,
+      "loss": 0.7052,
+      "step": 4769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0289890637924575e-06,
+      "loss": 0.7662,
+      "step": 4770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0251085387787994e-06,
+      "loss": 0.6895,
+      "step": 4771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021229412511436e-06,
+      "loss": 0.7739,
+      "step": 4772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017351685898502e-06,
+      "loss": 0.7408,
+      "step": 4773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013475359847774e-06,
+      "loss": 0.6938,
+      "step": 4774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0096004352667295e-06,
+      "loss": 0.671,
+      "step": 4775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.005726913062494e-06,
+      "loss": 0.7372,
+      "step": 4776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00185479414188e-06,
+      "loss": 0.7754,
+      "step": 4777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9979840794113624e-06,
+      "loss": 0.7979,
+      "step": 4778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994114769777098e-06,
+      "loss": 0.7049,
+      "step": 4779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990246866144896e-06,
+      "loss": 0.6616,
+      "step": 4780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986380369420262e-06,
+      "loss": 0.6757,
+      "step": 4781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982515280508346e-06,
+      "loss": 0.7945,
+      "step": 4782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978651600313993e-06,
+      "loss": 0.7432,
+      "step": 4783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974789329741696e-06,
+      "loss": 0.6757,
+      "step": 4784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970928469695634e-06,
+      "loss": 0.7404,
+      "step": 4785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.967069021079647e-06,
+      "loss": 0.6899,
+      "step": 4786
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.96321098479725e-06,
+      "loss": 0.7764,
+      "step": 4787
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959354361751624e-06,
+      "loss": 0.7583,
+      "step": 4788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.955499152845622e-06,
+      "loss": 0.756,
+      "step": 4789
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.951645358981756e-06,
+      "loss": 0.6539,
+      "step": 4790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947792981062228e-06,
+      "loss": 0.737,
+      "step": 4791
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9439420199888836e-06,
+      "loss": 0.7079,
+      "step": 4792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940092476663253e-06,
+      "loss": 0.777,
+      "step": 4793
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93624435198653e-06,
+      "loss": 0.7319,
+      "step": 4794
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932397646859576e-06,
+      "loss": 0.7533,
+      "step": 4795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92855236218292e-06,
+      "loss": 0.7216,
+      "step": 4796
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924708498856762e-06,
+      "loss": 0.7767,
+      "step": 4797
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920866057780957e-06,
+      "loss": 0.7446,
+      "step": 4798
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9170250398550465e-06,
+      "loss": 0.7006,
+      "step": 4799
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913185445978223e-06,
+      "loss": 0.7959,
+      "step": 4800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909347277049349e-06,
+      "loss": 0.7784,
+      "step": 4801
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905510533966959e-06,
+      "loss": 0.7308,
+      "step": 4802
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90167521762925e-06,
+      "loss": 0.7682,
+      "step": 4803
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897841328934084e-06,
+      "loss": 0.8037,
+      "step": 4804
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89400886877899e-06,
+      "loss": 0.7585,
+      "step": 4805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890177838061167e-06,
+      "loss": 0.7148,
+      "step": 4806
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886348237677465e-06,
+      "loss": 0.6594,
+      "step": 4807
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.882520068524416e-06,
+      "loss": 0.7356,
+      "step": 4808
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878693331498208e-06,
+      "loss": 0.6991,
+      "step": 4809
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874868027494696e-06,
+      "loss": 0.7314,
+      "step": 4810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.871044157409398e-06,
+      "loss": 0.8053,
+      "step": 4811
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8672217221374995e-06,
+      "loss": 0.7519,
+      "step": 4812
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863400722573848e-06,
+      "loss": 0.7606,
+      "step": 4813
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8595811596129565e-06,
+      "loss": 0.6926,
+      "step": 4814
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855763034148995e-06,
+      "loss": 0.3347,
+      "step": 4815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851946347075804e-06,
+      "loss": 0.7783,
+      "step": 4816
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848131099286888e-06,
+      "loss": 0.7413,
+      "step": 4817
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844317291675409e-06,
+      "loss": 0.7295,
+      "step": 4818
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840504925134198e-06,
+      "loss": 0.7136,
+      "step": 4819
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836694000555745e-06,
+      "loss": 0.7144,
+      "step": 4820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832884518832195e-06,
+      "loss": 0.7552,
+      "step": 4821
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829076480855377e-06,
+      "loss": 0.6798,
+      "step": 4822
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825269887516754e-06,
+      "loss": 0.6898,
+      "step": 4823
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821464739707479e-06,
+      "loss": 0.7204,
+      "step": 4824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817661038318341e-06,
+      "loss": 0.771,
+      "step": 4825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8138587842398056e-06,
+      "loss": 0.7077,
+      "step": 4826
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8100579783619973e-06,
+      "loss": 0.763,
+      "step": 4827
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8062586215747034e-06,
+      "loss": 0.3241,
+      "step": 4828
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8024607147673564e-06,
+      "loss": 0.764,
+      "step": 4829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7986642588290766e-06,
+      "loss": 0.7371,
+      "step": 4830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7948692546486167e-06,
+      "loss": 0.7274,
+      "step": 4831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7910757031144164e-06,
+      "loss": 0.7585,
+      "step": 4832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7872836051145488e-06,
+      "loss": 0.736,
+      "step": 4833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7834929615367657e-06,
+      "loss": 0.7928,
+      "step": 4834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7797037732684705e-06,
+      "loss": 0.7448,
+      "step": 4835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.775916041196728e-06,
+      "loss": 0.367,
+      "step": 4836
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.772129766208262e-06,
+      "loss": 0.7196,
+      "step": 4837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7683449491894574e-06,
+      "loss": 0.7643,
+      "step": 4838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7645615910263457e-06,
+      "loss": 0.7019,
+      "step": 4839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7607796926046393e-06,
+      "loss": 0.7303,
+      "step": 4840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7569992548096833e-06,
+      "loss": 0.7613,
+      "step": 4841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.753220278526506e-06,
+      "loss": 0.7415,
+      "step": 4842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7494427646397714e-06,
+      "loss": 0.775,
+      "step": 4843
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7456667140338154e-06,
+      "loss": 0.7915,
+      "step": 4844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7418921275926245e-06,
+      "loss": 0.7161,
+      "step": 4845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7381190061998485e-06,
+      "loss": 0.3445,
+      "step": 4846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7343473507387817e-06,
+      "loss": 0.6876,
+      "step": 4847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7305771620923968e-06,
+      "loss": 0.7429,
+      "step": 4848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.726808441143297e-06,
+      "loss": 0.6789,
+      "step": 4849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7230411887737672e-06,
+      "loss": 0.726,
+      "step": 4850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7192754058657275e-06,
+      "loss": 0.8154,
+      "step": 4851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7155110933007667e-06,
+      "loss": 0.7197,
+      "step": 4852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7117482519601243e-06,
+      "loss": 0.7168,
+      "step": 4853
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.707986882724698e-06,
+      "loss": 0.6798,
+      "step": 4854
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7042269864750393e-06,
+      "loss": 0.7236,
+      "step": 4855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7004685640913583e-06,
+      "loss": 0.7695,
+      "step": 4856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6967116164535056e-06,
+      "loss": 0.3756,
+      "step": 4857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6929561444410134e-06,
+      "loss": 0.715,
+      "step": 4858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.689202148933039e-06,
+      "loss": 0.6996,
+      "step": 4859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6854496308084207e-06,
+      "loss": 0.6931,
+      "step": 4860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.681698590945628e-06,
+      "loss": 0.7343,
+      "step": 4861
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.677949030222798e-06,
+      "loss": 0.6662,
+      "step": 4862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6742009495177187e-06,
+      "loss": 0.795,
+      "step": 4863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6704543497078347e-06,
+      "loss": 0.7912,
+      "step": 4864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.666709231670229e-06,
+      "loss": 0.7576,
+      "step": 4865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6629655962816622e-06,
+      "loss": 0.8117,
+      "step": 4866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6592234444185228e-06,
+      "loss": 0.7621,
+      "step": 4867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6554827769568747e-06,
+      "loss": 0.7137,
+      "step": 4868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6517435947724158e-06,
+      "loss": 0.6903,
+      "step": 4869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6480058987405042e-06,
+      "loss": 0.7511,
+      "step": 4870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.644269689736153e-06,
+      "loss": 0.6868,
+      "step": 4871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.640534968634022e-06,
+      "loss": 0.3766,
+      "step": 4872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.636801736308424e-06,
+      "loss": 0.7004,
+      "step": 4873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.633069993633328e-06,
+      "loss": 0.7232,
+      "step": 4874
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6293397414823406e-06,
+      "loss": 0.6618,
+      "step": 4875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625610980728741e-06,
+      "loss": 0.8177,
+      "step": 4876
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6218837122454386e-06,
+      "loss": 0.7036,
+      "step": 4877
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6181579369050045e-06,
+      "loss": 0.719,
+      "step": 4878
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6144336555796575e-06,
+      "loss": 0.8047,
+      "step": 4879
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6107108691412672e-06,
+      "loss": 0.6656,
+      "step": 4880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.606989578461354e-06,
+      "loss": 0.6793,
+      "step": 4881
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6032697844110896e-06,
+      "loss": 0.6841,
+      "step": 4882
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.599551487861287e-06,
+      "loss": 0.6988,
+      "step": 4883
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.595834689682416e-06,
+      "loss": 0.7185,
+      "step": 4884
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.592119390744595e-06,
+      "loss": 0.6639,
+      "step": 4885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5884055919175907e-06,
+      "loss": 0.7348,
+      "step": 4886
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58469329407082e-06,
+      "loss": 0.7117,
+      "step": 4887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5809824980733445e-06,
+      "loss": 0.74,
+      "step": 4888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5772732047938772e-06,
+      "loss": 0.3284,
+      "step": 4889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.573565415100779e-06,
+      "loss": 0.7807,
+      "step": 4890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.569859129862062e-06,
+      "loss": 0.7383,
+      "step": 4891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5661543499453753e-06,
+      "loss": 0.7221,
+      "step": 4892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5624510762180277e-06,
+      "loss": 0.7124,
+      "step": 4893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558749309546968e-06,
+      "loss": 0.7011,
+      "step": 4894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5550490507987966e-06,
+      "loss": 0.706,
+      "step": 4895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.55135030083976e-06,
+      "loss": 0.779,
+      "step": 4896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5476530605357516e-06,
+      "loss": 0.7416,
+      "step": 4897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543957330752301e-06,
+      "loss": 0.7497,
+      "step": 4898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5402631123546085e-06,
+      "loss": 0.6833,
+      "step": 4899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5365704062074936e-06,
+      "loss": 0.7838,
+      "step": 4900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5328792131754374e-06,
+      "loss": 0.7449,
+      "step": 4901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5291895341225648e-06,
+      "loss": 0.3287,
+      "step": 4902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5255013699126417e-06,
+      "loss": 0.7054,
+      "step": 4903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5218147214090845e-06,
+      "loss": 0.7696,
+      "step": 4904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5181295894749567e-06,
+      "loss": 0.7809,
+      "step": 4905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5144459749729498e-06,
+      "loss": 0.7488,
+      "step": 4906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5107638787654285e-06,
+      "loss": 0.3468,
+      "step": 4907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5070833017143725e-06,
+      "loss": 0.7333,
+      "step": 4908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.503404244681432e-06,
+      "loss": 0.7221,
+      "step": 4909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4997267085278807e-06,
+      "loss": 0.6722,
+      "step": 4910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4960506941146475e-06,
+      "loss": 0.681,
+      "step": 4911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4923762023023023e-06,
+      "loss": 0.7289,
+      "step": 4912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4887032339510587e-06,
+      "loss": 0.7332,
+      "step": 4913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.485031789920772e-06,
+      "loss": 0.6991,
+      "step": 4914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.481361871070947e-06,
+      "loss": 0.6798,
+      "step": 4915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4776934782607176e-06,
+      "loss": 0.7882,
+      "step": 4916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4740266123488807e-06,
+      "loss": 0.7163,
+      "step": 4917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4703612741938543e-06,
+      "loss": 0.7637,
+      "step": 4918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4666974646537145e-06,
+      "loss": 0.655,
+      "step": 4919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.463035184586172e-06,
+      "loss": 0.7447,
+      "step": 4920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4593744348485814e-06,
+      "loss": 0.7034,
+      "step": 4921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.45571521629794e-06,
+      "loss": 0.7719,
+      "step": 4922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4520575297908886e-06,
+      "loss": 0.7074,
+      "step": 4923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4484013761836975e-06,
+      "loss": 0.7275,
+      "step": 4924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444746756332299e-06,
+      "loss": 0.332,
+      "step": 4925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4410936710922406e-06,
+      "loss": 0.7446,
+      "step": 4926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437442121318739e-06,
+      "loss": 0.6887,
+      "step": 4927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433792107866627e-06,
+      "loss": 0.7777,
+      "step": 4928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.430143631590389e-06,
+      "loss": 0.302,
+      "step": 4929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4264966933441502e-06,
+      "loss": 0.6858,
+      "step": 4930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422851293981676e-06,
+      "loss": 0.65,
+      "step": 4931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4192074343563598e-06,
+      "loss": 0.7042,
+      "step": 4932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4155651153212556e-06,
+      "loss": 0.6496,
+      "step": 4933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4119243377290336e-06,
+      "loss": 0.6732,
+      "step": 4934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.408285102432026e-06,
+      "loss": 0.7386,
+      "step": 4935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4046474102821846e-06,
+      "loss": 0.6809,
+      "step": 4936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.401011262131111e-06,
+      "loss": 0.7167,
+      "step": 4937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3973766588300407e-06,
+      "loss": 0.746,
+      "step": 4938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.39374360122985e-06,
+      "loss": 0.7713,
+      "step": 4939
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3901120901810516e-06,
+      "loss": 0.7496,
+      "step": 4940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.386482126533801e-06,
+      "loss": 0.7286,
+      "step": 4941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.382853711137878e-06,
+      "loss": 0.6896,
+      "step": 4942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3792268448427213e-06,
+      "loss": 0.7174,
+      "step": 4943
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.375601528497383e-06,
+      "loss": 0.7122,
+      "step": 4944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.371977762950577e-06,
+      "loss": 0.7297,
+      "step": 4945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3683555490506304e-06,
+      "loss": 0.697,
+      "step": 4946
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3647348876455244e-06,
+      "loss": 0.694,
+      "step": 4947
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3611157795828676e-06,
+      "loss": 0.7293,
+      "step": 4948
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.357498225709912e-06,
+      "loss": 0.6947,
+      "step": 4949
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3538822268735327e-06,
+      "loss": 0.6708,
+      "step": 4950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3502677839202624e-06,
+      "loss": 0.6898,
+      "step": 4951
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.346654897696243e-06,
+      "loss": 0.7469,
+      "step": 4952
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3430435690472797e-06,
+      "loss": 0.7284,
+      "step": 4953
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3394337988187896e-06,
+      "loss": 0.7776,
+      "step": 4954
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3358255878558366e-06,
+      "loss": 0.7604,
+      "step": 4955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3322189370031175e-06,
+      "loss": 0.736,
+      "step": 4956
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3286138471049657e-06,
+      "loss": 0.3523,
+      "step": 4957
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.325010319005345e-06,
+      "loss": 0.7415,
+      "step": 4958
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3214083535478615e-06,
+      "loss": 0.7508,
+      "step": 4959
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3178079515757422e-06,
+      "loss": 0.6741,
+      "step": 4960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.314209113931859e-06,
+      "loss": 0.8113,
+      "step": 4961
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3106118414587143e-06,
+      "loss": 0.6267,
+      "step": 4962
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.307016134998444e-06,
+      "loss": 0.7348,
+      "step": 4963
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.303421995392818e-06,
+      "loss": 0.6503,
+      "step": 4964
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2998294234832393e-06,
+      "loss": 0.6971,
+      "step": 4965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2962384201107435e-06,
+      "loss": 0.6918,
+      "step": 4966
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2926489861160017e-06,
+      "loss": 0.7669,
+      "step": 4967
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.289061122339309e-06,
+      "loss": 0.7523,
+      "step": 4968
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.285474829620601e-06,
+      "loss": 0.7211,
+      "step": 4969
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2818901087994437e-06,
+      "loss": 0.6947,
+      "step": 4970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2783069607150354e-06,
+      "loss": 0.7356,
+      "step": 4971
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2747253862062057e-06,
+      "loss": 0.6988,
+      "step": 4972
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.271145386111415e-06,
+      "loss": 0.7191,
+      "step": 4973
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2675669612687565e-06,
+      "loss": 0.3297,
+      "step": 4974
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2639901125159546e-06,
+      "loss": 0.758,
+      "step": 4975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2604148406903647e-06,
+      "loss": 0.6809,
+      "step": 4976
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2568411466289684e-06,
+      "loss": 0.6497,
+      "step": 4977
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2532690311683845e-06,
+      "loss": 0.6715,
+      "step": 4978
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2496984951448597e-06,
+      "loss": 0.735,
+      "step": 4979
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2461295393942704e-06,
+      "loss": 0.7222,
+      "step": 4980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2425621647521242e-06,
+      "loss": 0.757,
+      "step": 4981
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2389963720535623e-06,
+      "loss": 0.7289,
+      "step": 4982
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2354321621333386e-06,
+      "loss": 0.7365,
+      "step": 4983
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2318695358258644e-06,
+      "loss": 0.7603,
+      "step": 4984
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2283084939651545e-06,
+      "loss": 0.6739,
+      "step": 4985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2247490373848665e-06,
+      "loss": 0.7456,
+      "step": 4986
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2211911669182838e-06,
+      "loss": 0.7264,
+      "step": 4987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2176348833983184e-06,
+      "loss": 0.7271,
+      "step": 4988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2140801876575103e-06,
+      "loss": 0.3171,
+      "step": 4989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.210527080528031e-06,
+      "loss": 0.695,
+      "step": 4990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2069755628416687e-06,
+      "loss": 0.7648,
+      "step": 4991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.20342563542986e-06,
+      "loss": 0.747,
+      "step": 4992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1998772991236448e-06,
+      "loss": 0.7116,
+      "step": 4993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.196330554753716e-06,
+      "loss": 0.7553,
+      "step": 4994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1927854031503715e-06,
+      "loss": 0.7815,
+      "step": 4995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.189241845143548e-06,
+      "loss": 0.6799,
+      "step": 4996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185699881562807e-06,
+      "loss": 0.7151,
+      "step": 4997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.182159513237337e-06,
+      "loss": 0.7311,
+      "step": 4998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1786207409959526e-06,
+      "loss": 0.6747,
+      "step": 4999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1750835656670964e-06,
+      "loss": 0.7427,
+      "step": 5000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171547988078828e-06,
+      "loss": 0.7298,
+      "step": 5001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.168014009058853e-06,
+      "loss": 0.7874,
+      "step": 5002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1644816294344793e-06,
+      "loss": 0.7843,
+      "step": 5003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1609508500326558e-06,
+      "loss": 0.7867,
+      "step": 5004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.157421671679951e-06,
+      "loss": 0.7809,
+      "step": 5005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1538940952025597e-06,
+      "loss": 0.7153,
+      "step": 5006
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.150368121426304e-06,
+      "loss": 0.7096,
+      "step": 5007
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.14684375117663e-06,
+      "loss": 0.6365,
+      "step": 5008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1433209852785983e-06,
+      "loss": 0.6827,
+      "step": 5009
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.139799824556915e-06,
+      "loss": 0.7038,
+      "step": 5010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.136280269835884e-06,
+      "loss": 0.7351,
+      "step": 5011
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132762321939463e-06,
+      "loss": 0.6918,
+      "step": 5012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1292459816912067e-06,
+      "loss": 0.6937,
+      "step": 5013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.125731249914308e-06,
+      "loss": 0.7372,
+      "step": 5014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1222181274315787e-06,
+      "loss": 0.7332,
+      "step": 5015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.11870661506546e-06,
+      "loss": 0.7303,
+      "step": 5016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.115196713638e-06,
+      "loss": 0.7569,
+      "step": 5017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1116884239708957e-06,
+      "loss": 0.7629,
+      "step": 5018
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.108181746885437e-06,
+      "loss": 0.7064,
+      "step": 5019
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1046766832025653e-06,
+      "loss": 0.7723,
+      "step": 5020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10117323374282e-06,
+      "loss": 0.7508,
+      "step": 5021
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0976713993263763e-06,
+      "loss": 0.8124,
+      "step": 5022
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.094171180773028e-06,
+      "loss": 0.7331,
+      "step": 5023
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0906725789021906e-06,
+      "loss": 0.7343,
+      "step": 5024
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0871755945329e-06,
+      "loss": 0.6859,
+      "step": 5025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0836802284838172e-06,
+      "loss": 0.7552,
+      "step": 5026
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.080186481573214e-06,
+      "loss": 0.6718,
+      "step": 5027
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.076694354619003e-06,
+      "loss": 0.7493,
+      "step": 5028
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0732038484386915e-06,
+      "loss": 0.7224,
+      "step": 5029
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.069714963849434e-06,
+      "loss": 0.7291,
+      "step": 5030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0662277016679833e-06,
+      "loss": 0.7075,
+      "step": 5031
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0627420627107253e-06,
+      "loss": 0.6715,
+      "step": 5032
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0592580477936606e-06,
+      "loss": 0.6764,
+      "step": 5033
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0557756577324172e-06,
+      "loss": 0.7673,
+      "step": 5034
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0522948933422237e-06,
+      "loss": 0.7404,
+      "step": 5035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.048815755437957e-06,
+      "loss": 0.7205,
+      "step": 5036
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0453382448340828e-06,
+      "loss": 0.7924,
+      "step": 5037
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.041862362344713e-06,
+      "loss": 0.6719,
+      "step": 5038
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0383881087835576e-06,
+      "loss": 0.744,
+      "step": 5039
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0349154849639563e-06,
+      "loss": 0.7136,
+      "step": 5040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.031444491698864e-06,
+      "loss": 0.7504,
+      "step": 5041
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.027975129800853e-06,
+      "loss": 0.7291,
+      "step": 5042
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0245074000821175e-06,
+      "loss": 0.7393,
+      "step": 5043
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0210413033544684e-06,
+      "loss": 0.7467,
+      "step": 5044
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017576840429327e-06,
+      "loss": 0.7397,
+      "step": 5045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0141140121177427e-06,
+      "loss": 0.6745,
+      "step": 5046
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.010652819230375e-06,
+      "loss": 0.6542,
+      "step": 5047
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.007193262577506e-06,
+      "loss": 0.7261,
+      "step": 5048
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.003735342969031e-06,
+      "loss": 0.7602,
+      "step": 5049
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0002790612144608e-06,
+      "loss": 0.7422,
+      "step": 5050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.996824418122928e-06,
+      "loss": 0.6838,
+      "step": 5051
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.99337141450318e-06,
+      "loss": 0.6909,
+      "step": 5052
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.989920051163573e-06,
+      "loss": 0.7336,
+      "step": 5053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.986470328912089e-06,
+      "loss": 0.6972,
+      "step": 5054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.983022248556321e-06,
+      "loss": 0.7246,
+      "step": 5055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9795758109034787e-06,
+      "loss": 0.7831,
+      "step": 5056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9761310167603874e-06,
+      "loss": 0.6507,
+      "step": 5057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.972687866933488e-06,
+      "loss": 0.8378,
+      "step": 5058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.969246362228838e-06,
+      "loss": 0.3351,
+      "step": 5059
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.965806503452098e-06,
+      "loss": 0.7242,
+      "step": 5060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.962368291408566e-06,
+      "loss": 0.7639,
+      "step": 5061
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9589317269031326e-06,
+      "loss": 0.6539,
+      "step": 5062
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9554968107403127e-06,
+      "loss": 0.7235,
+      "step": 5063
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.952063543724236e-06,
+      "loss": 0.7449,
+      "step": 5064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.948631926658644e-06,
+      "loss": 0.7985,
+      "step": 5065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9452019603468906e-06,
+      "loss": 0.7565,
+      "step": 5066
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.941773645591949e-06,
+      "loss": 0.7381,
+      "step": 5067
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.938346983196393e-06,
+      "loss": 0.6645,
+      "step": 5068
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934921973962429e-06,
+      "loss": 0.6959,
+      "step": 5069
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.931498618691857e-06,
+      "loss": 0.692,
+      "step": 5070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9280769181861023e-06,
+      "loss": 0.7259,
+      "step": 5071
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9246568732461976e-06,
+      "loss": 0.6379,
+      "step": 5072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9212384846727893e-06,
+      "loss": 0.3254,
+      "step": 5073
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9178217532661368e-06,
+      "loss": 0.7933,
+      "step": 5074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.914406679826114e-06,
+      "loss": 0.7441,
+      "step": 5075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9109932651521934e-06,
+      "loss": 0.6129,
+      "step": 5076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.907581510043481e-06,
+      "loss": 0.7366,
+      "step": 5077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.904171415298672e-06,
+      "loss": 0.7263,
+      "step": 5078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9007629817160945e-06,
+      "loss": 0.7876,
+      "step": 5079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8973562100936668e-06,
+      "loss": 0.3767,
+      "step": 5080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8939511012289324e-06,
+      "loss": 0.3781,
+      "step": 5081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8905476559190404e-06,
+      "loss": 0.6917,
+      "step": 5082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8871458749607505e-06,
+      "loss": 0.6225,
+      "step": 5083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.883745759150435e-06,
+      "loss": 0.7174,
+      "step": 5084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8803473092840772e-06,
+      "loss": 0.7923,
+      "step": 5085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8769505261572572e-06,
+      "loss": 0.7084,
+      "step": 5086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.873555410565191e-06,
+      "loss": 0.7542,
+      "step": 5087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.870161963302677e-06,
+      "loss": 0.7377,
+      "step": 5088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8667701851641384e-06,
+      "loss": 0.7151,
+      "step": 5089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.863380076943606e-06,
+      "loss": 0.7118,
+      "step": 5090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.859991639434716e-06,
+      "loss": 0.6709,
+      "step": 5091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.856604873430717e-06,
+      "loss": 0.7381,
+      "step": 5092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8532197797244663e-06,
+      "loss": 0.7044,
+      "step": 5093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.849836359108419e-06,
+      "loss": 0.6987,
+      "step": 5094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8464546123746608e-06,
+      "loss": 0.743,
+      "step": 5095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8430745403148586e-06,
+      "loss": 0.6755,
+      "step": 5096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.839696143720314e-06,
+      "loss": 0.7501,
+      "step": 5097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.836319423381916e-06,
+      "loss": 0.7118,
+      "step": 5098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.832944380090168e-06,
+      "loss": 0.7369,
+      "step": 5099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829571014635184e-06,
+      "loss": 0.6886,
+      "step": 5100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8261993278066836e-06,
+      "loss": 0.679,
+      "step": 5101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.822829320393985e-06,
+      "loss": 0.7889,
+      "step": 5102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.819460993186032e-06,
+      "loss": 0.7039,
+      "step": 5103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816094346971352e-06,
+      "loss": 0.6319,
+      "step": 5104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.812729382538102e-06,
+      "loss": 0.7291,
+      "step": 5105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8093661006740235e-06,
+      "loss": 0.6375,
+      "step": 5106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8060045021664804e-06,
+      "loss": 0.7144,
+      "step": 5107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8026445878024343e-06,
+      "loss": 0.7676,
+      "step": 5108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.799286358368456e-06,
+      "loss": 0.7756,
+      "step": 5109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7959298146507197e-06,
+      "loss": 0.6367,
+      "step": 5110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7925749574350093e-06,
+      "loss": 0.3656,
+      "step": 5111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7892217875067017e-06,
+      "loss": 0.7513,
+      "step": 5112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7858703056507983e-06,
+      "loss": 0.7483,
+      "step": 5113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.782520512651885e-06,
+      "loss": 0.6786,
+      "step": 5114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7791724092941718e-06,
+      "loss": 0.6878,
+      "step": 5115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7758259963614552e-06,
+      "loss": 0.7178,
+      "step": 5116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7724812746371464e-06,
+      "loss": 0.7246,
+      "step": 5117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.769138244904258e-06,
+      "loss": 0.6869,
+      "step": 5118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7657969079454118e-06,
+      "loss": 0.7106,
+      "step": 5119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.762457264542817e-06,
+      "loss": 0.7145,
+      "step": 5120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7591193154783113e-06,
+      "loss": 0.6901,
+      "step": 5121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7557830615333125e-06,
+      "loss": 0.3596,
+      "step": 5122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7524485034888536e-06,
+      "loss": 0.6841,
+      "step": 5123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7491156421255695e-06,
+      "loss": 0.7524,
+      "step": 5124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7457844782236953e-06,
+      "loss": 0.7486,
+      "step": 5125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.742455012563071e-06,
+      "loss": 0.6482,
+      "step": 5126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7391272459231365e-06,
+      "loss": 0.6651,
+      "step": 5127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7358011790829362e-06,
+      "loss": 0.7966,
+      "step": 5128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.73247681282112e-06,
+      "loss": 0.6962,
+      "step": 5129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729154147915928e-06,
+      "loss": 0.3104,
+      "step": 5130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7258331851452126e-06,
+      "loss": 0.6648,
+      "step": 5131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7225139252864265e-06,
+      "loss": 0.7523,
+      "step": 5132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719196369116619e-06,
+      "loss": 0.6368,
+      "step": 5133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7158805174124445e-06,
+      "loss": 0.7335,
+      "step": 5134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7125663709501595e-06,
+      "loss": 0.7109,
+      "step": 5135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7092539305056185e-06,
+      "loss": 0.777,
+      "step": 5136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.705943196854275e-06,
+      "loss": 0.6167,
+      "step": 5137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7026341707711844e-06,
+      "loss": 0.6406,
+      "step": 5138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6993268530310057e-06,
+      "loss": 0.669,
+      "step": 5139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6960212444079947e-06,
+      "loss": 0.6894,
+      "step": 5140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.692717345676008e-06,
+      "loss": 0.6972,
+      "step": 5141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6894151576085003e-06,
+      "loss": 0.6792,
+      "step": 5142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.68611468097853e-06,
+      "loss": 0.7111,
+      "step": 5143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.682815916558752e-06,
+      "loss": 0.7896,
+      "step": 5144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6795188651214133e-06,
+      "loss": 0.7321,
+      "step": 5145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6762235274383775e-06,
+      "loss": 0.7474,
+      "step": 5146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6729299042810895e-06,
+      "loss": 0.3596,
+      "step": 5147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.669637996420601e-06,
+      "loss": 0.7639,
+      "step": 5148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6663478046275624e-06,
+      "loss": 0.765,
+      "step": 5149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6630593296722207e-06,
+      "loss": 0.7108,
+      "step": 5150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6597725723244195e-06,
+      "loss": 0.6882,
+      "step": 5151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6564875333536078e-06,
+      "loss": 0.6691,
+      "step": 5152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.653204213528815e-06,
+      "loss": 0.7699,
+      "step": 5153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6499226136186928e-06,
+      "loss": 0.7469,
+      "step": 5154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.646642734391468e-06,
+      "loss": 0.6933,
+      "step": 5155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.643364576614976e-06,
+      "loss": 0.6905,
+      "step": 5156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6400881410566483e-06,
+      "loss": 0.7203,
+      "step": 5157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.63681342848351e-06,
+      "loss": 0.7077,
+      "step": 5158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6335404396621855e-06,
+      "loss": 0.7872,
+      "step": 5159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.630269175358898e-06,
+      "loss": 0.7353,
+      "step": 5160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6269996363394535e-06,
+      "loss": 0.3267,
+      "step": 5161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.623731823369278e-06,
+      "loss": 0.7377,
+      "step": 5162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.620465737213367e-06,
+      "loss": 0.678,
+      "step": 5163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.617201378636335e-06,
+      "loss": 0.7663,
+      "step": 5164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.613938748402375e-06,
+      "loss": 0.7182,
+      "step": 5165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6106778472752836e-06,
+      "loss": 0.7772,
+      "step": 5166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6074186760184516e-06,
+      "loss": 0.7577,
+      "step": 5167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6041612353948644e-06,
+      "loss": 0.6993,
+      "step": 5168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6009055261671e-06,
+      "loss": 0.7332,
+      "step": 5169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5976515490973385e-06,
+      "loss": 0.7161,
+      "step": 5170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5943993049473404e-06,
+      "loss": 0.6804,
+      "step": 5171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5911487944784786e-06,
+      "loss": 0.7006,
+      "step": 5172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5879000184517033e-06,
+      "loss": 0.675,
+      "step": 5173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5846529776275697e-06,
+      "loss": 0.7293,
+      "step": 5174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.581407672766223e-06,
+      "loss": 0.7205,
+      "step": 5175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5781641046274006e-06,
+      "loss": 0.7016,
+      "step": 5176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.574922273970437e-06,
+      "loss": 0.7013,
+      "step": 5177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.57168218155426e-06,
+      "loss": 0.363,
+      "step": 5178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5684438281373803e-06,
+      "loss": 0.7836,
+      "step": 5179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5652072144779194e-06,
+      "loss": 0.7241,
+      "step": 5180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.561972341333572e-06,
+      "loss": 0.6555,
+      "step": 5181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.558739209461646e-06,
+      "loss": 0.7401,
+      "step": 5182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.555507819619021e-06,
+      "loss": 0.7192,
+      "step": 5183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5522781725621814e-06,
+      "loss": 0.3287,
+      "step": 5184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.549050269047203e-06,
+      "loss": 0.7402,
+      "step": 5185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545824109829752e-06,
+      "loss": 0.6915,
+      "step": 5186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5425996956650766e-06,
+      "loss": 0.6852,
+      "step": 5187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5393770273080375e-06,
+      "loss": 0.3669,
+      "step": 5188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5361561055130625e-06,
+      "loss": 0.8045,
+      "step": 5189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5329369310341945e-06,
+      "loss": 0.6868,
+      "step": 5190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5297195046250456e-06,
+      "loss": 0.6206,
+      "step": 5191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.526503827038832e-06,
+      "loss": 0.7155,
+      "step": 5192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.523289899028356e-06,
+      "loss": 0.7265,
+      "step": 5193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5200777213460136e-06,
+      "loss": 0.7398,
+      "step": 5194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5168672947437857e-06,
+      "loss": 0.6854,
+      "step": 5195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5136586199732503e-06,
+      "loss": 0.7745,
+      "step": 5196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5104516977855633e-06,
+      "loss": 0.7509,
+      "step": 5197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5072465289314886e-06,
+      "loss": 0.338,
+      "step": 5198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5040431141613598e-06,
+      "loss": 0.6466,
+      "step": 5199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5008414542251148e-06,
+      "loss": 0.6866,
+      "step": 5200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4976415498722727e-06,
+      "loss": 0.7679,
+      "step": 5201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494443401851945e-06,
+      "loss": 0.3462,
+      "step": 5202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4912470109128305e-06,
+      "loss": 0.7398,
+      "step": 5203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4880523778032217e-06,
+      "loss": 0.7316,
+      "step": 5204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4848595032709856e-06,
+      "loss": 0.6326,
+      "step": 5205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481668388063598e-06,
+      "loss": 0.7318,
+      "step": 5206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478479032928104e-06,
+      "loss": 0.7394,
+      "step": 5207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.475291438611147e-06,
+      "loss": 0.7188,
+      "step": 5208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4721056058589565e-06,
+      "loss": 0.7552,
+      "step": 5209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.468921535417348e-06,
+      "loss": 0.7219,
+      "step": 5210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.465739228031726e-06,
+      "loss": 0.678,
+      "step": 5211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4625586844470805e-06,
+      "loss": 0.5857,
+      "step": 5212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4593799054079903e-06,
+      "loss": 0.7692,
+      "step": 5213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4562028916586243e-06,
+      "loss": 0.6769,
+      "step": 5214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.453027643942727e-06,
+      "loss": 0.6355,
+      "step": 5215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.44985416300364e-06,
+      "loss": 0.6317,
+      "step": 5216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.446682449584289e-06,
+      "loss": 0.7376,
+      "step": 5217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4435125044271833e-06,
+      "loss": 0.7368,
+      "step": 5218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440344328274421e-06,
+      "loss": 0.7125,
+      "step": 5219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.437177921867686e-06,
+      "loss": 0.6527,
+      "step": 5220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4340132859482467e-06,
+      "loss": 0.7028,
+      "step": 5221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.430850421256955e-06,
+      "loss": 0.6843,
+      "step": 5222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4276893285342507e-06,
+      "loss": 0.7746,
+      "step": 5223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4245300085201594e-06,
+      "loss": 0.7447,
+      "step": 5224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42137246195429e-06,
+      "loss": 0.6749,
+      "step": 5225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4182166895758387e-06,
+      "loss": 0.7507,
+      "step": 5226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.415062692123582e-06,
+      "loss": 0.7124,
+      "step": 5227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4119104703358865e-06,
+      "loss": 0.7441,
+      "step": 5228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4087600249507004e-06,
+      "loss": 0.6909,
+      "step": 5229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4056113567055484e-06,
+      "loss": 0.7731,
+      "step": 5230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4024644663375574e-06,
+      "loss": 0.7033,
+      "step": 5231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3993193545834182e-06,
+      "loss": 0.7062,
+      "step": 5232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.396176022179418e-06,
+      "loss": 0.7818,
+      "step": 5233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.393034469861424e-06,
+      "loss": 0.6849,
+      "step": 5234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.389894698364885e-06,
+      "loss": 0.728,
+      "step": 5235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3867567084248356e-06,
+      "loss": 0.739,
+      "step": 5236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.383620500775894e-06,
+      "loss": 0.7089,
+      "step": 5237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.380486076152251e-06,
+      "loss": 0.7375,
+      "step": 5238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.377353435287698e-06,
+      "loss": 0.6914,
+      "step": 5239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3742225789155927e-06,
+      "loss": 0.6946,
+      "step": 5240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.371093507768882e-06,
+      "loss": 0.7241,
+      "step": 5241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.367966222580097e-06,
+      "loss": 0.7215,
+      "step": 5242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364840724081344e-06,
+      "loss": 0.7172,
+      "step": 5243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3617170130043177e-06,
+      "loss": 0.6376,
+      "step": 5244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3585950900802936e-06,
+      "loss": 0.6674,
+      "step": 5245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3554749560401167e-06,
+      "loss": 0.3256,
+      "step": 5246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352356611614237e-06,
+      "loss": 0.6887,
+      "step": 5247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3492400575326578e-06,
+      "loss": 0.6475,
+      "step": 5248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.34612529452499e-06,
+      "loss": 0.4082,
+      "step": 5249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3430123233204027e-06,
+      "loss": 0.6547,
+      "step": 5250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.339901144647658e-06,
+      "loss": 0.7343,
+      "step": 5251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336791759235096e-06,
+      "loss": 0.7245,
+      "step": 5252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.333684167810637e-06,
+      "loss": 0.7385,
+      "step": 5253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3305783711017792e-06,
+      "loss": 0.3526,
+      "step": 5254
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.327474369835605e-06,
+      "loss": 0.722,
+      "step": 5255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3243721647387673e-06,
+      "loss": 0.748,
+      "step": 5256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.321271756537514e-06,
+      "loss": 0.7298,
+      "step": 5257
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.318173145957655e-06,
+      "loss": 0.7968,
+      "step": 5258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.315076333724592e-06,
+      "loss": 0.737,
+      "step": 5259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3119813205633003e-06,
+      "loss": 0.6752,
+      "step": 5260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.308888107198334e-06,
+      "loss": 0.7139,
+      "step": 5261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.305796694353829e-06,
+      "loss": 0.6939,
+      "step": 5262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.302707082753498e-06,
+      "loss": 0.784,
+      "step": 5263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.299619273120626e-06,
+      "loss": 0.7352,
+      "step": 5264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2965332661780905e-06,
+      "loss": 0.6719,
+      "step": 5265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2934490626483273e-06,
+      "loss": 0.3216,
+      "step": 5266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290366663253374e-06,
+      "loss": 0.7049,
+      "step": 5267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2872860687148225e-06,
+      "loss": 0.6671,
+      "step": 5268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.284207279753855e-06,
+      "loss": 0.6178,
+      "step": 5269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2811302970912307e-06,
+      "loss": 0.7515,
+      "step": 5270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2780551214472846e-06,
+      "loss": 0.7846,
+      "step": 5271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2749817535419207e-06,
+      "loss": 0.66,
+      "step": 5272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2719101940946366e-06,
+      "loss": 0.739,
+      "step": 5273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.268840443824486e-06,
+      "loss": 0.8299,
+      "step": 5274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265772503450122e-06,
+      "loss": 0.797,
+      "step": 5275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262706373689754e-06,
+      "loss": 0.7253,
+      "step": 5276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.259642055261175e-06,
+      "loss": 0.7444,
+      "step": 5277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2565795488817575e-06,
+      "loss": 0.6488,
+      "step": 5278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2535188552684463e-06,
+      "loss": 0.683,
+      "step": 5279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.250459975137761e-06,
+      "loss": 0.7493,
+      "step": 5280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2474029092058004e-06,
+      "loss": 0.7431,
+      "step": 5281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2443476581882295e-06,
+      "loss": 0.791,
+      "step": 5282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2412942228003043e-06,
+      "loss": 0.7347,
+      "step": 5283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2382426037568384e-06,
+      "loss": 0.7341,
+      "step": 5284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2351928017722326e-06,
+      "loss": 0.7262,
+      "step": 5285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2321448175604555e-06,
+      "loss": 0.8328,
+      "step": 5286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229098651835053e-06,
+      "loss": 0.6603,
+      "step": 5287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2260543053091466e-06,
+      "loss": 0.7049,
+      "step": 5288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.223011778695432e-06,
+      "loss": 0.7065,
+      "step": 5289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2199710727061684e-06,
+      "loss": 0.6878,
+      "step": 5290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.216932188053208e-06,
+      "loss": 0.7049,
+      "step": 5291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.213895125447959e-06,
+      "loss": 0.6962,
+      "step": 5292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.210859885601412e-06,
+      "loss": 0.6297,
+      "step": 5293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207826469224128e-06,
+      "loss": 0.6293,
+      "step": 5294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.204794877026244e-06,
+      "loss": 0.698,
+      "step": 5295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2017651097174676e-06,
+      "loss": 0.7196,
+      "step": 5296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.198737168007079e-06,
+      "loss": 0.7939,
+      "step": 5297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.195711052603934e-06,
+      "loss": 0.7163,
+      "step": 5298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.192686764216454e-06,
+      "loss": 0.6981,
+      "step": 5299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1896643035526388e-06,
+      "loss": 0.7823,
+      "step": 5300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18664367132006e-06,
+      "loss": 0.7005,
+      "step": 5301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1836248682258585e-06,
+      "loss": 0.7512,
+      "step": 5302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1806078949767497e-06,
+      "loss": 0.7057,
+      "step": 5303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1775927522790174e-06,
+      "loss": 0.7589,
+      "step": 5304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1745794408385214e-06,
+      "loss": 0.7402,
+      "step": 5305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1715679613606897e-06,
+      "loss": 0.7606,
+      "step": 5306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1685583145505194e-06,
+      "loss": 0.7419,
+      "step": 5307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165550501112582e-06,
+      "loss": 0.7393,
+      "step": 5308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.162544521751019e-06,
+      "loss": 0.7339,
+      "step": 5309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1595403771695435e-06,
+      "loss": 0.6897,
+      "step": 5310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1565380680714354e-06,
+      "loss": 0.6882,
+      "step": 5311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1535375951595525e-06,
+      "loss": 0.7205,
+      "step": 5312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150538959136309e-06,
+      "loss": 0.3454,
+      "step": 5313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.147542160703707e-06,
+      "loss": 0.7183,
+      "step": 5314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1445472005633016e-06,
+      "loss": 0.7777,
+      "step": 5315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1415540794162328e-06,
+      "loss": 0.7915,
+      "step": 5316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1385627979631952e-06,
+      "loss": 0.7126,
+      "step": 5317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1355733569044633e-06,
+      "loss": 0.7265,
+      "step": 5318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1325857569398768e-06,
+      "loss": 0.6926,
+      "step": 5319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1295999987688456e-06,
+      "loss": 0.5959,
+      "step": 5320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.126616083090347e-06,
+      "loss": 0.6591,
+      "step": 5321
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1236340106029297e-06,
+      "loss": 0.7101,
+      "step": 5322
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1206537820047037e-06,
+      "loss": 0.7195,
+      "step": 5323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1176753979933597e-06,
+      "loss": 0.7247,
+      "step": 5324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1146988592661446e-06,
+      "loss": 0.7082,
+      "step": 5325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1117241665198786e-06,
+      "loss": 0.7505,
+      "step": 5326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1087513204509502e-06,
+      "loss": 0.7356,
+      "step": 5327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1057803217553153e-06,
+      "loss": 0.7636,
+      "step": 5328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102811171128496e-06,
+      "loss": 0.7156,
+      "step": 5329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0998438692655855e-06,
+      "loss": 0.3141,
+      "step": 5330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0968784168612333e-06,
+      "loss": 0.559,
+      "step": 5331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0939148146096733e-06,
+      "loss": 0.7225,
+      "step": 5332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.090953063204688e-06,
+      "loss": 0.7488,
+      "step": 5333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0879931633396446e-06,
+      "loss": 0.6807,
+      "step": 5334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08503511570746e-06,
+      "loss": 0.7224,
+      "step": 5335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0820789210006285e-06,
+      "loss": 0.6418,
+      "step": 5336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.079124579911206e-06,
+      "loss": 0.6717,
+      "step": 5337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.076172093130817e-06,
+      "loss": 0.8408,
+      "step": 5338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0732214613506498e-06,
+      "loss": 0.7499,
+      "step": 5339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0702726852614628e-06,
+      "loss": 0.6984,
+      "step": 5340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0673257655535672e-06,
+      "loss": 0.6788,
+      "step": 5341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0643807029168604e-06,
+      "loss": 0.8205,
+      "step": 5342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0614374980407848e-06,
+      "loss": 0.7825,
+      "step": 5343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0584961516143598e-06,
+      "loss": 0.6668,
+      "step": 5344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.055556664326167e-06,
+      "loss": 0.7395,
+      "step": 5345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.052619036864352e-06,
+      "loss": 0.7256,
+      "step": 5346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0496832699166237e-06,
+      "loss": 0.682,
+      "step": 5347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.046749364170262e-06,
+      "loss": 0.6636,
+      "step": 5348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043817320312097e-06,
+      "loss": 0.7626,
+      "step": 5349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0408871390285423e-06,
+      "loss": 0.8292,
+      "step": 5350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0379588210055555e-06,
+      "loss": 0.6392,
+      "step": 5351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0350323669286775e-06,
+      "loss": 0.6485,
+      "step": 5352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0321077774829957e-06,
+      "loss": 0.7485,
+      "step": 5353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.029185053353171e-06,
+      "loss": 0.6842,
+      "step": 5354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.026264195223425e-06,
+      "loss": 0.7316,
+      "step": 5355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0233452037775427e-06,
+      "loss": 0.7282,
+      "step": 5356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0204280796988696e-06,
+      "loss": 0.7236,
+      "step": 5357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.017512823670322e-06,
+      "loss": 0.7712,
+      "step": 5358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.014599436374364e-06,
+      "loss": 0.7505,
+      "step": 5359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0116879184930403e-06,
+      "loss": 0.7849,
+      "step": 5360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.008778270707944e-06,
+      "loss": 0.7345,
+      "step": 5361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.005870493700237e-06,
+      "loss": 0.6829,
+      "step": 5362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0029645881506397e-06,
+      "loss": 0.7508,
+      "step": 5363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.000060554739439e-06,
+      "loss": 0.6191,
+      "step": 5364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997158394146481e-06,
+      "loss": 0.6722,
+      "step": 5365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9942581070511734e-06,
+      "loss": 0.7788,
+      "step": 5366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9913596941324796e-06,
+      "loss": 0.7327,
+      "step": 5367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988463156068938e-06,
+      "loss": 0.6298,
+      "step": 5368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9855684935386355e-06,
+      "loss": 0.7131,
+      "step": 5369
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9826757072192237e-06,
+      "loss": 0.6999,
+      "step": 5370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.979784797787916e-06,
+      "loss": 0.7303,
+      "step": 5371
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9768957659214883e-06,
+      "loss": 0.735,
+      "step": 5372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.974008612296272e-06,
+      "loss": 0.676,
+      "step": 5373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9711233375881656e-06,
+      "loss": 0.7364,
+      "step": 5374
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9682399424726184e-06,
+      "loss": 0.6978,
+      "step": 5375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9653584276246473e-06,
+      "loss": 0.6882,
+      "step": 5376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962478793718826e-06,
+      "loss": 0.6979,
+      "step": 5377
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9596010414292887e-06,
+      "loss": 0.6994,
+      "step": 5378
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956725171429729e-06,
+      "loss": 0.6698,
+      "step": 5379
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9538511843933994e-06,
+      "loss": 0.6833,
+      "step": 5380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9509790809931126e-06,
+      "loss": 0.7289,
+      "step": 5381
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94810886190124e-06,
+      "loss": 0.6229,
+      "step": 5382
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9452405277897136e-06,
+      "loss": 0.7046,
+      "step": 5383
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9423740793300162e-06,
+      "loss": 0.5825,
+      "step": 5384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9395095171931976e-06,
+      "loss": 0.7473,
+      "step": 5385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936646842049864e-06,
+      "loss": 0.8077,
+      "step": 5386
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.933786054570179e-06,
+      "loss": 0.6952,
+      "step": 5387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9309271554238662e-06,
+      "loss": 0.6911,
+      "step": 5388
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.928070145280202e-06,
+      "loss": 0.7754,
+      "step": 5389
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925215024808027e-06,
+      "loss": 0.3432,
+      "step": 5390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9223617946757376e-06,
+      "loss": 0.6579,
+      "step": 5391
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9195104555512824e-06,
+      "loss": 0.7054,
+      "step": 5392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916661008102174e-06,
+      "loss": 0.6945,
+      "step": 5393
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.913813452995478e-06,
+      "loss": 0.7479,
+      "step": 5394
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9109677908978196e-06,
+      "loss": 0.3454,
+      "step": 5395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9081240224753804e-06,
+      "loss": 0.3427,
+      "step": 5396
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9052821483939e-06,
+      "loss": 0.6292,
+      "step": 5397
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9024421693186646e-06,
+      "loss": 0.8465,
+      "step": 5398
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8996040859145358e-06,
+      "loss": 0.7159,
+      "step": 5399
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8967678988459104e-06,
+      "loss": 0.7516,
+      "step": 5400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8939336087767613e-06,
+      "loss": 0.6675,
+      "step": 5401
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8911012163705989e-06,
+      "loss": 0.7165,
+      "step": 5402
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8882707222905006e-06,
+      "loss": 0.7009,
+      "step": 5403
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8854421271990964e-06,
+      "loss": 0.7327,
+      "step": 5404
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8826154317585722e-06,
+      "loss": 0.7473,
+      "step": 5405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8797906366306695e-06,
+      "loss": 0.7454,
+      "step": 5406
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8769677424766863e-06,
+      "loss": 0.7747,
+      "step": 5407
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8741467499574651e-06,
+      "loss": 0.6046,
+      "step": 5408
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8713276597334228e-06,
+      "loss": 0.7617,
+      "step": 5409
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.868510472464512e-06,
+      "loss": 0.7395,
+      "step": 5410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8656951888102504e-06,
+      "loss": 0.6757,
+      "step": 5411
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8628818094297063e-06,
+      "loss": 0.6556,
+      "step": 5412
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8600703349815053e-06,
+      "loss": 0.7254,
+      "step": 5413
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8572607661238228e-06,
+      "loss": 0.7395,
+      "step": 5414
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8544531035143932e-06,
+      "loss": 0.7104,
+      "step": 5415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8516473478104957e-06,
+      "loss": 0.7291,
+      "step": 5416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8488434996689774e-06,
+      "loss": 0.7072,
+      "step": 5417
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.84604155974622e-06,
+      "loss": 0.6826,
+      "step": 5418
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.843241528698182e-06,
+      "loss": 0.6405,
+      "step": 5419
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8404434071803511e-06,
+      "loss": 0.7197,
+      "step": 5420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8376471958477838e-06,
+      "loss": 0.7797,
+      "step": 5421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8348528953550837e-06,
+      "loss": 0.7725,
+      "step": 5422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8320605063564068e-06,
+      "loss": 0.7621,
+      "step": 5423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8292700295054644e-06,
+      "loss": 0.7761,
+      "step": 5424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8264814654555197e-06,
+      "loss": 0.6991,
+      "step": 5425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8236948148593802e-06,
+      "loss": 0.7289,
+      "step": 5426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.820910078369421e-06,
+      "loss": 0.7336,
+      "step": 5427
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.818127256637554e-06,
+      "loss": 0.7551,
+      "step": 5428
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8153463503152513e-06,
+      "loss": 0.7932,
+      "step": 5429
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8125673600535331e-06,
+      "loss": 0.6878,
+      "step": 5430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8097902865029738e-06,
+      "loss": 0.6938,
+      "step": 5431
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8070151303136974e-06,
+      "loss": 0.7382,
+      "step": 5432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80424189213538e-06,
+      "loss": 0.7291,
+      "step": 5433
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8014705726172433e-06,
+      "loss": 0.6884,
+      "step": 5434
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7987011724080717e-06,
+      "loss": 0.687,
+      "step": 5435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7959336921561832e-06,
+      "loss": 0.689,
+      "step": 5436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7931681325094686e-06,
+      "loss": 0.7058,
+      "step": 5437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7904044941153464e-06,
+      "loss": 0.711,
+      "step": 5438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7876427776208005e-06,
+      "loss": 0.7435,
+      "step": 5439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.784882983672358e-06,
+      "loss": 0.7686,
+      "step": 5440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7821251129160978e-06,
+      "loss": 0.7596,
+      "step": 5441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7793691659976497e-06,
+      "loss": 0.7788,
+      "step": 5442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7766151435621949e-06,
+      "loss": 0.7069,
+      "step": 5443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7738630462544515e-06,
+      "loss": 0.6766,
+      "step": 5444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.771112874718709e-06,
+      "loss": 0.7455,
+      "step": 5445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7683646295987856e-06,
+      "loss": 0.7358,
+      "step": 5446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7656183115380577e-06,
+      "loss": 0.727,
+      "step": 5447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.762873921179452e-06,
+      "loss": 0.6518,
+      "step": 5448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7601314591654384e-06,
+      "loss": 0.6596,
+      "step": 5449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7573909261380405e-06,
+      "loss": 0.7332,
+      "step": 5450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7546523227388312e-06,
+      "loss": 0.7213,
+      "step": 5451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7519156496089185e-06,
+      "loss": 0.6942,
+      "step": 5452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.749180907388982e-06,
+      "loss": 0.7538,
+      "step": 5453
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7464480967192255e-06,
+      "loss": 0.629,
+      "step": 5454
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7437172182394146e-06,
+      "loss": 0.7632,
+      "step": 5455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7409882725888594e-06,
+      "loss": 0.7066,
+      "step": 5456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7382612604064174e-06,
+      "loss": 0.7746,
+      "step": 5457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7355361823304916e-06,
+      "loss": 0.7163,
+      "step": 5458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7328130389990383e-06,
+      "loss": 0.7318,
+      "step": 5459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7300918310495507e-06,
+      "loss": 0.7261,
+      "step": 5460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.727372559119077e-06,
+      "loss": 0.707,
+      "step": 5461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7246552238442094e-06,
+      "loss": 0.6964,
+      "step": 5462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7219398258610864e-06,
+      "loss": 0.7435,
+      "step": 5463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7192263658053964e-06,
+      "loss": 0.3468,
+      "step": 5464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.716514844312368e-06,
+      "loss": 0.72,
+      "step": 5465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7138052620167823e-06,
+      "loss": 0.7088,
+      "step": 5466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7110976195529617e-06,
+      "loss": 0.7902,
+      "step": 5467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708391917554779e-06,
+      "loss": 0.7218,
+      "step": 5468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056881566556461e-06,
+      "loss": 0.6834,
+      "step": 5469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029863374885258e-06,
+      "loss": 0.7378,
+      "step": 5470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7002864606859248e-06,
+      "loss": 0.6211,
+      "step": 5471
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6975885268798965e-06,
+      "loss": 0.3636,
+      "step": 5472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6948925367020363e-06,
+      "loss": 0.7387,
+      "step": 5473
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.69219849078349e-06,
+      "loss": 0.6891,
+      "step": 5474
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6895063897549368e-06,
+      "loss": 0.8006,
+      "step": 5475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6868162342466188e-06,
+      "loss": 0.7446,
+      "step": 5476
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6841280248883052e-06,
+      "loss": 0.7507,
+      "step": 5477
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6814417623093182e-06,
+      "loss": 0.653,
+      "step": 5478
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6787574471385238e-06,
+      "loss": 0.6983,
+      "step": 5479
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6760750800043302e-06,
+      "loss": 0.8102,
+      "step": 5480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6733946615346908e-06,
+      "loss": 0.7616,
+      "step": 5481
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6707161923571047e-06,
+      "loss": 0.7052,
+      "step": 5482
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6680396730986058e-06,
+      "loss": 0.7711,
+      "step": 5483
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6653651043857867e-06,
+      "loss": 0.7397,
+      "step": 5484
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6626924868447646e-06,
+      "loss": 0.6904,
+      "step": 5485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.660021821101222e-06,
+      "loss": 0.6487,
+      "step": 5486
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6573531077803639e-06,
+      "loss": 0.6504,
+      "step": 5487
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6546863475069496e-06,
+      "loss": 0.7205,
+      "step": 5488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6520215409052775e-06,
+      "loss": 0.6984,
+      "step": 5489
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6493586885991908e-06,
+      "loss": 0.6352,
+      "step": 5490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6466977912120742e-06,
+      "loss": 0.7048,
+      "step": 5491
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6440388493668568e-06,
+      "loss": 0.7195,
+      "step": 5492
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.641381863685999e-06,
+      "loss": 0.7454,
+      "step": 5493
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6387268347915241e-06,
+      "loss": 0.7479,
+      "step": 5494
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.636073763304975e-06,
+      "loss": 0.6529,
+      "step": 5495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6334226498474514e-06,
+      "loss": 0.699,
+      "step": 5496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6307734950395882e-06,
+      "loss": 0.7701,
+      "step": 5497
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6281262995015635e-06,
+      "loss": 0.7564,
+      "step": 5498
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.625481063853097e-06,
+      "loss": 0.6537,
+      "step": 5499
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6228377887134516e-06,
+      "loss": 0.7331,
+      "step": 5500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6201964747014209e-06,
+      "loss": 0.7155,
+      "step": 5501
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.617557122435357e-06,
+      "loss": 0.6901,
+      "step": 5502
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6149197325331333e-06,
+      "loss": 0.3537,
+      "step": 5503
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6122843056121828e-06,
+      "loss": 0.7712,
+      "step": 5504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6096508422894642e-06,
+      "loss": 0.6477,
+      "step": 5505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6070193431814817e-06,
+      "loss": 0.7618,
+      "step": 5506
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6043898089042809e-06,
+      "loss": 0.6934,
+      "step": 5507
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6017622400734478e-06,
+      "loss": 0.6898,
+      "step": 5508
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.599136637304105e-06,
+      "loss": 0.7572,
+      "step": 5509
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5965130012109209e-06,
+      "loss": 0.7636,
+      "step": 5510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5938913324080897e-06,
+      "loss": 0.7136,
+      "step": 5511
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5912716315093669e-06,
+      "loss": 0.7108,
+      "step": 5512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5886538991280266e-06,
+      "loss": 0.7565,
+      "step": 5513
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5860381358768929e-06,
+      "loss": 0.6806,
+      "step": 5514
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.583424342368326e-06,
+      "loss": 0.7136,
+      "step": 5515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5808125192142277e-06,
+      "loss": 0.6356,
+      "step": 5516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5782026670260342e-06,
+      "loss": 0.8301,
+      "step": 5517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.575594786414726e-06,
+      "loss": 0.6618,
+      "step": 5518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5729888779908109e-06,
+      "loss": 0.7088,
+      "step": 5519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5703849423643525e-06,
+      "loss": 0.7473,
+      "step": 5520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.567782980144933e-06,
+      "loss": 0.7346,
+      "step": 5521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5651829919416917e-06,
+      "loss": 0.7354,
+      "step": 5522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5625849783632897e-06,
+      "loss": 0.7206,
+      "step": 5523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5599889400179336e-06,
+      "loss": 0.724,
+      "step": 5524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5573948775133685e-06,
+      "loss": 0.7207,
+      "step": 5525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.554802791456873e-06,
+      "loss": 0.7528,
+      "step": 5526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5522126824552664e-06,
+      "loss": 0.7705,
+      "step": 5527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5496245511149044e-06,
+      "loss": 0.7576,
+      "step": 5528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5470383980416748e-06,
+      "loss": 0.7502,
+      "step": 5529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5444542238410131e-06,
+      "loss": 0.6993,
+      "step": 5530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5418720291178791e-06,
+      "loss": 0.6781,
+      "step": 5531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5392918144767766e-06,
+      "loss": 0.7274,
+      "step": 5532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.536713580521746e-06,
+      "loss": 0.6657,
+      "step": 5533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.534137327856361e-06,
+      "loss": 0.7054,
+      "step": 5534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5315630570837326e-06,
+      "loss": 0.6786,
+      "step": 5535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.528990768806511e-06,
+      "loss": 0.7468,
+      "step": 5536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5264204636268742e-06,
+      "loss": 0.7423,
+      "step": 5537
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5238521421465447e-06,
+      "loss": 0.6946,
+      "step": 5538
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5212858049667745e-06,
+      "loss": 0.322,
+      "step": 5539
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5187214526883564e-06,
+      "loss": 0.6341,
+      "step": 5540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5161590859116125e-06,
+      "loss": 0.6216,
+      "step": 5541
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5135987052364065e-06,
+      "loss": 0.7579,
+      "step": 5542
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.511040311262132e-06,
+      "loss": 0.7519,
+      "step": 5543
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.508483904587722e-06,
+      "loss": 0.6369,
+      "step": 5544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5059294858116369e-06,
+      "loss": 0.7094,
+      "step": 5545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5033770555318795e-06,
+      "loss": 0.6344,
+      "step": 5546
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5008266143459825e-06,
+      "loss": 0.7006,
+      "step": 5547
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4982781628510169e-06,
+      "loss": 0.6103,
+      "step": 5548
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4957317016435825e-06,
+      "loss": 0.7544,
+      "step": 5549
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4931872313198192e-06,
+      "loss": 0.7979,
+      "step": 5550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4906447524753986e-06,
+      "loss": 0.7376,
+      "step": 5551
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4881042657055188e-06,
+      "loss": 0.3343,
+      "step": 5552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4855657716049266e-06,
+      "loss": 0.7289,
+      "step": 5553
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4830292707678874e-06,
+      "loss": 0.7702,
+      "step": 5554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4804947637882084e-06,
+      "loss": 0.7595,
+      "step": 5555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4779622512592283e-06,
+      "loss": 0.72,
+      "step": 5556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.475431733773819e-06,
+      "loss": 0.721,
+      "step": 5557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4729032119243836e-06,
+      "loss": 0.6853,
+      "step": 5558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4703766863028624e-06,
+      "loss": 0.7226,
+      "step": 5559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4678521575007175e-06,
+      "loss": 0.705,
+      "step": 5560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4653296261089622e-06,
+      "loss": 0.7543,
+      "step": 5561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4628090927181227e-06,
+      "loss": 0.6975,
+      "step": 5562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4602905579182692e-06,
+      "loss": 0.6742,
+      "step": 5563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4577740222990012e-06,
+      "loss": 0.7502,
+      "step": 5564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4552594864494495e-06,
+      "loss": 0.6917,
+      "step": 5565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.452746950958277e-06,
+      "loss": 0.3386,
+      "step": 5566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4502364164136816e-06,
+      "loss": 0.74,
+      "step": 5567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4477278834033814e-06,
+      "loss": 0.6892,
+      "step": 5568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4452213525146442e-06,
+      "loss": 0.804,
+      "step": 5569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4427168243342494e-06,
+      "loss": 0.7238,
+      "step": 5570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.440214299448527e-06,
+      "loss": 0.7737,
+      "step": 5571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4377137784433203e-06,
+      "loss": 0.3432,
+      "step": 5572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.435215261904015e-06,
+      "loss": 0.6657,
+      "step": 5573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4327187504155227e-06,
+      "loss": 0.5966,
+      "step": 5574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.430224244562287e-06,
+      "loss": 0.7671,
+      "step": 5575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4277317449282834e-06,
+      "loss": 0.6862,
+      "step": 5576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.425241252097016e-06,
+      "loss": 0.7434,
+      "step": 5577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.422752766651515e-06,
+      "loss": 0.3509,
+      "step": 5578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4202662891743524e-06,
+      "loss": 0.724,
+      "step": 5579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4177818202476156e-06,
+      "loss": 0.6613,
+      "step": 5580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4152993604529318e-06,
+      "loss": 0.7375,
+      "step": 5581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4128189103714551e-06,
+      "loss": 0.7334,
+      "step": 5582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4103404705838686e-06,
+      "loss": 0.7295,
+      "step": 5583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4078640416703849e-06,
+      "loss": 0.7275,
+      "step": 5584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.405389624210749e-06,
+      "loss": 0.7928,
+      "step": 5585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4029172187842243e-06,
+      "loss": 0.7213,
+      "step": 5586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4004468259696214e-06,
+      "loss": 0.7919,
+      "step": 5587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3979784463452584e-06,
+      "loss": 0.7778,
+      "step": 5588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395512080489002e-06,
+      "loss": 0.7,
+      "step": 5589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.393047728978234e-06,
+      "loss": 0.7388,
+      "step": 5590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3905853923898694e-06,
+      "loss": 0.6534,
+      "step": 5591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3881250713003525e-06,
+      "loss": 0.7602,
+      "step": 5592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3856667662856526e-06,
+      "loss": 0.6598,
+      "step": 5593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3832104779212696e-06,
+      "loss": 0.7445,
+      "step": 5594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3807562067822334e-06,
+      "loss": 0.7486,
+      "step": 5595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3783039534430919e-06,
+      "loss": 0.6784,
+      "step": 5596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3758537184779363e-06,
+      "loss": 0.7055,
+      "step": 5597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.37340550246037e-06,
+      "loss": 0.8074,
+      "step": 5598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.370959305963533e-06,
+      "loss": 0.6962,
+      "step": 5599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3685151295600885e-06,
+      "loss": 0.7568,
+      "step": 5600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.366072973822229e-06,
+      "loss": 0.6727,
+      "step": 5601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3636328393216724e-06,
+      "loss": 0.6709,
+      "step": 5602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3611947266296677e-06,
+      "loss": 0.655,
+      "step": 5603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.358758636316979e-06,
+      "loss": 0.7593,
+      "step": 5604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.356324568953914e-06,
+      "loss": 0.7152,
+      "step": 5605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3538925251102885e-06,
+      "loss": 0.7338,
+      "step": 5606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3514625053554643e-06,
+      "loss": 0.6423,
+      "step": 5607
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3490345102583103e-06,
+      "loss": 0.6884,
+      "step": 5608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3466085403872332e-06,
+      "loss": 0.6883,
+      "step": 5609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.344184596310163e-06,
+      "loss": 0.3042,
+      "step": 5610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3417626785945525e-06,
+      "loss": 0.6958,
+      "step": 5611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339342787807385e-06,
+      "loss": 0.7294,
+      "step": 5612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3369249245151682e-06,
+      "loss": 0.7045,
+      "step": 5613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3345090892839262e-06,
+      "loss": 0.7369,
+      "step": 5614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3320952826792255e-06,
+      "loss": 0.7536,
+      "step": 5615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3296835052661417e-06,
+      "loss": 0.7986,
+      "step": 5616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.327273757609283e-06,
+      "loss": 0.6512,
+      "step": 5617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3248660402727809e-06,
+      "loss": 0.7467,
+      "step": 5618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3224603538202929e-06,
+      "loss": 0.6789,
+      "step": 5619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3200566988149987e-06,
+      "loss": 0.7469,
+      "step": 5620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3176550758196072e-06,
+      "loss": 0.3466,
+      "step": 5621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.315255485396343e-06,
+      "loss": 0.6899,
+      "step": 5622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.312857928106962e-06,
+      "loss": 0.7118,
+      "step": 5623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3104624045127423e-06,
+      "loss": 0.752,
+      "step": 5624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3080689151744852e-06,
+      "loss": 0.8049,
+      "step": 5625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.305677460652517e-06,
+      "loss": 0.6902,
+      "step": 5626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3032880415066862e-06,
+      "loss": 0.7042,
+      "step": 5627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3009006582963668e-06,
+      "loss": 0.7561,
+      "step": 5628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.298515311580455e-06,
+      "loss": 0.7324,
+      "step": 5629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2961320019173673e-06,
+      "loss": 0.7586,
+      "step": 5630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2937507298650475e-06,
+      "loss": 0.7003,
+      "step": 5631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.291371495980962e-06,
+      "loss": 0.8007,
+      "step": 5632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2889943008220983e-06,
+      "loss": 0.3215,
+      "step": 5633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2866191449449683e-06,
+      "loss": 0.6884,
+      "step": 5634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2842460289056047e-06,
+      "loss": 0.3126,
+      "step": 5635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2818749532595652e-06,
+      "loss": 0.6794,
+      "step": 5636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.279505918561923e-06,
+      "loss": 0.3273,
+      "step": 5637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2771389253672873e-06,
+      "loss": 0.7715,
+      "step": 5638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2747739742297739e-06,
+      "loss": 0.6989,
+      "step": 5639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2724110657030298e-06,
+      "loss": 0.7647,
+      "step": 5640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2700502003402204e-06,
+      "loss": 0.6525,
+      "step": 5641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.267691378694036e-06,
+      "loss": 0.6788,
+      "step": 5642
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2653346013166845e-06,
+      "loss": 0.7401,
+      "step": 5643
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2629798687598994e-06,
+      "loss": 0.7378,
+      "step": 5644
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.260627181574927e-06,
+      "loss": 0.7209,
+      "step": 5645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2582765403125508e-06,
+      "loss": 0.7232,
+      "step": 5646
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2559279455230566e-06,
+      "loss": 0.7116,
+      "step": 5647
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2535813977562627e-06,
+      "loss": 0.7768,
+      "step": 5648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2512368975615076e-06,
+      "loss": 0.7061,
+      "step": 5649
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.248894445487645e-06,
+      "loss": 0.6921,
+      "step": 5650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2465540420830545e-06,
+      "loss": 0.3455,
+      "step": 5651
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2442156878956369e-06,
+      "loss": 0.6811,
+      "step": 5652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2418793834728015e-06,
+      "loss": 0.6907,
+      "step": 5653
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2395451293614968e-06,
+      "loss": 0.7261,
+      "step": 5654
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2372129261081734e-06,
+      "loss": 0.6744,
+      "step": 5655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2348827742588154e-06,
+      "loss": 0.662,
+      "step": 5656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2325546743589168e-06,
+      "loss": 0.7379,
+      "step": 5657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.230228626953497e-06,
+      "loss": 0.6691,
+      "step": 5658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2279046325870925e-06,
+      "loss": 0.7467,
+      "step": 5659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2255826918037595e-06,
+      "loss": 0.6838,
+      "step": 5660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.223262805147074e-06,
+      "loss": 0.7846,
+      "step": 5661
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.220944973160133e-06,
+      "loss": 0.7599,
+      "step": 5662
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2186291963855445e-06,
+      "loss": 0.6964,
+      "step": 5663
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2163154753654505e-06,
+      "loss": 0.75,
+      "step": 5664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.214003810641493e-06,
+      "loss": 0.6483,
+      "step": 5665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2116942027548473e-06,
+      "loss": 0.7155,
+      "step": 5666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2093866522462017e-06,
+      "loss": 0.6232,
+      "step": 5667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2070811596557608e-06,
+      "loss": 0.7435,
+      "step": 5668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2047777255232517e-06,
+      "loss": 0.6663,
+      "step": 5669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.20247635038792e-06,
+      "loss": 0.3092,
+      "step": 5670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.200177034788519e-06,
+      "loss": 0.8118,
+      "step": 5671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.197879779263338e-06,
+      "loss": 0.7826,
+      "step": 5672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1955845843501634e-06,
+      "loss": 0.6992,
+      "step": 5673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1932914505863202e-06,
+      "loss": 0.7472,
+      "step": 5674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1910003785086333e-06,
+      "loss": 0.3309,
+      "step": 5675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1887113686534534e-06,
+      "loss": 0.7316,
+      "step": 5676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1864244215566467e-06,
+      "loss": 0.6878,
+      "step": 5677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.184139537753598e-06,
+      "loss": 0.6948,
+      "step": 5678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1818567177792073e-06,
+      "loss": 0.7343,
+      "step": 5679
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1795759621678937e-06,
+      "loss": 0.7214,
+      "step": 5680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1772972714535858e-06,
+      "loss": 0.7931,
+      "step": 5681
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1750206461697421e-06,
+      "loss": 0.6594,
+      "step": 5682
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1727460868493245e-06,
+      "loss": 0.691,
+      "step": 5683
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.170473594024818e-06,
+      "loss": 0.8221,
+      "step": 5684
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1682031682282213e-06,
+      "loss": 0.3258,
+      "step": 5685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1659348099910517e-06,
+      "loss": 0.6402,
+      "step": 5686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1636685198443409e-06,
+      "loss": 0.3349,
+      "step": 5687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.161404298318638e-06,
+      "loss": 0.7378,
+      "step": 5688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.159142145944001e-06,
+      "loss": 0.767,
+      "step": 5689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1568820632500165e-06,
+      "loss": 0.7283,
+      "step": 5690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1546240507657713e-06,
+      "loss": 0.3291,
+      "step": 5691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1523681090198835e-06,
+      "loss": 0.7769,
+      "step": 5692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1501142385404718e-06,
+      "loss": 0.7556,
+      "step": 5693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1478624398551785e-06,
+      "loss": 0.6835,
+      "step": 5694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1456127134911588e-06,
+      "loss": 0.7481,
+      "step": 5695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1433650599750822e-06,
+      "loss": 0.708,
+      "step": 5696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1411194798331348e-06,
+      "loss": 0.7187,
+      "step": 5697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.138875973591017e-06,
+      "loss": 0.6966,
+      "step": 5698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1366345417739388e-06,
+      "loss": 0.6746,
+      "step": 5699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1343951849066304e-06,
+      "loss": 0.7631,
+      "step": 5700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1321579035133356e-06,
+      "loss": 0.3207,
+      "step": 5701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.129922698117809e-06,
+      "loss": 0.6996,
+      "step": 5702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1276895692433231e-06,
+      "loss": 0.7168,
+      "step": 5703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.125458517412661e-06,
+      "loss": 0.6643,
+      "step": 5704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1232295431481222e-06,
+      "loss": 0.7381,
+      "step": 5705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1210026469715197e-06,
+      "loss": 0.7551,
+      "step": 5706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.118777829404174e-06,
+      "loss": 0.7275,
+      "step": 5707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1165550909669287e-06,
+      "loss": 0.7378,
+      "step": 5708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.114334432180133e-06,
+      "loss": 0.6756,
+      "step": 5709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1121158535636544e-06,
+      "loss": 0.6959,
+      "step": 5710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1098993556368697e-06,
+      "loss": 0.6675,
+      "step": 5711
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1076849389186705e-06,
+      "loss": 0.6381,
+      "step": 5712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1054726039274633e-06,
+      "loss": 0.7542,
+      "step": 5713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1032623511811568e-06,
+      "loss": 0.6631,
+      "step": 5714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1010541811971897e-06,
+      "loss": 0.6754,
+      "step": 5715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0988480944924984e-06,
+      "loss": 0.6479,
+      "step": 5716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0966440915835364e-06,
+      "loss": 0.6875,
+      "step": 5717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.094442172986272e-06,
+      "loss": 0.6772,
+      "step": 5718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0922423392161807e-06,
+      "loss": 0.7122,
+      "step": 5719
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0900445907882551e-06,
+      "loss": 0.3309,
+      "step": 5720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0878489282169967e-06,
+      "loss": 0.6662,
+      "step": 5721
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.085655352016416e-06,
+      "loss": 0.7422,
+      "step": 5722
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0834638627000427e-06,
+      "loss": 0.6421,
+      "step": 5723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.08127446078091e-06,
+      "loss": 0.7026,
+      "step": 5724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0790871467715658e-06,
+      "loss": 0.741,
+      "step": 5725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0769019211840703e-06,
+      "loss": 0.7222,
+      "step": 5726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0747187845299934e-06,
+      "loss": 0.7309,
+      "step": 5727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.072537737320416e-06,
+      "loss": 0.7105,
+      "step": 5728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0703587800659332e-06,
+      "loss": 0.787,
+      "step": 5729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0681819132766392e-06,
+      "loss": 0.6894,
+      "step": 5730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0660071374621583e-06,
+      "loss": 0.7846,
+      "step": 5731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.063834453131607e-06,
+      "loss": 0.3437,
+      "step": 5732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0616638607936213e-06,
+      "loss": 0.6792,
+      "step": 5733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0594953609563452e-06,
+      "loss": 0.7042,
+      "step": 5734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0573289541274344e-06,
+      "loss": 0.7183,
+      "step": 5735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0551646408140536e-06,
+      "loss": 0.6202,
+      "step": 5736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0530024215228784e-06,
+      "loss": 0.7298,
+      "step": 5737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0508422967600872e-06,
+      "loss": 0.7481,
+      "step": 5738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0486842670313824e-06,
+      "loss": 0.7954,
+      "step": 5739
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0465283328419596e-06,
+      "loss": 0.6688,
+      "step": 5740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0443744946965395e-06,
+      "loss": 0.741,
+      "step": 5741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.042222753099338e-06,
+      "loss": 0.6704,
+      "step": 5742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0400731085540894e-06,
+      "loss": 0.7626,
+      "step": 5743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0379255615640339e-06,
+      "loss": 0.7811,
+      "step": 5744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.03578011263192e-06,
+      "loss": 0.6438,
+      "step": 5745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0336367622600064e-06,
+      "loss": 0.7526,
+      "step": 5746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0314955109500635e-06,
+      "loss": 0.713,
+      "step": 5747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0293563592033595e-06,
+      "loss": 0.746,
+      "step": 5748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.027219307520686e-06,
+      "loss": 0.6712,
+      "step": 5749
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0250843564023304e-06,
+      "loss": 0.7107,
+      "step": 5750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0229515063480965e-06,
+      "loss": 0.6934,
+      "step": 5751
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0208207578572904e-06,
+      "loss": 0.7912,
+      "step": 5752
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0186921114287318e-06,
+      "loss": 0.7094,
+      "step": 5753
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0165655675607433e-06,
+      "loss": 0.7176,
+      "step": 5754
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0144411267511601e-06,
+      "loss": 0.6886,
+      "step": 5755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0123187894973175e-06,
+      "loss": 0.7422,
+      "step": 5756
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0101985562960693e-06,
+      "loss": 0.7294,
+      "step": 5757
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.008080427643765e-06,
+      "loss": 0.8015,
+      "step": 5758
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0059644040362725e-06,
+      "loss": 0.7155,
+      "step": 5759
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0038504859689569e-06,
+      "loss": 0.7984,
+      "step": 5760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0017386739366974e-06,
+      "loss": 0.7772,
+      "step": 5761
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.996289684338778e-07,
+      "loss": 0.7417,
+      "step": 5762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.975213699543873e-07,
+      "loss": 0.7179,
+      "step": 5763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.954158789916247e-07,
+      "loss": 0.7162,
+      "step": 5764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.933124960384944e-07,
+      "loss": 0.7315,
+      "step": 5765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.912112215874026e-07,
+      "loss": 0.7568,
+      "step": 5766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.891120561302725e-07,
+      "loss": 0.7088,
+      "step": 5767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.870150001585233e-07,
+      "loss": 0.7289,
+      "step": 5768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.849200541630843e-07,
+      "loss": 0.6938,
+      "step": 5769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828272186343923e-07,
+      "loss": 0.6847,
+      "step": 5770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.807364940623876e-07,
+      "loss": 0.7002,
+      "step": 5771
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.786478809365185e-07,
+      "loss": 0.7282,
+      "step": 5772
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.765613797457374e-07,
+      "loss": 0.7733,
+      "step": 5773
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.74476990978499e-07,
+      "loss": 0.7591,
+      "step": 5774
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.723947151227753e-07,
+      "loss": 0.7431,
+      "step": 5775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.703145526660274e-07,
+      "loss": 0.7093,
+      "step": 5776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.682365040952336e-07,
+      "loss": 0.7576,
+      "step": 5777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.661605698968723e-07,
+      "loss": 0.7196,
+      "step": 5778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.640867505569295e-07,
+      "loss": 0.7422,
+      "step": 5779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.620150465608923e-07,
+      "loss": 0.8069,
+      "step": 5780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.599454583937584e-07,
+      "loss": 0.6569,
+      "step": 5781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.578779865400245e-07,
+      "loss": 0.7437,
+      "step": 5782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.558126314836958e-07,
+      "loss": 0.6944,
+      "step": 5783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.537493937082775e-07,
+      "loss": 0.3639,
+      "step": 5784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.516882736967837e-07,
+      "loss": 0.7089,
+      "step": 5785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.496292719317312e-07,
+      "loss": 0.7679,
+      "step": 5786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.475723888951394e-07,
+      "loss": 0.8078,
+      "step": 5787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.455176250685338e-07,
+      "loss": 0.7764,
+      "step": 5788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.434649809329432e-07,
+      "loss": 0.7557,
+      "step": 5789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.414144569688998e-07,
+      "loss": 0.7614,
+      "step": 5790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.393660536564408e-07,
+      "loss": 0.758,
+      "step": 5791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.373197714751025e-07,
+      "loss": 0.7825,
+      "step": 5792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.352756109039296e-07,
+      "loss": 0.7136,
+      "step": 5793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.332335724214691e-07,
+      "loss": 0.7448,
+      "step": 5794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.311936565057689e-07,
+      "loss": 0.7091,
+      "step": 5795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.291558636343833e-07,
+      "loss": 0.7118,
+      "step": 5796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.271201942843666e-07,
+      "loss": 0.646,
+      "step": 5797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.250866489322796e-07,
+      "loss": 0.7116,
+      "step": 5798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.230552280541782e-07,
+      "loss": 0.657,
+      "step": 5799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.210259321256332e-07,
+      "loss": 0.711,
+      "step": 5800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.189987616217055e-07,
+      "loss": 0.7009,
+      "step": 5801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.169737170169657e-07,
+      "loss": 0.6931,
+      "step": 5802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.149507987854856e-07,
+      "loss": 0.7874,
+      "step": 5803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.129300074008385e-07,
+      "loss": 0.7098,
+      "step": 5804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.109113433360983e-07,
+      "loss": 0.6993,
+      "step": 5805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.08894807063846e-07,
+      "loss": 0.6546,
+      "step": 5806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.068803990561548e-07,
+      "loss": 0.7393,
+      "step": 5807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.048681197846132e-07,
+      "loss": 0.6663,
+      "step": 5808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.028579697202988e-07,
+      "loss": 0.729,
+      "step": 5809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.008499493337964e-07,
+      "loss": 0.7145,
+      "step": 5810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.98844059095193e-07,
+      "loss": 0.7698,
+      "step": 5811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.968402994740754e-07,
+      "loss": 0.7353,
+      "step": 5812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.948386709395318e-07,
+      "loss": 0.692,
+      "step": 5813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.928391739601528e-07,
+      "loss": 0.6837,
+      "step": 5814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.908418090040239e-07,
+      "loss": 0.6623,
+      "step": 5815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.888465765387433e-07,
+      "loss": 0.7216,
+      "step": 5816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.868534770313974e-07,
+      "loss": 0.6974,
+      "step": 5817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.84862510948582e-07,
+      "loss": 0.6635,
+      "step": 5818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.828736787563885e-07,
+      "loss": 0.7034,
+      "step": 5819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.808869809204124e-07,
+      "loss": 0.7025,
+      "step": 5820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.789024179057449e-07,
+      "loss": 0.8018,
+      "step": 5821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.769199901769853e-07,
+      "loss": 0.6923,
+      "step": 5822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.749396981982205e-07,
+      "loss": 0.7268,
+      "step": 5823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.729615424330529e-07,
+      "loss": 0.6849,
+      "step": 5824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.709855233445696e-07,
+      "loss": 0.7101,
+      "step": 5825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.690116413953709e-07,
+      "loss": 0.6526,
+      "step": 5826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.670398970475468e-07,
+      "loss": 0.7707,
+      "step": 5827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.650702907626907e-07,
+      "loss": 0.7289,
+      "step": 5828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.631028230018967e-07,
+      "loss": 0.7192,
+      "step": 5829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.611374942257567e-07,
+      "loss": 0.7794,
+      "step": 5830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.591743048943624e-07,
+      "loss": 0.7577,
+      "step": 5831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.572132554673052e-07,
+      "loss": 0.6566,
+      "step": 5832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.552543464036689e-07,
+      "loss": 0.7443,
+      "step": 5833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.532975781620511e-07,
+      "loss": 0.7277,
+      "step": 5834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.513429512005333e-07,
+      "loss": 0.7228,
+      "step": 5835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.493904659767027e-07,
+      "loss": 0.6995,
+      "step": 5836
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.474401229476436e-07,
+      "loss": 0.702,
+      "step": 5837
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.45491922569941e-07,
+      "loss": 0.7232,
+      "step": 5838
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.435458652996742e-07,
+      "loss": 0.7045,
+      "step": 5839
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.416019515924267e-07,
+      "loss": 0.6662,
+      "step": 5840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.396601819032702e-07,
+      "loss": 0.7175,
+      "step": 5841
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.377205566867896e-07,
+      "loss": 0.6996,
+      "step": 5842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.357830763970498e-07,
+      "loss": 0.7247,
+      "step": 5843
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.3384774148763e-07,
+      "loss": 0.6662,
+      "step": 5844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.319145524115968e-07,
+      "loss": 0.688,
+      "step": 5845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.29983509621517e-07,
+      "loss": 0.6894,
+      "step": 5846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.280546135694567e-07,
+      "loss": 0.6773,
+      "step": 5847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.261278647069781e-07,
+      "loss": 0.7463,
+      "step": 5848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.242032634851416e-07,
+      "loss": 0.7503,
+      "step": 5849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.222808103545044e-07,
+      "loss": 0.7521,
+      "step": 5850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.203605057651165e-07,
+      "loss": 0.7105,
+      "step": 5851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.184423501665351e-07,
+      "loss": 0.7186,
+      "step": 5852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.165263440078031e-07,
+      "loss": 0.6993,
+      "step": 5853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.146124877374684e-07,
+      "loss": 0.6624,
+      "step": 5854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.127007818035715e-07,
+      "loss": 0.6868,
+      "step": 5855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.1079122665365e-07,
+      "loss": 0.7172,
+      "step": 5856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.088838227347395e-07,
+      "loss": 0.7482,
+      "step": 5857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.06978570493372e-07,
+      "loss": 0.6821,
+      "step": 5858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.050754703755692e-07,
+      "loss": 0.6973,
+      "step": 5859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.031745228268628e-07,
+      "loss": 0.7743,
+      "step": 5860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.012757282922656e-07,
+      "loss": 0.7414,
+      "step": 5861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.993790872162965e-07,
+      "loss": 0.7355,
+      "step": 5862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.974846000429659e-07,
+      "loss": 0.6849,
+      "step": 5863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.955922672157812e-07,
+      "loss": 0.7164,
+      "step": 5864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.937020891777448e-07,
+      "loss": 0.6681,
+      "step": 5865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.918140663713547e-07,
+      "loss": 0.7329,
+      "step": 5866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.899281992386054e-07,
+      "loss": 0.7439,
+      "step": 5867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.880444882209881e-07,
+      "loss": 0.6894,
+      "step": 5868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.861629337594834e-07,
+      "loss": 0.3305,
+      "step": 5869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.842835362945722e-07,
+      "loss": 0.7777,
+      "step": 5870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.824062962662283e-07,
+      "loss": 0.7004,
+      "step": 5871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.805312141139232e-07,
+      "loss": 0.7619,
+      "step": 5872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.786582902766205e-07,
+      "loss": 0.7687,
+      "step": 5873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.767875251927781e-07,
+      "loss": 0.743,
+      "step": 5874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.749189193003514e-07,
+      "loss": 0.6811,
+      "step": 5875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.730524730367872e-07,
+      "loss": 0.7135,
+      "step": 5876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.711881868390292e-07,
+      "loss": 0.6893,
+      "step": 5877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.693260611435128e-07,
+      "loss": 0.7029,
+      "step": 5878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.674660963861702e-07,
+      "loss": 0.3686,
+      "step": 5879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.656082930024277e-07,
+      "loss": 0.6633,
+      "step": 5880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.63752651427202e-07,
+      "loss": 0.693,
+      "step": 5881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.618991720949087e-07,
+      "loss": 0.6672,
+      "step": 5882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.600478554394563e-07,
+      "loss": 0.665,
+      "step": 5883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.581987018942394e-07,
+      "loss": 0.7595,
+      "step": 5884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.563517118921593e-07,
+      "loss": 0.6495,
+      "step": 5885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.545068858655991e-07,
+      "loss": 0.7213,
+      "step": 5886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.526642242464422e-07,
+      "loss": 0.71,
+      "step": 5887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.508237274660612e-07,
+      "loss": 0.7176,
+      "step": 5888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.489853959553283e-07,
+      "loss": 0.7064,
+      "step": 5889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.471492301445981e-07,
+      "loss": 0.7516,
+      "step": 5890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.453152304637301e-07,
+      "loss": 0.7725,
+      "step": 5891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.434833973420663e-07,
+      "loss": 0.7209,
+      "step": 5892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.416537312084515e-07,
+      "loss": 0.7951,
+      "step": 5893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.398262324912131e-07,
+      "loss": 0.7652,
+      "step": 5894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.380009016181777e-07,
+      "loss": 0.8079,
+      "step": 5895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.361777390166635e-07,
+      "loss": 0.7006,
+      "step": 5896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.343567451134792e-07,
+      "loss": 0.7359,
+      "step": 5897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.32537920334927e-07,
+      "loss": 0.7189,
+      "step": 5898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.307212651068041e-07,
+      "loss": 0.7473,
+      "step": 5899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.289067798543903e-07,
+      "loss": 0.7528,
+      "step": 5900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.270944650024714e-07,
+      "loss": 0.701,
+      "step": 5901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.252843209753124e-07,
+      "loss": 0.6771,
+      "step": 5902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.234763481966767e-07,
+      "loss": 0.7457,
+      "step": 5903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.216705470898178e-07,
+      "loss": 0.7449,
+      "step": 5904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.198669180774831e-07,
+      "loss": 0.6917,
+      "step": 5905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.180654615819071e-07,
+      "loss": 0.6849,
+      "step": 5906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.162661780248203e-07,
+      "loss": 0.709,
+      "step": 5907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.144690678274379e-07,
+      "loss": 0.7061,
+      "step": 5908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.126741314104768e-07,
+      "loss": 0.7145,
+      "step": 5909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.10881369194133e-07,
+      "loss": 0.7773,
+      "step": 5910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.090907815981063e-07,
+      "loss": 0.6836,
+      "step": 5911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.073023690415737e-07,
+      "loss": 0.7529,
+      "step": 5912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.055161319432124e-07,
+      "loss": 0.7349,
+      "step": 5913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.037320707211881e-07,
+      "loss": 0.6787,
+      "step": 5914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.019501857931565e-07,
+      "loss": 0.7542,
+      "step": 5915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.001704775762641e-07,
+      "loss": 0.7117,
+      "step": 5916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.983929464871497e-07,
+      "loss": 0.7298,
+      "step": 5917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.966175929419349e-07,
+      "loss": 0.297,
+      "step": 5918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.948444173562452e-07,
+      "loss": 0.773,
+      "step": 5919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.930734201451817e-07,
+      "loss": 0.7133,
+      "step": 5920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.913046017233438e-07,
+      "loss": 0.7519,
+      "step": 5921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.895379625048204e-07,
+      "loss": 0.703,
+      "step": 5922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.877735029031873e-07,
+      "loss": 0.621,
+      "step": 5923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.860112233315131e-07,
+      "loss": 0.7051,
+      "step": 5924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.842511242023542e-07,
+      "loss": 0.6536,
+      "step": 5925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.824932059277545e-07,
+      "loss": 0.6684,
+      "step": 5926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.807374689192548e-07,
+      "loss": 0.6757,
+      "step": 5927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.78983913587874e-07,
+      "loss": 0.3049,
+      "step": 5928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.772325403441327e-07,
+      "loss": 0.6697,
+      "step": 5929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.754833495980295e-07,
+      "loss": 0.7305,
+      "step": 5930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.737363417590603e-07,
+      "loss": 0.7095,
+      "step": 5931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.719915172362046e-07,
+      "loss": 0.7645,
+      "step": 5932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.702488764379345e-07,
+      "loss": 0.7416,
+      "step": 5933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.685084197722091e-07,
+      "loss": 0.6893,
+      "step": 5934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.667701476464761e-07,
+      "loss": 0.6747,
+      "step": 5935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.650340604676697e-07,
+      "loss": 0.6941,
+      "step": 5936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.633001586422194e-07,
+      "loss": 0.7479,
+      "step": 5937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.61568442576036e-07,
+      "loss": 0.7331,
+      "step": 5938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.598389126745209e-07,
+      "loss": 0.7108,
+      "step": 5939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.581115693425644e-07,
+      "loss": 0.7741,
+      "step": 5940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.563864129845443e-07,
+      "loss": 0.7833,
+      "step": 5941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.546634440043287e-07,
+      "loss": 0.6844,
+      "step": 5942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.529426628052704e-07,
+      "loss": 0.7591,
+      "step": 5943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.512240697902095e-07,
+      "loss": 0.7015,
+      "step": 5944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.495076653614785e-07,
+      "loss": 0.7404,
+      "step": 5945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.477934499208927e-07,
+      "loss": 0.7094,
+      "step": 5946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.460814238697577e-07,
+      "loss": 0.7097,
+      "step": 5947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.443715876088652e-07,
+      "loss": 0.7988,
+      "step": 5948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.42663941538495e-07,
+      "loss": 0.7002,
+      "step": 5949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.409584860584139e-07,
+      "loss": 0.7244,
+      "step": 5950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.39255221567876e-07,
+      "loss": 0.685,
+      "step": 5951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.375541484656244e-07,
+      "loss": 0.7794,
+      "step": 5952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.358552671498841e-07,
+      "loss": 0.6822,
+      "step": 5953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.341585780183701e-07,
+      "loss": 0.7612,
+      "step": 5954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.324640814682859e-07,
+      "loss": 0.6455,
+      "step": 5955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.307717778963207e-07,
+      "loss": 0.7614,
+      "step": 5956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.290816676986467e-07,
+      "loss": 0.7232,
+      "step": 5957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.273937512709283e-07,
+      "loss": 0.7209,
+      "step": 5958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.257080290083129e-07,
+      "loss": 0.642,
+      "step": 5959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.240245013054358e-07,
+      "loss": 0.3149,
+      "step": 5960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.223431685564141e-07,
+      "loss": 0.7453,
+      "step": 5961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.206640311548573e-07,
+      "loss": 0.7547,
+      "step": 5962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.189870894938587e-07,
+      "loss": 0.6701,
+      "step": 5963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.173123439659956e-07,
+      "loss": 0.6924,
+      "step": 5964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.156397949633342e-07,
+      "loss": 0.7475,
+      "step": 5965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.139694428774234e-07,
+      "loss": 0.7318,
+      "step": 5966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.123012880993007e-07,
+      "loss": 0.6918,
+      "step": 5967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.10635331019489e-07,
+      "loss": 0.7456,
+      "step": 5968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.08971572027991e-07,
+      "loss": 0.7526,
+      "step": 5969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.07310011514306e-07,
+      "loss": 0.6952,
+      "step": 5970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.056506498674064e-07,
+      "loss": 0.7211,
+      "step": 5971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.03993487475757e-07,
+      "loss": 0.8029,
+      "step": 5972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.023385247273084e-07,
+      "loss": 0.7773,
+      "step": 5973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.006857620094931e-07,
+      "loss": 0.7273,
+      "step": 5974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.990351997092259e-07,
+      "loss": 0.7494,
+      "step": 5975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.973868382129155e-07,
+      "loss": 0.7131,
+      "step": 5976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.957406779064457e-07,
+      "loss": 0.7428,
+      "step": 5977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.940967191751945e-07,
+      "loss": 0.3409,
+      "step": 5978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.924549624040143e-07,
+      "loss": 0.7249,
+      "step": 5979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.908154079772488e-07,
+      "loss": 0.719,
+      "step": 5980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.891780562787242e-07,
+      "loss": 0.6785,
+      "step": 5981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.875429076917516e-07,
+      "loss": 0.7599,
+      "step": 5982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.85909962599126e-07,
+      "loss": 0.7322,
+      "step": 5983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.842792213831272e-07,
+      "loss": 0.7505,
+      "step": 5984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.826506844255142e-07,
+      "loss": 0.7337,
+      "step": 5985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.810243521075398e-07,
+      "loss": 0.6982,
+      "step": 5986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.794002248099307e-07,
+      "loss": 0.6817,
+      "step": 5987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.777783029129036e-07,
+      "loss": 0.6973,
+      "step": 5988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.761585867961572e-07,
+      "loss": 0.6692,
+      "step": 5989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.745410768388726e-07,
+      "loss": 0.697,
+      "step": 5990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.729257734197158e-07,
+      "loss": 0.7014,
+      "step": 5991
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.713126769168375e-07,
+      "loss": 0.7369,
+      "step": 5992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.697017877078659e-07,
+      "loss": 0.7231,
+      "step": 5993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.680931061699224e-07,
+      "loss": 0.7297,
+      "step": 5994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.664866326796004e-07,
+      "loss": 0.7054,
+      "step": 5995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.648823676129866e-07,
+      "loss": 0.7279,
+      "step": 5996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.63280311345642e-07,
+      "loss": 0.7357,
+      "step": 5997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.616804642526164e-07,
+      "loss": 0.6619,
+      "step": 5998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.600828267084401e-07,
+      "loss": 0.72,
+      "step": 5999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.584873990871276e-07,
+      "loss": 0.7063,
+      "step": 6000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.568941817621731e-07,
+      "loss": 0.7302,
+      "step": 6001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.553031751065574e-07,
+      "loss": 0.6213,
+      "step": 6002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.53714379492738e-07,
+      "loss": 0.761,
+      "step": 6003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.521277952926641e-07,
+      "loss": 0.7277,
+      "step": 6004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.505434228777562e-07,
+      "loss": 0.7192,
+      "step": 6005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.489612626189245e-07,
+      "loss": 0.6497,
+      "step": 6006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.473813148865581e-07,
+      "loss": 0.7348,
+      "step": 6007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.458035800505312e-07,
+      "loss": 0.7339,
+      "step": 6008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.442280584801951e-07,
+      "loss": 0.734,
+      "step": 6009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.426547505443902e-07,
+      "loss": 0.7732,
+      "step": 6010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.410836566114286e-07,
+      "loss": 0.7629,
+      "step": 6011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.395147770491149e-07,
+      "loss": 0.6483,
+      "step": 6012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.379481122247243e-07,
+      "loss": 0.7658,
+      "step": 6013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.363836625050268e-07,
+      "loss": 0.747,
+      "step": 6014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.348214282562602e-07,
+      "loss": 0.7532,
+      "step": 6015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.332614098441535e-07,
+      "loss": 0.6905,
+      "step": 6016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.317036076339111e-07,
+      "loss": 0.6848,
+      "step": 6017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.301480219902222e-07,
+      "loss": 0.7891,
+      "step": 6018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.285946532772545e-07,
+      "loss": 0.753,
+      "step": 6019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.270435018586606e-07,
+      "loss": 0.7607,
+      "step": 6020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.254945680975665e-07,
+      "loss": 0.703,
+      "step": 6021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2394785235659e-07,
+      "loss": 0.7266,
+      "step": 6022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.22403354997818e-07,
+      "loss": 0.6317,
+      "step": 6023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.20861076382827e-07,
+      "loss": 0.7691,
+      "step": 6024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.193210168726692e-07,
+      "loss": 0.757,
+      "step": 6025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.177831768278785e-07,
+      "loss": 0.7068,
+      "step": 6026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.162475566084701e-07,
+      "loss": 0.7428,
+      "step": 6027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.147141565739411e-07,
+      "loss": 0.7665,
+      "step": 6028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.13182977083262e-07,
+      "loss": 0.7372,
+      "step": 6029
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.116540184948926e-07,
+      "loss": 0.685,
+      "step": 6030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.101272811667645e-07,
+      "loss": 0.7135,
+      "step": 6031
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.086027654562952e-07,
+      "loss": 0.5965,
+      "step": 6032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.070804717203803e-07,
+      "loss": 0.6595,
+      "step": 6033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.055604003153936e-07,
+      "loss": 0.7013,
+      "step": 6034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.040425515971914e-07,
+      "loss": 0.7685,
+      "step": 6035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.025269259211075e-07,
+      "loss": 0.6506,
+      "step": 6036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.010135236419578e-07,
+      "loss": 0.7034,
+      "step": 6037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.995023451140313e-07,
+      "loss": 0.7428,
+      "step": 6038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.97993390691105e-07,
+      "loss": 0.7236,
+      "step": 6039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.964866607264307e-07,
+      "loss": 0.6686,
+      "step": 6040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.949821555727386e-07,
+      "loss": 0.671,
+      "step": 6041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9347987558224e-07,
+      "loss": 0.6738,
+      "step": 6042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.919798211066251e-07,
+      "loss": 0.7252,
+      "step": 6043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.904819924970638e-07,
+      "loss": 0.6937,
+      "step": 6044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.889863901042025e-07,
+      "loss": 0.7088,
+      "step": 6045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.874930142781675e-07,
+      "loss": 0.7411,
+      "step": 6046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.86001865368565e-07,
+      "loss": 0.6549,
+      "step": 6047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.845129437244778e-07,
+      "loss": 0.6697,
+      "step": 6048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.830262496944693e-07,
+      "loss": 0.7375,
+      "step": 6049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.815417836265812e-07,
+      "loss": 0.7621,
+      "step": 6050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.800595458683332e-07,
+      "loss": 0.7643,
+      "step": 6051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.785795367667201e-07,
+      "loss": 0.7181,
+      "step": 6052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.771017566682224e-07,
+      "loss": 0.7794,
+      "step": 6053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7562620591878995e-07,
+      "loss": 0.6606,
+      "step": 6054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7415288486386103e-07,
+      "loss": 0.3497,
+      "step": 6055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7268179384833967e-07,
+      "loss": 0.7998,
+      "step": 6056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712129332166182e-07,
+      "loss": 0.6979,
+      "step": 6057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6974630331256153e-07,
+      "loss": 0.6834,
+      "step": 6058
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.682819044795162e-07,
+      "loss": 0.7379,
+      "step": 6059
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6681973706029783e-07,
+      "loss": 0.6947,
+      "step": 6060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.653598013972116e-07,
+      "loss": 0.6815,
+      "step": 6061
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.639020978320308e-07,
+      "loss": 0.708,
+      "step": 6062
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6244662670601345e-07,
+      "loss": 0.706,
+      "step": 6063
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6099338835988827e-07,
+      "loss": 0.6718,
+      "step": 6064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5954238313386415e-07,
+      "loss": 0.6625,
+      "step": 6065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.580936113676282e-07,
+      "loss": 0.7925,
+      "step": 6066
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5664707340034475e-07,
+      "loss": 0.6797,
+      "step": 6067
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.552027695706518e-07,
+      "loss": 0.7115,
+      "step": 6068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.537607002166711e-07,
+      "loss": 0.695,
+      "step": 6069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5232086567598923e-07,
+      "loss": 0.7592,
+      "step": 6070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.508832662856866e-07,
+      "loss": 0.6827,
+      "step": 6071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.49447902382304e-07,
+      "loss": 0.6648,
+      "step": 6072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4801477430186815e-07,
+      "loss": 0.6917,
+      "step": 6073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.465838823798807e-07,
+      "loss": 0.333,
+      "step": 6074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.451552269513193e-07,
+      "loss": 0.716,
+      "step": 6075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.437288083506375e-07,
+      "loss": 0.7066,
+      "step": 6076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.423046269117659e-07,
+      "loss": 0.7754,
+      "step": 6077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.408826829681101e-07,
+      "loss": 0.7506,
+      "step": 6078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.394629768525549e-07,
+      "loss": 0.6986,
+      "step": 6079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3804550889745665e-07,
+      "loss": 0.7173,
+      "step": 6080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.366302794346544e-07,
+      "loss": 0.7552,
+      "step": 6081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.352172887954553e-07,
+      "loss": 0.7148,
+      "step": 6082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3380653731064703e-07,
+      "loss": 0.7044,
+      "step": 6083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.323980253104931e-07,
+      "loss": 0.7104,
+      "step": 6084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.30991753124731e-07,
+      "loss": 0.7216,
+      "step": 6085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.29587721082575e-07,
+      "loss": 0.6993,
+      "step": 6086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2818592951271664e-07,
+      "loss": 0.6894,
+      "step": 6087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2678637874331685e-07,
+      "loss": 0.7718,
+      "step": 6088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.253890691020213e-07,
+      "loss": 0.6787,
+      "step": 6089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2399400091594154e-07,
+      "loss": 0.6617,
+      "step": 6090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.226011745116698e-07,
+      "loss": 0.6849,
+      "step": 6091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.21210590215273e-07,
+      "loss": 0.7466,
+      "step": 6092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.198222483522929e-07,
+      "loss": 0.7099,
+      "step": 6093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.18436149247744e-07,
+      "loss": 0.7187,
+      "step": 6094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1705229322612227e-07,
+      "loss": 0.7155,
+      "step": 6095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.156706806113875e-07,
+      "loss": 0.7433,
+      "step": 6096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1429131172698644e-07,
+      "loss": 0.703,
+      "step": 6097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1291418689583083e-07,
+      "loss": 0.6578,
+      "step": 6098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1153930644031506e-07,
+      "loss": 0.6576,
+      "step": 6099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.101666706823004e-07,
+      "loss": 0.7109,
+      "step": 6100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0879627994312884e-07,
+      "loss": 0.6525,
+      "step": 6101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.074281345436137e-07,
+      "loss": 0.7264,
+      "step": 6102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.060622348040444e-07,
+      "loss": 0.7037,
+      "step": 6103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0469858104418167e-07,
+      "loss": 0.7449,
+      "step": 6104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0333717358326474e-07,
+      "loss": 0.7184,
+      "step": 6105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.019780127400008e-07,
+      "loss": 0.7339,
+      "step": 6106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.006210988325798e-07,
+      "loss": 0.7746,
+      "step": 6107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.992664321786566e-07,
+      "loss": 0.7181,
+      "step": 6108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9791401309536627e-07,
+      "loss": 0.7575,
+      "step": 6109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9656384189931564e-07,
+      "loss": 0.6534,
+      "step": 6110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9521591890658405e-07,
+      "loss": 0.8142,
+      "step": 6111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.93870244432728e-07,
+      "loss": 0.7356,
+      "step": 6112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.925268187927733e-07,
+      "loss": 0.753,
+      "step": 6113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.911856423012217e-07,
+      "loss": 0.7247,
+      "step": 6114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.898467152720487e-07,
+      "loss": 0.6731,
+      "step": 6115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.885100380187012e-07,
+      "loss": 0.6754,
+      "step": 6116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.871756108541014e-07,
+      "loss": 0.7405,
+      "step": 6117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8584343409064474e-07,
+      "loss": 0.7266,
+      "step": 6118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.845135080401985e-07,
+      "loss": 0.7651,
+      "step": 6119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8318583301410474e-07,
+      "loss": 0.7175,
+      "step": 6120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.81860409323177e-07,
+      "loss": 0.7421,
+      "step": 6121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8053723727770274e-07,
+      "loss": 0.681,
+      "step": 6122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.792163171874408e-07,
+      "loss": 0.7071,
+      "step": 6123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.778976493616238e-07,
+      "loss": 0.7191,
+      "step": 6124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7658123410895806e-07,
+      "loss": 0.6848,
+      "step": 6125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.752670717376228e-07,
+      "loss": 0.6543,
+      "step": 6126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.739551625552662e-07,
+      "loss": 0.7364,
+      "step": 6127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7264550686901493e-07,
+      "loss": 0.6497,
+      "step": 6128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.713381049854592e-07,
+      "loss": 0.6892,
+      "step": 6129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.700329572106742e-07,
+      "loss": 0.6519,
+      "step": 6130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.687300638501945e-07,
+      "loss": 0.6558,
+      "step": 6131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.674294252090349e-07,
+      "loss": 0.7439,
+      "step": 6132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6613104159168076e-07,
+      "loss": 0.6939,
+      "step": 6133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6483491330208784e-07,
+      "loss": 0.6801,
+      "step": 6134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.635410406436857e-07,
+      "loss": 0.724,
+      "step": 6135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6224942391937637e-07,
+      "loss": 0.62,
+      "step": 6136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.60960063431528e-07,
+      "loss": 0.7039,
+      "step": 6137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.596729594819914e-07,
+      "loss": 0.6842,
+      "step": 6138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.583881123720767e-07,
+      "loss": 0.6944,
+      "step": 6139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5710552240257767e-07,
+      "loss": 0.6966,
+      "step": 6140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.558251898737497e-07,
+      "loss": 0.7209,
+      "step": 6141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5454711508532525e-07,
+      "loss": 0.7457,
+      "step": 6142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.532712983365061e-07,
+      "loss": 0.7657,
+      "step": 6143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5199773992596777e-07,
+      "loss": 0.7583,
+      "step": 6144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5072644015185286e-07,
+      "loss": 0.7445,
+      "step": 6145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.494573993117811e-07,
+      "loss": 0.728,
+      "step": 6146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4819061770283603e-07,
+      "loss": 0.7145,
+      "step": 6147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469260956215803e-07,
+      "loss": 0.6488,
+      "step": 6148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4566383336404165e-07,
+      "loss": 0.7294,
+      "step": 6149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.444038312257203e-07,
+      "loss": 0.6959,
+      "step": 6150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4314608950158925e-07,
+      "loss": 0.6691,
+      "step": 6151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.418906084860907e-07,
+      "loss": 0.7328,
+      "step": 6152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.406373884731362e-07,
+      "loss": 0.6884,
+      "step": 6153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3938642975611337e-07,
+      "loss": 0.6381,
+      "step": 6154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.381377326278712e-07,
+      "loss": 0.6787,
+      "step": 6155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3689129738074034e-07,
+      "loss": 0.77,
+      "step": 6156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.356471243065107e-07,
+      "loss": 0.6603,
+      "step": 6157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3440521369645483e-07,
+      "loss": 0.7236,
+      "step": 6158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3316556584130353e-07,
+      "loss": 0.7014,
+      "step": 6159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.319281810312658e-07,
+      "loss": 0.3279,
+      "step": 6160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.306930595560176e-07,
+      "loss": 0.3257,
+      "step": 6161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.294602017047077e-07,
+      "loss": 0.7246,
+      "step": 6162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.282296077659497e-07,
+      "loss": 0.3361,
+      "step": 6163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.270012780278342e-07,
+      "loss": 0.7729,
+      "step": 6164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.257752127779157e-07,
+      "loss": 0.692,
+      "step": 6165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.245514123032234e-07,
+      "loss": 0.7005,
+      "step": 6166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.233298768902504e-07,
+      "loss": 0.3347,
+      "step": 6167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.221106068249669e-07,
+      "loss": 0.6902,
+      "step": 6168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2089360239280555e-07,
+      "loss": 0.7316,
+      "step": 6169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.196788638786752e-07,
+      "loss": 0.7715,
+      "step": 6170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.184663915669495e-07,
+      "loss": 0.7147,
+      "step": 6171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1725618574147353e-07,
+      "loss": 0.7498,
+      "step": 6172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.160482466855586e-07,
+      "loss": 0.722,
+      "step": 6173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1484257468199387e-07,
+      "loss": 0.722,
+      "step": 6174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.13639170013027e-07,
+      "loss": 0.6497,
+      "step": 6175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1243803296038154e-07,
+      "loss": 0.7593,
+      "step": 6176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.112391638052492e-07,
+      "loss": 0.7358,
+      "step": 6177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.100425628282899e-07,
+      "loss": 0.7002,
+      "step": 6178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0884823030963184e-07,
+      "loss": 0.7751,
+      "step": 6179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.076561665288746e-07,
+      "loss": 0.7485,
+      "step": 6180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0646637176508286e-07,
+      "loss": 0.7866,
+      "step": 6181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.052788462967948e-07,
+      "loss": 0.6994,
+      "step": 6182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.040935904020126e-07,
+      "loss": 0.7259,
+      "step": 6183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0291060435821306e-07,
+      "loss": 0.7617,
+      "step": 6184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0172988844233475e-07,
+      "loss": 0.7128,
+      "step": 6185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0055144293078875e-07,
+      "loss": 0.707,
+      "step": 6186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.993752680994544e-07,
+      "loss": 0.755,
+      "step": 6187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.982013642236781e-07,
+      "loss": 0.7431,
+      "step": 6188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9702973157827663e-07,
+      "loss": 0.6467,
+      "step": 6189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.958603704375351e-07,
+      "loss": 0.7549,
+      "step": 6190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9469328107520344e-07,
+      "loss": 0.3327,
+      "step": 6191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9352846376450197e-07,
+      "loss": 0.7147,
+      "step": 6192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9236591877811917e-07,
+      "loss": 0.7304,
+      "step": 6193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.91205646388214e-07,
+      "loss": 0.7775,
+      "step": 6194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.900476468664082e-07,
+      "loss": 0.8008,
+      "step": 6195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.888919204837948e-07,
+      "loss": 0.7798,
+      "step": 6196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8773846751093404e-07,
+      "loss": 0.7587,
+      "step": 6197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8658728821785665e-07,
+      "loss": 0.708,
+      "step": 6198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8543838287405367e-07,
+      "loss": 0.6815,
+      "step": 6199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.842917517484911e-07,
+      "loss": 0.3508,
+      "step": 6200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.831473951095998e-07,
+      "loss": 0.7339,
+      "step": 6201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.820053132252776e-07,
+      "loss": 0.7703,
+      "step": 6202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.808655063628918e-07,
+      "loss": 0.737,
+      "step": 6203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.797279747892745e-07,
+      "loss": 0.6428,
+      "step": 6204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7859271877072716e-07,
+      "loss": 0.7328,
+      "step": 6205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.774597385730182e-07,
+      "loss": 0.7367,
+      "step": 6206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7632903446138226e-07,
+      "loss": 0.6328,
+      "step": 6207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7520060670052087e-07,
+      "loss": 0.7185,
+      "step": 6208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7407445555460486e-07,
+      "loss": 0.7782,
+      "step": 6209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.729505812872701e-07,
+      "loss": 0.7434,
+      "step": 6210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7182898416162065e-07,
+      "loss": 0.7017,
+      "step": 6211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.707096644402274e-07,
+      "loss": 0.7409,
+      "step": 6212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.695926223851264e-07,
+      "loss": 0.7289,
+      "step": 6213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.684778582578218e-07,
+      "loss": 0.7469,
+      "step": 6214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6736537231928593e-07,
+      "loss": 0.6866,
+      "step": 6215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6625516482995384e-07,
+      "loss": 0.7165,
+      "step": 6216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6514723604973094e-07,
+      "loss": 0.6597,
+      "step": 6217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.640415862379886e-07,
+      "loss": 0.7491,
+      "step": 6218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6293821565356203e-07,
+      "loss": 0.645,
+      "step": 6219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6183712455475683e-07,
+      "loss": 0.755,
+      "step": 6220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.607383131993424e-07,
+      "loss": 0.6874,
+      "step": 6221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.596417818445529e-07,
+      "loss": 0.7144,
+      "step": 6222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5854753074709416e-07,
+      "loss": 0.6952,
+      "step": 6223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.574555601631312e-07,
+      "loss": 0.772,
+      "step": 6224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5636587034830187e-07,
+      "loss": 0.6934,
+      "step": 6225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.552784615577053e-07,
+      "loss": 0.6722,
+      "step": 6226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5419333404590797e-07,
+      "loss": 0.6439,
+      "step": 6227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.531104880669444e-07,
+      "loss": 0.7513,
+      "step": 6228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.520299238743118e-07,
+      "loss": 0.7466,
+      "step": 6229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.509516417209734e-07,
+      "loss": 0.748,
+      "step": 6230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4987564185936286e-07,
+      "loss": 0.763,
+      "step": 6231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.48801924541372e-07,
+      "loss": 0.3317,
+      "step": 6232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4773049001836747e-07,
+      "loss": 0.7289,
+      "step": 6233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4666133854117205e-07,
+      "loss": 0.8045,
+      "step": 6234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.455944703600799e-07,
+      "loss": 0.6597,
+      "step": 6235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.44529885724849e-07,
+      "loss": 0.7446,
+      "step": 6236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.434675848847023e-07,
+      "loss": 0.7518,
+      "step": 6237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.424075680883309e-07,
+      "loss": 0.6896,
+      "step": 6238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4134983558388836e-07,
+      "loss": 0.6653,
+      "step": 6239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4029438761899116e-07,
+      "loss": 0.7309,
+      "step": 6240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.392412244407294e-07,
+      "loss": 0.7086,
+      "step": 6241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3819034629564698e-07,
+      "loss": 0.701,
+      "step": 6242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3714175342976265e-07,
+      "loss": 0.7089,
+      "step": 6243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.360954460885556e-07,
+      "loss": 0.7192,
+      "step": 6244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3505142451696883e-07,
+      "loss": 0.6868,
+      "step": 6245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3400968895941346e-07,
+      "loss": 0.693,
+      "step": 6246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.329702396597655e-07,
+      "loss": 0.7629,
+      "step": 6247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3193307686136035e-07,
+      "loss": 0.6174,
+      "step": 6248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3089820080700597e-07,
+      "loss": 0.7587,
+      "step": 6249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.298656117389675e-07,
+      "loss": 0.7223,
+      "step": 6250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.288353098989815e-07,
+      "loss": 0.725,
+      "step": 6251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2780729552824398e-07,
+      "loss": 0.7116,
+      "step": 6252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2678156886741687e-07,
+      "loss": 0.6742,
+      "step": 6253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2575813015662917e-07,
+      "loss": 0.7239,
+      "step": 6254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2473697963546925e-07,
+      "loss": 0.7053,
+      "step": 6255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2371811754299367e-07,
+      "loss": 0.6826,
+      "step": 6256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2270154411772495e-07,
+      "loss": 0.6943,
+      "step": 6257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.216872595976416e-07,
+      "loss": 0.7153,
+      "step": 6258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2067526422019592e-07,
+      "loss": 0.6549,
+      "step": 6259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196655582222984e-07,
+      "loss": 0.5782,
+      "step": 6260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1865814184032662e-07,
+      "loss": 0.6425,
+      "step": 6261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.176530153101186e-07,
+      "loss": 0.3563,
+      "step": 6262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1665017886698058e-07,
+      "loss": 0.7141,
+      "step": 6263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1564963274568028e-07,
+      "loss": 0.728,
+      "step": 6264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1465137718044925e-07,
+      "loss": 0.7158,
+      "step": 6265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1365541240498167e-07,
+      "loss": 0.7248,
+      "step": 6266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.12661738652441e-07,
+      "loss": 0.6885,
+      "step": 6267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1167035615544562e-07,
+      "loss": 0.7284,
+      "step": 6268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1068126514608546e-07,
+      "loss": 0.7279,
+      "step": 6269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0969446585590857e-07,
+      "loss": 0.7462,
+      "step": 6270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0870995851593023e-07,
+      "loss": 0.7862,
+      "step": 6271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0772774335662714e-07,
+      "loss": 0.7763,
+      "step": 6272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0674782060793985e-07,
+      "loss": 0.7544,
+      "step": 6273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.057701904992726e-07,
+      "loss": 0.7186,
+      "step": 6274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0479485325949233e-07,
+      "loss": 0.6515,
+      "step": 6275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.038218091169275e-07,
+      "loss": 0.7527,
+      "step": 6276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.028510582993748e-07,
+      "loss": 0.7176,
+      "step": 6277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0188260103408907e-07,
+      "loss": 0.8046,
+      "step": 6278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.009164375477901e-07,
+      "loss": 0.7178,
+      "step": 6279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9995256806666143e-07,
+      "loss": 0.6829,
+      "step": 6280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9899099281634805e-07,
+      "loss": 0.6224,
+      "step": 6281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.980317120219599e-07,
+      "loss": 0.7542,
+      "step": 6282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9707472590806854e-07,
+      "loss": 0.6919,
+      "step": 6283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9612003469870578e-07,
+      "loss": 0.72,
+      "step": 6284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9516763861737066e-07,
+      "loss": 0.7597,
+      "step": 6285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.942175378870237e-07,
+      "loss": 0.7298,
+      "step": 6286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9326973273008477e-07,
+      "loss": 0.3437,
+      "step": 6287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9232422336844081e-07,
+      "loss": 0.7584,
+      "step": 6288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9138101002343922e-07,
+      "loss": 0.7081,
+      "step": 6289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9044009291589116e-07,
+      "loss": 0.7536,
+      "step": 6290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8950147226606595e-07,
+      "loss": 0.7055,
+      "step": 6291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.885651482937012e-07,
+      "loss": 0.7893,
+      "step": 6292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8763112121799265e-07,
+      "loss": 0.7421,
+      "step": 6293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8669939125760094e-07,
+      "loss": 0.6927,
+      "step": 6294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8576995863064717e-07,
+      "loss": 0.7046,
+      "step": 6295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8484282355471616e-07,
+      "loss": 0.7714,
+      "step": 6296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8391798624685208e-07,
+      "loss": 0.703,
+      "step": 6297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.829954469235662e-07,
+      "loss": 0.7376,
+      "step": 6298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820752058008246e-07,
+      "loss": 0.6826,
+      "step": 6299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.811572630940639e-07,
+      "loss": 0.7032,
+      "step": 6300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.802416190181744e-07,
+      "loss": 0.31,
+      "step": 6301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.793282737875146e-07,
+      "loss": 0.6981,
+      "step": 6302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7841722761590019e-07,
+      "loss": 0.7653,
+      "step": 6303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7750848071661274e-07,
+      "loss": 0.794,
+      "step": 6304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7660203330239322e-07,
+      "loss": 0.7343,
+      "step": 6305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.756978855854441e-07,
+      "loss": 0.745,
+      "step": 6306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7479603777742937e-07,
+      "loss": 0.6949,
+      "step": 6307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7389649008947795e-07,
+      "loss": 0.6552,
+      "step": 6308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7299924273217362e-07,
+      "loss": 0.5646,
+      "step": 6309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7210429591556832e-07,
+      "loss": 0.7133,
+      "step": 6310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7121164984917227e-07,
+      "loss": 0.6998,
+      "step": 6311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7032130474195718e-07,
+      "loss": 0.6656,
+      "step": 6312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6943326080235743e-07,
+      "loss": 0.6823,
+      "step": 6313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6854751823826675e-07,
+      "loss": 0.7622,
+      "step": 6314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6766407725704038e-07,
+      "loss": 0.7017,
+      "step": 6315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6678293806549728e-07,
+      "loss": 0.7024,
+      "step": 6316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.659041008699136e-07,
+      "loss": 0.7268,
+      "step": 6317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.650275658760314e-07,
+      "loss": 0.7134,
+      "step": 6318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.641533332890477e-07,
+      "loss": 0.7194,
+      "step": 6319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6328140331362652e-07,
+      "loss": 0.6912,
+      "step": 6320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6241177615388905e-07,
+      "loss": 0.7461,
+      "step": 6321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6154445201341907e-07,
+      "loss": 0.6378,
+      "step": 6322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6067943109526085e-07,
+      "loss": 0.6979,
+      "step": 6323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5981671360191908e-07,
+      "loss": 0.7156,
+      "step": 6324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5895629973535775e-07,
+      "loss": 0.7097,
+      "step": 6325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5809818969700575e-07,
+      "loss": 0.7392,
+      "step": 6326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.572423836877479e-07,
+      "loss": 0.7453,
+      "step": 6327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.563888819079351e-07,
+      "loss": 0.7185,
+      "step": 6328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5553768455737306e-07,
+      "loss": 0.7554,
+      "step": 6329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5468879183533015e-07,
+      "loss": 0.7039,
+      "step": 6330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5384220394053628e-07,
+      "loss": 0.7595,
+      "step": 6331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5299792107118183e-07,
+      "loss": 0.7757,
+      "step": 6332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5215594342491536e-07,
+      "loss": 0.7121,
+      "step": 6333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5131627119884917e-07,
+      "loss": 0.7093,
+      "step": 6334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.504789045895516e-07,
+      "loss": 0.7939,
+      "step": 6335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4964384379305696e-07,
+      "loss": 0.7208,
+      "step": 6336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4881108900485331e-07,
+      "loss": 0.7651,
+      "step": 6337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4798064041989247e-07,
+      "loss": 0.7511,
+      "step": 6338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4715249823258559e-07,
+      "loss": 0.7567,
+      "step": 6339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.463266626368054e-07,
+      "loss": 0.6773,
+      "step": 6340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4550313382588387e-07,
+      "loss": 0.715,
+      "step": 6341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4468191199261016e-07,
+      "loss": 0.7228,
+      "step": 6342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4386299732923603e-07,
+      "loss": 0.6885,
+      "step": 6343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4304639002747588e-07,
+      "loss": 0.6863,
+      "step": 6344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.422320902784957e-07,
+      "loss": 0.7027,
+      "step": 6345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.414200982729308e-07,
+      "loss": 0.6162,
+      "step": 6346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4061041420086908e-07,
+      "loss": 0.6821,
+      "step": 6347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3980303825186226e-07,
+      "loss": 0.741,
+      "step": 6348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3899797061491914e-07,
+      "loss": 0.7324,
+      "step": 6349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3819521147851122e-07,
+      "loss": 0.6433,
+      "step": 6350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3739476103056481e-07,
+      "loss": 0.7042,
+      "step": 6351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3659661945847224e-07,
+      "loss": 0.6741,
+      "step": 6352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.358007869490785e-07,
+      "loss": 0.7019,
+      "step": 6353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3500726368869233e-07,
+      "loss": 0.6921,
+      "step": 6354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3421604986308068e-07,
+      "loss": 0.7305,
+      "step": 6355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3342714565747094e-07,
+      "loss": 0.7436,
+      "step": 6356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3264055125654652e-07,
+      "loss": 0.7486,
+      "step": 6357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3185626684445452e-07,
+      "loss": 0.6987,
+      "step": 6358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.310742926047992e-07,
+      "loss": 0.7486,
+      "step": 6359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3029462872064302e-07,
+      "loss": 0.7109,
+      "step": 6360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2951727537450664e-07,
+      "loss": 0.6497,
+      "step": 6361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.287422327483745e-07,
+      "loss": 0.7357,
+      "step": 6362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.27969501023687e-07,
+      "loss": 0.7612,
+      "step": 6363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2719908038134164e-07,
+      "loss": 0.7495,
+      "step": 6364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2643097100169865e-07,
+      "loss": 0.6917,
+      "step": 6365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2566517306457525e-07,
+      "loss": 0.6367,
+      "step": 6366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2490168674924806e-07,
+      "loss": 0.6534,
+      "step": 6367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.241405122344519e-07,
+      "loss": 0.7075,
+      "step": 6368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2338164969838085e-07,
+      "loss": 0.7287,
+      "step": 6369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2262509931868838e-07,
+      "loss": 0.7833,
+      "step": 6370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2187086127248393e-07,
+      "loss": 0.7025,
+      "step": 6371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.211189357363407e-07,
+      "loss": 0.7176,
+      "step": 6372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2036932288628566e-07,
+      "loss": 0.7477,
+      "step": 6373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1962202289780623e-07,
+      "loss": 0.7559,
+      "step": 6374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1887703594584909e-07,
+      "loss": 0.7338,
+      "step": 6375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.18134362204817e-07,
+      "loss": 0.7479,
+      "step": 6376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1739400184857642e-07,
+      "loss": 0.6989,
+      "step": 6377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1665595505044424e-07,
+      "loss": 0.7221,
+      "step": 6378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1592022198320341e-07,
+      "loss": 0.6591,
+      "step": 6379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1518680281909167e-07,
+      "loss": 0.7409,
+      "step": 6380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1445569772980392e-07,
+      "loss": 0.7361,
+      "step": 6381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1372690688649546e-07,
+      "loss": 0.681,
+      "step": 6382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1300043045978094e-07,
+      "loss": 0.6943,
+      "step": 6383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1227626861972873e-07,
+      "loss": 0.3605,
+      "step": 6384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1155442153586993e-07,
+      "loss": 0.7157,
+      "step": 6385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1083488937719045e-07,
+      "loss": 0.6991,
+      "step": 6386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1011767231213665e-07,
+      "loss": 0.7882,
+      "step": 6387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0940277050861203e-07,
+      "loss": 0.6991,
+      "step": 6388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0869018413397714e-07,
+      "loss": 0.696,
+      "step": 6389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0797991335505187e-07,
+      "loss": 0.7354,
+      "step": 6390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0727195833811433e-07,
+      "loss": 0.6936,
+      "step": 6391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0656631924889749e-07,
+      "loss": 0.6855,
+      "step": 6392
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0586299625259699e-07,
+      "loss": 0.6626,
+      "step": 6393
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0516198951386114e-07,
+      "loss": 0.713,
+      "step": 6394
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0446329919680087e-07,
+      "loss": 0.7439,
+      "step": 6395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.037669254649809e-07,
+      "loss": 0.7468,
+      "step": 6396
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0307286848142417e-07,
+      "loss": 0.7106,
+      "step": 6397
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0238112840861514e-07,
+      "loss": 0.6895,
+      "step": 6398
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0169170540849094e-07,
+      "loss": 0.758,
+      "step": 6399
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0100459964244691e-07,
+      "loss": 0.714,
+      "step": 6400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0031981127133994e-07,
+      "loss": 0.3429,
+      "step": 6401
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.963734045548067e-08,
+      "loss": 0.6996,
+      "step": 6402
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.895718735463911e-08,
+      "loss": 0.6763,
+      "step": 6403
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.827935212804119e-08,
+      "loss": 0.3335,
+      "step": 6404
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.760383493436998e-08,
+      "loss": 0.7214,
+      "step": 6405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.693063593176788e-08,
+      "loss": 0.7419,
+      "step": 6406
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.625975527783327e-08,
+      "loss": 0.7774,
+      "step": 6407
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.559119312962273e-08,
+      "loss": 0.6764,
+      "step": 6408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.492494964364885e-08,
+      "loss": 0.6725,
+      "step": 6409
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.426102497588019e-08,
+      "loss": 0.7006,
+      "step": 6410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.359941928174799e-08,
+      "loss": 0.7603,
+      "step": 6411
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.294013271613278e-08,
+      "loss": 0.6851,
+      "step": 6412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.22831654333789e-08,
+      "loss": 0.7504,
+      "step": 6413
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.162851758728442e-08,
+      "loss": 0.7697,
+      "step": 6414
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.097618933110452e-08,
+      "loss": 0.7422,
+      "step": 6415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.032618081755152e-08,
+      "loss": 0.8136,
+      "step": 6416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.967849219879698e-08,
+      "loss": 0.6377,
+      "step": 6417
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.903312362646521e-08,
+      "loss": 0.7393,
+      "step": 6418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.839007525164201e-08,
+      "loss": 0.6781,
+      "step": 6419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.774934722486472e-08,
+      "loss": 0.6872,
+      "step": 6420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.711093969613226e-08,
+      "loss": 0.6222,
+      "step": 6421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.647485281489843e-08,
+      "loss": 0.6483,
+      "step": 6422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.584108673007296e-08,
+      "loss": 0.7282,
+      "step": 6423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.520964159002276e-08,
+      "loss": 0.7573,
+      "step": 6424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.45805175425729e-08,
+      "loss": 0.7163,
+      "step": 6425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.395371473500336e-08,
+      "loss": 0.6878,
+      "step": 6426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.332923331405229e-08,
+      "loss": 0.3421,
+      "step": 6427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.270707342591056e-08,
+      "loss": 0.6916,
+      "step": 6428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.208723521623052e-08,
+      "loss": 0.7735,
+      "step": 6429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.146971883011834e-08,
+      "loss": 0.7036,
+      "step": 6430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.085452441213726e-08,
+      "loss": 0.701,
+      "step": 6431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.024165210630652e-08,
+      "loss": 0.6916,
+      "step": 6432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.963110205610137e-08,
+      "loss": 0.736,
+      "step": 6433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.902287440445521e-08,
+      "loss": 0.7678,
+      "step": 6434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.841696929375641e-08,
+      "loss": 0.664,
+      "step": 6435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.781338686584928e-08,
+      "loss": 0.6977,
+      "step": 6436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.72121272620352e-08,
+      "loss": 0.7625,
+      "step": 6437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.66131906230716e-08,
+      "loss": 0.634,
+      "step": 6438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.601657708917076e-08,
+      "loss": 0.7017,
+      "step": 6439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.542228680000429e-08,
+      "loss": 0.675,
+      "step": 6440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.483031989469647e-08,
+      "loss": 0.6959,
+      "step": 6441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.424067651183087e-08,
+      "loss": 0.7241,
+      "step": 6442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.365335678944263e-08,
+      "loss": 0.7076,
+      "step": 6443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.306836086502844e-08,
+      "loss": 0.687,
+      "step": 6444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.248568887553653e-08,
+      "loss": 0.7456,
+      "step": 6445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.190534095737444e-08,
+      "loss": 0.6552,
+      "step": 6446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.13273172464013e-08,
+      "loss": 0.3463,
+      "step": 6447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.075161787793661e-08,
+      "loss": 0.7164,
+      "step": 6448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.017824298675369e-08,
+      "loss": 0.3329,
+      "step": 6449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.960719270708294e-08,
+      "loss": 0.6592,
+      "step": 6450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.903846717260742e-08,
+      "loss": 0.7447,
+      "step": 6451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.847206651646953e-08,
+      "loss": 0.7126,
+      "step": 6452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.790799087126543e-08,
+      "loss": 0.7845,
+      "step": 6453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.734624036904836e-08,
+      "loss": 0.6594,
+      "step": 6454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.678681514132424e-08,
+      "loss": 0.7378,
+      "step": 6455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.622971531905942e-08,
+      "loss": 0.7447,
+      "step": 6456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.56749410326718e-08,
+      "loss": 0.7724,
+      "step": 6457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.512249241203639e-08,
+      "loss": 0.7505,
+      "step": 6458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.457236958648305e-08,
+      "loss": 0.74,
+      "step": 6459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.40245726847999e-08,
+      "loss": 0.7396,
+      "step": 6460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.347910183522654e-08,
+      "loss": 0.7175,
+      "step": 6461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.293595716545975e-08,
+      "loss": 0.6696,
+      "step": 6462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.239513880265335e-08,
+      "loss": 0.7672,
+      "step": 6463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.185664687341386e-08,
+      "loss": 0.7459,
+      "step": 6464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.132048150380487e-08,
+      "loss": 0.7167,
+      "step": 6465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.078664281934599e-08,
+      "loss": 0.7352,
+      "step": 6466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.025513094500835e-08,
+      "loss": 0.7145,
+      "step": 6467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.972594600522352e-08,
+      "loss": 0.7087,
+      "step": 6468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.919908812387354e-08,
+      "loss": 0.7139,
+      "step": 6469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8674557424300835e-08,
+      "loss": 0.8137,
+      "step": 6470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.81523540292972e-08,
+      "loss": 0.7132,
+      "step": 6471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7632478061114826e-08,
+      "loss": 0.7395,
+      "step": 6472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.711492964145859e-08,
+      "loss": 0.3566,
+      "step": 6473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.659970889148714e-08,
+      "loss": 0.3704,
+      "step": 6474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.608681593181731e-08,
+      "loss": 0.674,
+      "step": 6475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.557625088251861e-08,
+      "loss": 0.7302,
+      "step": 6476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5068013863115444e-08,
+      "loss": 0.7502,
+      "step": 6477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.456210499259151e-08,
+      "loss": 0.7501,
+      "step": 6478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.405852438937764e-08,
+      "loss": 0.77,
+      "step": 6479
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.35572721713673e-08,
+      "loss": 0.7247,
+      "step": 6480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3058348455905514e-08,
+      "loss": 0.6695,
+      "step": 6481
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.256175335978997e-08,
+      "loss": 0.6729,
+      "step": 6482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.206748699927766e-08,
+      "loss": 0.6714,
+      "step": 6483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.157554949007826e-08,
+      "loss": 0.6818,
+      "step": 6484
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.10859409473552e-08,
+      "loss": 0.6867,
+      "step": 6485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0598661485730114e-08,
+      "loss": 0.7164,
+      "step": 6486
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0113711219274e-08,
+      "loss": 0.7119,
+      "step": 6487
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.963109026151713e-08,
+      "loss": 0.7063,
+      "step": 6488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9150798725443594e-08,
+      "loss": 0.7025,
+      "step": 6489
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8672836723491216e-08,
+      "loss": 0.6755,
+      "step": 6490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8197204367552706e-08,
+      "loss": 0.7275,
+      "step": 6491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.772390176897568e-08,
+      "loss": 0.7063,
+      "step": 6492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.725292903856149e-08,
+      "loss": 0.6825,
+      "step": 6493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.67842862865675e-08,
+      "loss": 0.7024,
+      "step": 6494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.631797362270485e-08,
+      "loss": 0.7003,
+      "step": 6495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5853991156139534e-08,
+      "loss": 0.7224,
+      "step": 6496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.539233899548912e-08,
+      "loss": 0.7511,
+      "step": 6497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.49330172488327e-08,
+      "loss": 0.6652,
+      "step": 6498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.44760260236965e-08,
+      "loss": 0.7006,
+      "step": 6499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.402136542706381e-08,
+      "loss": 0.6927,
+      "step": 6500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.356903556537395e-08,
+      "loss": 0.6277,
+      "step": 6501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.311903654451888e-08,
+      "loss": 0.7488,
+      "step": 6502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.267136846984321e-08,
+      "loss": 0.7348,
+      "step": 6503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.222603144615089e-08,
+      "loss": 0.7307,
+      "step": 6504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.17830255776952e-08,
+      "loss": 0.7296,
+      "step": 6505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1342350968186505e-08,
+      "loss": 0.7343,
+      "step": 6506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.090400772078784e-08,
+      "loss": 0.6889,
+      "step": 6507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0467995938117124e-08,
+      "loss": 0.6909,
+      "step": 6508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0034315722247144e-08,
+      "loss": 0.6854,
+      "step": 6509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.960296717470336e-08,
+      "loss": 0.7057,
+      "step": 6510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9173950396467206e-08,
+      "loss": 0.6927,
+      "step": 6511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.874726548797281e-08,
+      "loss": 0.7754,
+      "step": 6512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8322912549108024e-08,
+      "loss": 0.6317,
+      "step": 6513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.790089167921673e-08,
+      "loss": 0.7435,
+      "step": 6514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.748120297709434e-08,
+      "loss": 0.7558,
+      "step": 6515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7063846540993376e-08,
+      "loss": 0.7382,
+      "step": 6516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6648822468617897e-08,
+      "loss": 0.674,
+      "step": 6517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.623613085712685e-08,
+      "loss": 0.7528,
+      "step": 6518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.582577180313296e-08,
+      "loss": 0.7694,
+      "step": 6519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5417745402703817e-08,
+      "loss": 0.7784,
+      "step": 6520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.501205175135747e-08,
+      "loss": 0.7208,
+      "step": 6521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.460869094407127e-08,
+      "loss": 0.7438,
+      "step": 6522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.42076630752719e-08,
+      "loss": 0.3568,
+      "step": 6523
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3808968238842054e-08,
+      "loss": 0.6983,
+      "step": 6524
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3412606528118177e-08,
+      "loss": 0.7597,
+      "step": 6525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.301857803588937e-08,
+      "loss": 0.7245,
+      "step": 6526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2626882854400745e-08,
+      "loss": 0.7158,
+      "step": 6527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.223752107534783e-08,
+      "loss": 0.6679,
+      "step": 6528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.185049278988328e-08,
+      "loss": 0.6587,
+      "step": 6529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1465798088611274e-08,
+      "loss": 0.6975,
+      "step": 6530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1083437061590894e-08,
+      "loss": 0.792,
+      "step": 6531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.070340979833386e-08,
+      "loss": 0.7248,
+      "step": 6532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0325716387806794e-08,
+      "loss": 0.742,
+      "step": 6533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.995035691842896e-08,
+      "loss": 0.6918,
+      "step": 6534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.957733147807229e-08,
+      "loss": 0.6034,
+      "step": 6535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9206640154065823e-08,
+      "loss": 0.6553,
+      "step": 6536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.883828303318792e-08,
+      "loss": 0.6897,
+      "step": 6537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.847226020167404e-08,
+      "loss": 0.7783,
+      "step": 6538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.810857174521009e-08,
+      "loss": 0.6896,
+      "step": 6539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.774721774893796e-08,
+      "loss": 0.7588,
+      "step": 6540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7388198297452206e-08,
+      "loss": 0.3403,
+      "step": 6541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.703151347480004e-08,
+      "loss": 0.6686,
+      "step": 6542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.667716336448356e-08,
+      "loss": 0.8157,
+      "step": 6543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6325148049456408e-08,
+      "loss": 0.3431,
+      "step": 6544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5975467612129323e-08,
+      "loss": 0.7174,
+      "step": 6545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5628122134360167e-08,
+      "loss": 0.2803,
+      "step": 6546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5283111697467222e-08,
+      "loss": 0.6973,
+      "step": 6547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4940436382216994e-08,
+      "loss": 0.6869,
+      "step": 6548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4600096268831975e-08,
+      "loss": 0.7203,
+      "step": 6549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4262091436987323e-08,
+      "loss": 0.7412,
+      "step": 6550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3926421965810853e-08,
+      "loss": 0.7248,
+      "step": 6551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.359308793388304e-08,
+      "loss": 0.5823,
+      "step": 6552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3262089419241464e-08,
+      "loss": 0.7223,
+      "step": 6553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2933426499370803e-08,
+      "loss": 0.6791,
+      "step": 6554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2607099251216182e-08,
+      "loss": 0.3508,
+      "step": 6555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.228310775116871e-08,
+      "loss": 0.6933,
+      "step": 6556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1961452075078826e-08,
+      "loss": 0.7289,
+      "step": 6557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.164213229824519e-08,
+      "loss": 0.7058,
+      "step": 6558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1325148495423554e-08,
+      "loss": 0.3559,
+      "step": 6559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1010500740820116e-08,
+      "loss": 0.7676,
+      "step": 6560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0698189108095958e-08,
+      "loss": 0.6676,
+      "step": 6561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.038821367036259e-08,
+      "loss": 0.6711,
+      "step": 6562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0080574500189743e-08,
+      "loss": 0.7648,
+      "step": 6563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9775271669594252e-08,
+      "loss": 0.7125,
+      "step": 6564
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.947230525005006e-08,
+      "loss": 0.3622,
+      "step": 6565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.917167531248265e-08,
+      "loss": 0.3111,
+      "step": 6566
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8873381927270173e-08,
+      "loss": 0.796,
+      "step": 6567
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8577425164244544e-08,
+      "loss": 0.6801,
+      "step": 6568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8283805092690343e-08,
+      "loss": 0.6392,
+      "step": 6569
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7992521781345917e-08,
+      "loss": 0.6722,
+      "step": 6570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7703575298401165e-08,
+      "loss": 0.7107,
+      "step": 6571
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7416965711500865e-08,
+      "loss": 0.7193,
+      "step": 6572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7132693087739127e-08,
+      "loss": 0.7818,
+      "step": 6573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.685075749366716e-08,
+      "loss": 0.6584,
+      "step": 6574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.657115899528772e-08,
+      "loss": 0.3468,
+      "step": 6575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6293897658054004e-08,
+      "loss": 0.7302,
+      "step": 6576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6018973546875206e-08,
+      "loss": 0.7388,
+      "step": 6577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5746386726110952e-08,
+      "loss": 0.6888,
+      "step": 6578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.547613725957797e-08,
+      "loss": 0.7459,
+      "step": 6579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5208225210538998e-08,
+      "loss": 0.7019,
+      "step": 6580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4942650641716073e-08,
+      "loss": 0.7052,
+      "step": 6581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.467941361527947e-08,
+      "loss": 0.7176,
+      "step": 6582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4418514192855448e-08,
+      "loss": 0.3397,
+      "step": 6583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4159952435520708e-08,
+      "loss": 0.7314,
+      "step": 6584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3903728403805717e-08,
+      "loss": 0.7763,
+      "step": 6585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3649842157694715e-08,
+      "loss": 0.6931,
+      "step": 6586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3398293756621272e-08,
+      "loss": 0.7363,
+      "step": 6587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3149083259476059e-08,
+      "loss": 0.6681,
+      "step": 6588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2902210724599074e-08,
+      "loss": 0.6362,
+      "step": 6589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2657676209784087e-08,
+      "loss": 0.613,
+      "step": 6590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2415479772278638e-08,
+      "loss": 0.7245,
+      "step": 6591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2175621468780708e-08,
+      "loss": 0.7279,
+      "step": 6592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1938101355442045e-08,
+      "loss": 0.801,
+      "step": 6593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.170291948786928e-08,
+      "loss": 0.7098,
+      "step": 6594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1470075921116153e-08,
+      "loss": 0.7532,
+      "step": 6595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1239570709694614e-08,
+      "loss": 0.7593,
+      "step": 6596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1011403907567053e-08,
+      "loss": 0.6809,
+      "step": 6597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.07855755681463e-08,
+      "loss": 0.3309,
+      "step": 6598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0562085744301176e-08,
+      "loss": 0.6958,
+      "step": 6599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.034093448835094e-08,
+      "loss": 0.6244,
+      "step": 6600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0122121852068623e-08,
+      "loss": 0.6912,
+      "step": 6601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.905647886678804e-09,
+      "loss": 0.7287,
+      "step": 6602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.691512642858724e-09,
+      "loss": 0.7058,
+      "step": 6603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.479716170738284e-09,
+      "loss": 0.7325,
+      "step": 6604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.270258519900044e-09,
+      "loss": 0.7397,
+      "step": 6605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.063139739379223e-09,
+      "loss": 0.6673,
+      "step": 6606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.858359877661483e-09,
+      "loss": 0.3252,
+      "step": 6607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.655918982689582e-09,
+      "loss": 0.7289,
+      "step": 6608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.455817101853392e-09,
+      "loss": 0.6753,
+      "step": 6609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.258054281998772e-09,
+      "loss": 0.7521,
+      "step": 6610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.062630569422025e-09,
+      "loss": 0.7383,
+      "step": 6611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.869546009873219e-09,
+      "loss": 0.6684,
+      "step": 6612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.678800648553975e-09,
+      "loss": 0.7103,
+      "step": 6613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.490394530118572e-09,
+      "loss": 0.7071,
+      "step": 6614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.3043276986728415e-09,
+      "loss": 0.7161,
+      "step": 6615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.120600197777494e-09,
+      "loss": 0.746,
+      "step": 6616
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.939212070442569e-09,
+      "loss": 0.6983,
+      "step": 6617
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.760163359131877e-09,
+      "loss": 0.7492,
+      "step": 6618
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.583454105761888e-09,
+      "loss": 0.7113,
+      "step": 6619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.409084351701733e-09,
+      "loss": 0.6997,
+      "step": 6620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.237054137769871e-09,
+      "loss": 0.718,
+      "step": 6621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.067363504241863e-09,
+      "loss": 0.7061,
+      "step": 6622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.900012490840379e-09,
+      "loss": 0.7365,
+      "step": 6623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.7350011367451885e-09,
+      "loss": 0.6488,
+      "step": 6624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.572329480584282e-09,
+      "loss": 0.7062,
+      "step": 6625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.41199756044053e-09,
+      "loss": 0.6875,
+      "step": 6626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.254005413849461e-09,
+      "loss": 0.6802,
+      "step": 6627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.098353077795937e-09,
+      "loss": 0.6999,
+      "step": 6628
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.945040588719696e-09,
+      "loss": 0.7656,
+      "step": 6629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7940679825120295e-09,
+      "loss": 0.6695,
+      "step": 6630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.645435294514666e-09,
+      "loss": 0.7603,
+      "step": 6631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.499142559525327e-09,
+      "loss": 0.7271,
+      "step": 6632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.35518981178995e-09,
+      "loss": 0.3405,
+      "step": 6633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.213577085009357e-09,
+      "loss": 0.7434,
+      "step": 6634
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0743044123359166e-09,
+      "loss": 0.6871,
+      "step": 6635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.937371826372438e-09,
+      "loss": 0.7338,
+      "step": 6636
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.802779359177722e-09,
+      "loss": 0.7022,
+      "step": 6637
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.670527042258787e-09,
+      "loss": 0.6736,
+      "step": 6638
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5406149065764227e-09,
+      "loss": 0.34,
+      "step": 6639
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4130429825440792e-09,
+      "loss": 0.6617,
+      "step": 6640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.287811300026755e-09,
+      "loss": 0.8138,
+      "step": 6641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1649198883410004e-09,
+      "loss": 0.7043,
+      "step": 6642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0443687762571337e-09,
+      "loss": 0.3475,
+      "step": 6643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.926157991997025e-09,
+      "loss": 0.7632,
+      "step": 6644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.810287563232983e-09,
+      "loss": 0.7677,
+      "step": 6645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6967575170921965e-09,
+      "loss": 0.7845,
+      "step": 6646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5855678801511853e-09,
+      "loss": 0.6715,
+      "step": 6647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4767186784402376e-09,
+      "loss": 0.6778,
+      "step": 6648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3702099374423025e-09,
+      "loss": 0.7019,
+      "step": 6649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.266041682089659e-09,
+      "loss": 0.6684,
+      "step": 6650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.164213936770576e-09,
+      "loss": 0.7522,
+      "step": 6651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.064726725322652e-09,
+      "loss": 0.8273,
+      "step": 6652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967580071035036e-09,
+      "loss": 0.6895,
+      "step": 6653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8727739966517557e-09,
+      "loss": 0.6794,
+      "step": 6654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.78030852436617e-09,
+      "loss": 0.649,
+      "step": 6655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6901836758265178e-09,
+      "loss": 0.7428,
+      "step": 6656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6023994721292569e-09,
+      "loss": 0.6868,
+      "step": 6657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5169559338257256e-09,
+      "loss": 0.3262,
+      "step": 6658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4338530809188122e-09,
+      "loss": 0.7023,
+      "step": 6659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.353090932864065e-09,
+      "loss": 0.7286,
+      "step": 6660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2746695085663618e-09,
+      "loss": 0.7361,
+      "step": 6661
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1985888263865707e-09,
+      "loss": 0.7275,
+      "step": 6662
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1248489041337796e-09,
+      "loss": 0.7322,
+      "step": 6663
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0534497590719562e-09,
+      "loss": 0.7816,
+      "step": 6664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.843914079155082e-10,
+      "loss": 0.7608,
+      "step": 6665
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.176738668303931e-10,
+      "loss": 0.7548,
+      "step": 6666
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.532971514374489e-10,
+      "loss": 0.6891,
+      "step": 6667
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.912612768057326e-10,
+      "loss": 0.6295,
+      "step": 6668
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.31566257459182e-10,
+      "loss": 0.7287,
+      "step": 6669
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.742121073710639e-10,
+      "loss": 0.7427,
+      "step": 6670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.191988399706361e-10,
+      "loss": 0.7033,
+      "step": 6671
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.665264681342653e-10,
+      "loss": 0.7551,
+      "step": 6672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.161950041943087e-10,
+      "loss": 0.7075,
+      "step": 6673
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.682044599335633e-10,
+      "loss": 0.7938,
+      "step": 6674
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.22554846586376e-10,
+      "loss": 0.6667,
+      "step": 6675
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.792461748397536e-10,
+      "loss": 0.672,
+      "step": 6676
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.382784548311424e-10,
+      "loss": 0.6664,
+      "step": 6677
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.996516961539797e-10,
+      "loss": 0.6345,
+      "step": 6678
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6336590784770135e-10,
+      "loss": 0.73,
+      "step": 6679
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.294210984099543e-10,
+      "loss": 0.7788,
+      "step": 6680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.978172757854946e-10,
+      "loss": 0.6894,
+      "step": 6681
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6855444737284842e-10,
+      "loss": 0.6607,
+      "step": 6682
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.416326200232021e-10,
+      "loss": 0.7778,
+      "step": 6683
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1705180003929173e-10,
+      "loss": 0.7133,
+      "step": 6684
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.481199317540324e-11,
+      "loss": 0.6446,
+      "step": 6685
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.491320463737239e-11,
+      "loss": 0.7332,
+      "step": 6686
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.735543908480523e-11,
+      "loss": 0.6791,
+      "step": 6687
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.213870062663716e-11,
+      "loss": 0.7622,
+      "step": 6688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.926299282557388e-11,
+      "loss": 0.6584,
+      "step": 6689
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8728318696981106e-11,
+      "loss": 0.7222,
+      "step": 6690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0534680705553967e-11,
+      "loss": 0.7503,
+      "step": 6691
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.682080770868069e-12,
+      "loss": 0.6713,
+      "step": 6692
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1705202607181775e-12,
+      "loss": 0.461,
+      "step": 6693
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.4684,
+      "step": 6694
+    },
+    {
+      "epoch": 1.0,
+      "step": 6694,
+      "total_flos": 9260148522483712.0,
+      "train_loss": 0.7489719174853502,
+      "train_runtime": 47264.4408,
+      "train_samples_per_second": 18.127,
+      "train_steps_per_second": 0.142
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 6694,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 2000,
+  "total_flos": 9260148522483712.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}