diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,43956 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999317079833367,
+  "eval_steps": 500,
+  "global_step": 7321,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.545454545454545e-07,
+      "loss": 2.1103,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.09090909090909e-07,
+      "loss": 2.1609,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3636363636363636e-06,
+      "loss": 2.0738,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.818181818181818e-06,
+      "loss": 2.0953,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2727272727272728e-06,
+      "loss": 2.0827,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 1.9101,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1818181818181817e-06,
+      "loss": 2.0924,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.636363636363636e-06,
+      "loss": 2.0125,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0909090909090915e-06,
+      "loss": 1.9581,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 1.9089,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 0.5988,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 2.0005,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.909090909090909e-06,
+      "loss": 1.959,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.363636363636363e-06,
+      "loss": 1.8246,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.818181818181818e-06,
+      "loss": 1.8106,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.272727272727272e-06,
+      "loss": 0.6463,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.727272727272727e-06,
+      "loss": 1.7463,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 1.8108,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.636363636363637e-06,
+      "loss": 1.7236,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 1.6741,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.545454545454547e-06,
+      "loss": 1.6477,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4955,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0454545454545455e-05,
+      "loss": 1.526,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 1.4734,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1363636363636365e-05,
+      "loss": 1.5265,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1818181818181819e-05,
+      "loss": 1.4022,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2272727272727273e-05,
+      "loss": 1.37,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2727272727272727e-05,
+      "loss": 1.2933,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.318181818181818e-05,
+      "loss": 1.3234,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 1.3561,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.409090909090909e-05,
+      "loss": 1.3005,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4545454545454545e-05,
+      "loss": 1.3756,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 1.2263,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 1.2206,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.590909090909091e-05,
+      "loss": 1.2433,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 1.2993,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6818181818181818e-05,
+      "loss": 1.2719,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 1.2642,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.772727272727273e-05,
+      "loss": 1.2562,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 1.221,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8636363636363638e-05,
+      "loss": 0.6383,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9090909090909094e-05,
+      "loss": 1.1582,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9545454545454546e-05,
+      "loss": 1.1923,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 1.15,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0454545454545457e-05,
+      "loss": 1.1962,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.090909090909091e-05,
+      "loss": 1.1671,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1363636363636362e-05,
+      "loss": 1.1158,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1818181818181818e-05,
+      "loss": 1.1949,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2272727272727274e-05,
+      "loss": 1.0462,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 1.1443,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.318181818181818e-05,
+      "loss": 1.1384,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3636363636363637e-05,
+      "loss": 1.0551,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4090909090909093e-05,
+      "loss": 1.095,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4545454545454545e-05,
+      "loss": 1.076,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 1.064,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5454545454545454e-05,
+      "loss": 1.116,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.590909090909091e-05,
+      "loss": 1.0715,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.636363636363636e-05,
+      "loss": 1.0302,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.681818181818182e-05,
+      "loss": 1.083,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 1.0223,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.772727272727273e-05,
+      "loss": 1.0323,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.818181818181818e-05,
+      "loss": 1.0862,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.863636363636364e-05,
+      "loss": 1.0923,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.909090909090909e-05,
+      "loss": 0.9805,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.954545454545455e-05,
+      "loss": 1.0041,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 1.0307,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0454545454545456e-05,
+      "loss": 1.0501,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.090909090909091e-05,
+      "loss": 0.9571,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1363636363636365e-05,
+      "loss": 0.9571,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 0.9556,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2272727272727276e-05,
+      "loss": 0.9869,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.272727272727273e-05,
+      "loss": 1.0386,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.318181818181819e-05,
+      "loss": 0.9886,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3636363636363636e-05,
+      "loss": 1.0372,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.409090909090909e-05,
+      "loss": 0.961,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.454545454545455e-05,
+      "loss": 0.62,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-05,
+      "loss": 0.9605,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.545454545454546e-05,
+      "loss": 0.9954,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.590909090909091e-05,
+      "loss": 0.9707,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.9171,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.681818181818182e-05,
+      "loss": 0.9905,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7272727272727276e-05,
+      "loss": 0.9556,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7727272727272725e-05,
+      "loss": 0.9743,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.818181818181819e-05,
+      "loss": 0.9797,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8636363636363636e-05,
+      "loss": 0.9477,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.909090909090909e-05,
+      "loss": 0.5948,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.954545454545455e-05,
+      "loss": 0.8993,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 0.8877,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.045454545454546e-05,
+      "loss": 1.0158,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0909090909090915e-05,
+      "loss": 0.9701,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1363636363636364e-05,
+      "loss": 0.9439,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.181818181818182e-05,
+      "loss": 0.9236,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2272727272727275e-05,
+      "loss": 0.5656,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2727272727272724e-05,
+      "loss": 0.9129,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.318181818181819e-05,
+      "loss": 0.9633,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3636363636363636e-05,
+      "loss": 0.5078,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.409090909090909e-05,
+      "loss": 0.8773,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.454545454545455e-05,
+      "loss": 0.9151,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9149,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.8807,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5909090909090914e-05,
+      "loss": 0.8459,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.636363636363636e-05,
+      "loss": 0.8955,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.681818181818182e-05,
+      "loss": 0.8688,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7272727272727275e-05,
+      "loss": 0.8953,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.772727272727273e-05,
+      "loss": 0.938,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8181818181818186e-05,
+      "loss": 0.9157,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.863636363636364e-05,
+      "loss": 0.903,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.909090909090909e-05,
+      "loss": 1.0087,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9545454545454553e-05,
+      "loss": 0.8019,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 0.8173,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.045454545454545e-05,
+      "loss": 0.9318,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.090909090909091e-05,
+      "loss": 0.825,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.136363636363637e-05,
+      "loss": 0.8584,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.181818181818182e-05,
+      "loss": 0.8864,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2272727272727274e-05,
+      "loss": 0.8753,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.272727272727272e-05,
+      "loss": 0.9222,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3181818181818186e-05,
+      "loss": 0.8762,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.363636363636364e-05,
+      "loss": 0.8877,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.409090909090909e-05,
+      "loss": 0.8498,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 0.8863,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 0.8463,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.545454545454546e-05,
+      "loss": 0.8595,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5909090909090913e-05,
+      "loss": 0.8434,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.636363636363636e-05,
+      "loss": 0.9888,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6818181818181825e-05,
+      "loss": 0.8154,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.727272727272728e-05,
+      "loss": 0.8472,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.772727272727273e-05,
+      "loss": 0.8286,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.818181818181818e-05,
+      "loss": 0.8247,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8636363636363634e-05,
+      "loss": 0.8434,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.90909090909091e-05,
+      "loss": 0.8632,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9545454545454546e-05,
+      "loss": 0.8938,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-05,
+      "loss": 0.8538,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.045454545454545e-05,
+      "loss": 0.8542,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.090909090909091e-05,
+      "loss": 0.8695,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.136363636363636e-05,
+      "loss": 0.8521,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.181818181818182e-05,
+      "loss": 0.8897,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.227272727272727e-05,
+      "loss": 0.8776,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.272727272727273e-05,
+      "loss": 0.8318,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.318181818181818e-05,
+      "loss": 0.8497,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.363636363636364e-05,
+      "loss": 0.8802,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.40909090909091e-05,
+      "loss": 0.8482,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.454545454545455e-05,
+      "loss": 0.8554,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 0.8114,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.545454545454546e-05,
+      "loss": 0.8064,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.59090909090909e-05,
+      "loss": 0.8651,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.636363636363638e-05,
+      "loss": 0.7924,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.681818181818183e-05,
+      "loss": 0.8132,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.727272727272727e-05,
+      "loss": 0.8517,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.772727272727273e-05,
+      "loss": 0.8439,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.818181818181818e-05,
+      "loss": 0.8384,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.863636363636364e-05,
+      "loss": 0.819,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.90909090909091e-05,
+      "loss": 0.4539,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.954545454545455e-05,
+      "loss": 0.8428,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7e-05,
+      "loss": 0.8558,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.045454545454546e-05,
+      "loss": 0.8997,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.090909090909092e-05,
+      "loss": 0.8689,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.136363636363636e-05,
+      "loss": 0.8799,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.181818181818182e-05,
+      "loss": 0.4823,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.227272727272729e-05,
+      "loss": 0.8524,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 0.926,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.318181818181818e-05,
+      "loss": 0.8409,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.363636363636364e-05,
+      "loss": 0.8955,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.40909090909091e-05,
+      "loss": 0.856,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.454545454545455e-05,
+      "loss": 0.8426,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.8696,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.545454545454545e-05,
+      "loss": 0.8065,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.59090909090909e-05,
+      "loss": 0.814,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.636363636363637e-05,
+      "loss": 0.9308,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.681818181818182e-05,
+      "loss": 0.8207,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.727272727272727e-05,
+      "loss": 0.8389,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.772727272727273e-05,
+      "loss": 0.8733,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.818181818181818e-05,
+      "loss": 0.8679,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.863636363636364e-05,
+      "loss": 0.8303,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.90909090909091e-05,
+      "loss": 0.7991,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.954545454545455e-05,
+      "loss": 0.8232,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8e-05,
+      "loss": 0.8432,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.045454545454546e-05,
+      "loss": 0.7647,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.090909090909092e-05,
+      "loss": 0.4856,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.136363636363636e-05,
+      "loss": 0.7944,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.181818181818183e-05,
+      "loss": 0.8387,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.227272727272729e-05,
+      "loss": 0.833,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.272727272727273e-05,
+      "loss": 0.8389,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.318181818181818e-05,
+      "loss": 0.7823,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.363636363636364e-05,
+      "loss": 0.8415,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.40909090909091e-05,
+      "loss": 0.7843,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.454545454545455e-05,
+      "loss": 0.8177,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.5e-05,
+      "loss": 0.7322,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.545454545454545e-05,
+      "loss": 0.8011,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.590909090909092e-05,
+      "loss": 0.4671,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.636363636363637e-05,
+      "loss": 0.784,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.681818181818182e-05,
+      "loss": 0.8401,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.727272727272727e-05,
+      "loss": 0.7895,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.772727272727274e-05,
+      "loss": 0.8108,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.818181818181818e-05,
+      "loss": 0.7831,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.863636363636364e-05,
+      "loss": 0.8384,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.90909090909091e-05,
+      "loss": 0.8302,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.954545454545455e-05,
+      "loss": 0.8118,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9e-05,
+      "loss": 0.7676,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.045454545454546e-05,
+      "loss": 0.7904,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 0.7582,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.136363636363637e-05,
+      "loss": 0.802,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.181818181818183e-05,
+      "loss": 0.426,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.227272727272727e-05,
+      "loss": 0.7976,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.272727272727273e-05,
+      "loss": 0.803,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.318181818181818e-05,
+      "loss": 0.7461,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.363636363636364e-05,
+      "loss": 0.8091,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.40909090909091e-05,
+      "loss": 0.7707,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.454545454545455e-05,
+      "loss": 0.7711,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.5e-05,
+      "loss": 0.7898,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.545454545454546e-05,
+      "loss": 0.7553,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.590909090909092e-05,
+      "loss": 0.4562,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.636363636363637e-05,
+      "loss": 0.7711,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.681818181818181e-05,
+      "loss": 0.7847,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.727272727272728e-05,
+      "loss": 0.8014,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.772727272727274e-05,
+      "loss": 0.4386,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.818181818181818e-05,
+      "loss": 0.7585,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.863636363636364e-05,
+      "loss": 0.8378,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.909090909090911e-05,
+      "loss": 0.848,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.954545454545455e-05,
+      "loss": 0.792,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 0.8056,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999510671262e-05,
+      "loss": 0.8619,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999998042685148e-05,
+      "loss": 0.4141,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999559604194e-05,
+      "loss": 0.819,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999217074212e-05,
+      "loss": 0.431,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999987766786359e-05,
+      "loss": 0.8047,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999982384175517e-05,
+      "loss": 0.9006,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999976022910649e-05,
+      "loss": 0.8017,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999968682993e-05,
+      "loss": 0.7893,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999960364424007e-05,
+      "loss": 0.8334,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999951067205296e-05,
+      "loss": 0.7822,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999940791338688e-05,
+      "loss": 0.794,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999929536826196e-05,
+      "loss": 0.4415,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99991730367002e-05,
+      "loss": 0.7842,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999904091872556e-05,
+      "loss": 0.8352,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99988990143639e-05,
+      "loss": 0.8968,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9998747323643e-05,
+      "loss": 0.7828,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999858584659254e-05,
+      "loss": 0.8205,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999841458324413e-05,
+      "loss": 0.7945,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999823353363129e-05,
+      "loss": 0.7695,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999804269778948e-05,
+      "loss": 0.8219,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999784207575599e-05,
+      "loss": 0.4221,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999763166757016e-05,
+      "loss": 0.7901,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999741147327314e-05,
+      "loss": 0.6677,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999718149290802e-05,
+      "loss": 0.8085,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999694172651984e-05,
+      "loss": 0.8209,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99966921741555e-05,
+      "loss": 0.7637,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999643283586388e-05,
+      "loss": 0.7958,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999616371169571e-05,
+      "loss": 0.7887,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999588480170369e-05,
+      "loss": 0.8017,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999559610594239e-05,
+      "loss": 0.7753,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999529762446833e-05,
+      "loss": 0.8068,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999498935733993e-05,
+      "loss": 0.7714,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999467130461752e-05,
+      "loss": 0.7635,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999434346636336e-05,
+      "loss": 0.8382,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999400584264163e-05,
+      "loss": 0.4593,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999365843351838e-05,
+      "loss": 0.813,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999330123906164e-05,
+      "loss": 0.765,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99929342593413e-05,
+      "loss": 0.8213,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999255749442924e-05,
+      "loss": 0.7451,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999217094439914e-05,
+      "loss": 0.7408,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99917746093267e-05,
+      "loss": 0.9227,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999136848928947e-05,
+      "loss": 0.7786,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999095258436697e-05,
+      "loss": 0.8235,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999052689464059e-05,
+      "loss": 0.8374,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999009142019364e-05,
+      "loss": 0.7674,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998964616111137e-05,
+      "loss": 0.8147,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998919111748093e-05,
+      "loss": 0.8023,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998872628939139e-05,
+      "loss": 0.7687,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998825167693372e-05,
+      "loss": 0.7878,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998776728020083e-05,
+      "loss": 0.7991,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99872730992875e-05,
+      "loss": 0.7322,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99867691342905e-05,
+      "loss": 0.7849,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998625538530846e-05,
+      "loss": 0.8295,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998573185244192e-05,
+      "loss": 0.7252,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998519853579335e-05,
+      "loss": 0.8065,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998465543546718e-05,
+      "loss": 0.8323,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998410255156965e-05,
+      "loss": 0.7378,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998353988420902e-05,
+      "loss": 0.7921,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99829674334954e-05,
+      "loss": 0.8383,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998238519954086e-05,
+      "loss": 0.7341,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998179318245933e-05,
+      "loss": 0.7936,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998119138236672e-05,
+      "loss": 0.7585,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99805797993808e-05,
+      "loss": 0.7768,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997995843362129e-05,
+      "loss": 0.7788,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997932728520978e-05,
+      "loss": 0.7697,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997868635426983e-05,
+      "loss": 0.7779,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997803564092689e-05,
+      "loss": 0.779,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997737514530834e-05,
+      "loss": 0.8065,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997670486754342e-05,
+      "loss": 0.7862,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997602480776335e-05,
+      "loss": 0.7937,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997533496610125e-05,
+      "loss": 0.7175,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997463534269213e-05,
+      "loss": 0.7634,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997392593767292e-05,
+      "loss": 0.8047,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997320675118247e-05,
+      "loss": 0.7149,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997247778336158e-05,
+      "loss": 0.8137,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99717390343529e-05,
+      "loss": 0.7329,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997099050430104e-05,
+      "loss": 0.7888,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997023219335248e-05,
+      "loss": 0.7208,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996946410165571e-05,
+      "loss": 0.7323,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996868622936102e-05,
+      "loss": 0.793,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996789857662068e-05,
+      "loss": 0.789,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996710114358885e-05,
+      "loss": 0.813,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996629393042163e-05,
+      "loss": 0.7798,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996547693727698e-05,
+      "loss": 0.7646,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996465016431486e-05,
+      "loss": 0.7827,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996381361169706e-05,
+      "loss": 0.7125,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996296727958734e-05,
+      "loss": 0.8258,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996211116815133e-05,
+      "loss": 0.7821,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996124527755663e-05,
+      "loss": 0.7388,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99603696079727e-05,
+      "loss": 0.7566,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995948415957095e-05,
+      "loss": 0.738,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995858893252466e-05,
+      "loss": 0.7593,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99576839270091e-05,
+      "loss": 0.7628,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995676914320136e-05,
+      "loss": 0.8311,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995584458128054e-05,
+      "loss": 0.7835,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995491024142757e-05,
+      "loss": 0.7361,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995396612382535e-05,
+      "loss": 0.7384,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995301222865865e-05,
+      "loss": 0.7823,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99520485561142e-05,
+      "loss": 0.77,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995107510638062e-05,
+      "loss": 0.7664,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995009187964844e-05,
+      "loss": 0.7792,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994909887611009e-05,
+      "loss": 0.7368,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994809609595995e-05,
+      "loss": 0.7154,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99470835393943e-05,
+      "loss": 0.7566,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994606120661133e-05,
+      "loss": 0.7643,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994502909781114e-05,
+      "loss": 0.7283,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994398721319573e-05,
+      "loss": 0.6962,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994293555296904e-05,
+      "loss": 0.718,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994187411733693e-05,
+      "loss": 0.7698,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994080290650713e-05,
+      "loss": 0.7846,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993972192068932e-05,
+      "loss": 0.759,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99386311600951e-05,
+      "loss": 0.7851,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993753062493794e-05,
+      "loss": 0.7541,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993642031543326e-05,
+      "loss": 0.8265,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993530023179838e-05,
+      "loss": 0.7606,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993417037425254e-05,
+      "loss": 0.7163,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993303074301689e-05,
+      "loss": 0.7692,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993188133831448e-05,
+      "loss": 0.4432,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993072216037031e-05,
+      "loss": 0.7393,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992955320941123e-05,
+      "loss": 0.7822,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992837448566608e-05,
+      "loss": 0.7526,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992718598936554e-05,
+      "loss": 0.7734,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992598772074226e-05,
+      "loss": 0.7358,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992477968003077e-05,
+      "loss": 0.8353,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992356186746752e-05,
+      "loss": 0.7586,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992233428329088e-05,
+      "loss": 0.7982,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99210969277411e-05,
+      "loss": 0.8168,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991984980106042e-05,
+      "loss": 0.7907,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99185929034929e-05,
+      "loss": 0.7952,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991732623528457e-05,
+      "loss": 0.7742,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991604979668335e-05,
+      "loss": 0.8401,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99147635879391e-05,
+      "loss": 0.7497,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991346760930354e-05,
+      "loss": 0.7403,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991216186103036e-05,
+      "loss": 0.7375,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991084634337511e-05,
+      "loss": 0.7661,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990952105659533e-05,
+      "loss": 0.7424,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990818600095035e-05,
+      "loss": 0.7686,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990684117670152e-05,
+      "loss": 0.7826,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990548658411207e-05,
+      "loss": 0.7118,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990412222344714e-05,
+      "loss": 0.7247,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990274809497375e-05,
+      "loss": 0.7531,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990136419896088e-05,
+      "loss": 0.7422,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989997053567941e-05,
+      "loss": 0.7784,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989856710540209e-05,
+      "loss": 0.7769,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989715390840366e-05,
+      "loss": 0.7471,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989573094496069e-05,
+      "loss": 0.783,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989429821535172e-05,
+      "loss": 0.7715,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989285571985718e-05,
+      "loss": 0.8011,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989140345875939e-05,
+      "loss": 0.3936,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988994143234262e-05,
+      "loss": 0.716,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988846964089305e-05,
+      "loss": 0.7353,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988698808469874e-05,
+      "loss": 0.7911,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988549676404965e-05,
+      "loss": 0.7463,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988399567923774e-05,
+      "loss": 0.7517,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988248483055677e-05,
+      "loss": 0.7288,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988096421830246e-05,
+      "loss": 0.7411,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987943384277247e-05,
+      "loss": 0.6704,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987789370426635e-05,
+      "loss": 0.7279,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98763438030855e-05,
+      "loss": 0.7917,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987478413953333e-05,
+      "loss": 0.7631,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98732147139151e-05,
+      "loss": 0.7705,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987163552653802e-05,
+      "loss": 0.745,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987004657771115e-05,
+      "loss": 0.7264,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986844786774552e-05,
+      "loss": 0.71,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986683939695403e-05,
+      "loss": 0.7291,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986522116565153e-05,
+      "loss": 0.8262,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986359317415475e-05,
+      "loss": 0.7993,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986195542278231e-05,
+      "loss": 0.7208,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986030791185484e-05,
+      "loss": 0.7667,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985865064169473e-05,
+      "loss": 0.7452,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985698361262642e-05,
+      "loss": 0.7425,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985530682497616e-05,
+      "loss": 0.8215,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985362027907219e-05,
+      "loss": 0.7281,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985192397524459e-05,
+      "loss": 0.8274,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985021791382537e-05,
+      "loss": 0.8046,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98485020951485e-05,
+      "loss": 0.7262,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984677651954978e-05,
+      "loss": 0.864,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984504118736698e-05,
+      "loss": 0.7937,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984329609893976e-05,
+      "loss": 0.72,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984154125460969e-05,
+      "loss": 0.7685,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983977665472024e-05,
+      "loss": 0.7646,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983800229961678e-05,
+      "loss": 0.7913,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983621818964666e-05,
+      "loss": 0.7841,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983442432515904e-05,
+      "loss": 0.7395,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983262070650504e-05,
+      "loss": 0.7288,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98308073340377e-05,
+      "loss": 0.7775,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982898420811195e-05,
+      "loss": 0.7487,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982715132908464e-05,
+      "loss": 0.7452,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982530869731451e-05,
+      "loss": 0.8021,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982345631316223e-05,
+      "loss": 0.727,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982159417699036e-05,
+      "loss": 0.7777,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981972228916339e-05,
+      "loss": 0.7357,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981784065004768e-05,
+      "loss": 0.7427,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981594926001157e-05,
+      "loss": 0.7567,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981404811942525e-05,
+      "loss": 0.7481,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981213722866081e-05,
+      "loss": 0.7498,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981021658809229e-05,
+      "loss": 0.7961,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980828619809561e-05,
+      "loss": 0.7577,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980634605904862e-05,
+      "loss": 0.8006,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980439617133106e-05,
+      "loss": 0.8028,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980243653532459e-05,
+      "loss": 0.8006,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980046715141277e-05,
+      "loss": 0.8064,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979848801998106e-05,
+      "loss": 0.7422,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979649914141686e-05,
+      "loss": 0.7431,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979450051610942e-05,
+      "loss": 0.7845,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979249214444997e-05,
+      "loss": 0.7791,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97904740268316e-05,
+      "loss": 0.7351,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978844616364933e-05,
+      "loss": 0.7565,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978640855530003e-05,
+      "loss": 0.7304,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97843612021826e-05,
+      "loss": 0.8589,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978230410469771e-05,
+      "loss": 0.7515,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978023726324801e-05,
+      "loss": 0.6904,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977816067823807e-05,
+      "loss": 0.7267,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977607435007432e-05,
+      "loss": 0.7802,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977397827916512e-05,
+      "loss": 0.7847,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977187246592076e-05,
+      "loss": 0.7406,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976975691075338e-05,
+      "loss": 0.6969,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97676316140771e-05,
+      "loss": 0.7401,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97654965763079e-05,
+      "loss": 0.7301,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976335179786363e-05,
+      "loss": 0.7667,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976119727916414e-05,
+      "loss": 0.7787,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97590330206311e-05,
+      "loss": 0.7521,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975685902268816e-05,
+      "loss": 0.8123,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975467528576082e-05,
+      "loss": 0.6829,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97524818102765e-05,
+      "loss": 0.7561,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975027859666455e-05,
+      "loss": 0.7255,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97480656453562e-05,
+      "loss": 0.6976,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974584295678459e-05,
+      "loss": 0.6753,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974361053138478e-05,
+      "loss": 0.7023,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974136836959372e-05,
+      "loss": 0.7814,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973911647185026e-05,
+      "loss": 0.684,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973685483859518e-05,
+      "loss": 0.7011,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973458347027115e-05,
+      "loss": 0.7506,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973230236732276e-05,
+      "loss": 0.7429,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973001153019648e-05,
+      "loss": 0.725,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97277109593407e-05,
+      "loss": 0.8182,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972540065520572e-05,
+      "loss": 0.7595,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972308061824375e-05,
+      "loss": 0.8095,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972075084890886e-05,
+      "loss": 0.8105,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971841134765709e-05,
+      "loss": 0.744,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971606211494633e-05,
+      "loss": 0.4501,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971370315123643e-05,
+      "loss": 0.746,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971133445698908e-05,
+      "loss": 0.7421,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970895603266794e-05,
+      "loss": 0.8274,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970656787873851e-05,
+      "loss": 0.7991,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970416999566825e-05,
+      "loss": 0.7582,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97017623839265e-05,
+      "loss": 0.7872,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96993450439845e-05,
+      "loss": 0.7898,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96969179763154e-05,
+      "loss": 0.8012,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969448118139423e-05,
+      "loss": 0.7995,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969203465969801e-05,
+      "loss": 0.7202,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968957841170554e-05,
+      "loss": 0.7939,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96871124378976e-05,
+      "loss": 0.7679,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968463673875687e-05,
+      "loss": 0.7567,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968215131476792e-05,
+      "loss": 0.7797,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967965616641724e-05,
+      "loss": 0.6951,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967715129419319e-05,
+      "loss": 0.7846,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967463669858603e-05,
+      "loss": 0.6895,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9672112380088e-05,
+      "loss": 0.6864,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966957833919314e-05,
+      "loss": 0.7759,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966703457639748e-05,
+      "loss": 0.7496,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966448109219887e-05,
+      "loss": 0.7476,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966191788709716e-05,
+      "loss": 0.6949,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9659344961594e-05,
+      "loss": 0.7379,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965676231619304e-05,
+      "loss": 0.7688,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965416995139973e-05,
+      "loss": 0.7528,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965156786772154e-05,
+      "loss": 0.7403,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964895606566773e-05,
+      "loss": 0.7725,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964633454574953e-05,
+      "loss": 0.6715,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964370330848005e-05,
+      "loss": 0.7656,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964106235437433e-05,
+      "loss": 0.7484,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963841168394926e-05,
+      "loss": 0.762,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963575129772365e-05,
+      "loss": 0.7632,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963308119621826e-05,
+      "loss": 0.7137,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963040137995567e-05,
+      "loss": 0.7506,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962771184946045e-05,
+      "loss": 0.7389,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962501260525898e-05,
+      "loss": 0.7755,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962230364787962e-05,
+      "loss": 0.7088,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96195849778526e-05,
+      "loss": 0.7812,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961685659571e-05,
+      "loss": 0.725,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961411850198592e-05,
+      "loss": 0.7877,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961137069721626e-05,
+      "loss": 0.7524,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960861318193883e-05,
+      "loss": 0.729,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96058459566934e-05,
+      "loss": 0.7291,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960306902202157e-05,
+      "loss": 0.7396,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96002823784669e-05,
+      "loss": 0.7493,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959748602657481e-05,
+      "loss": 0.7996,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959467996689264e-05,
+      "loss": 0.7645,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959186419996962e-05,
+      "loss": 0.6856,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958903872635688e-05,
+      "loss": 0.7795,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958620354660747e-05,
+      "loss": 0.7853,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958335866127631e-05,
+      "loss": 0.7389,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958050407092024e-05,
+      "loss": 0.7093,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957763977609798e-05,
+      "loss": 0.7658,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957476577737019e-05,
+      "loss": 0.7818,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957188207529937e-05,
+      "loss": 0.7582,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956898867044999e-05,
+      "loss": 0.7214,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956608556338834e-05,
+      "loss": 0.8069,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956317275468267e-05,
+      "loss": 0.7654,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95602502449031e-05,
+      "loss": 0.7732,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955731803462168e-05,
+      "loss": 0.73,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95543761244123e-05,
+      "loss": 0.7457,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95514245148508e-05,
+      "loss": 0.7703,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954846320651492e-05,
+      "loss": 0.7225,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954549219998427e-05,
+      "loss": 0.7673,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954251149584036e-05,
+      "loss": 0.6502,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953952109466661e-05,
+      "loss": 0.8072,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953652099704835e-05,
+      "loss": 0.7037,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953351120357277e-05,
+      "loss": 0.4402,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953049171482901e-05,
+      "loss": 0.7169,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952746253140806e-05,
+      "loss": 0.7413,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952442365390283e-05,
+      "loss": 0.7001,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952137508290813e-05,
+      "loss": 0.7318,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951831681902064e-05,
+      "loss": 0.7529,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951524886283899e-05,
+      "loss": 0.7036,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951217121496365e-05,
+      "loss": 0.7296,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950908387599703e-05,
+      "loss": 0.7407,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950598684654342e-05,
+      "loss": 0.7373,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950288012720898e-05,
+      "loss": 0.7402,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949976371860183e-05,
+      "loss": 0.6383,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949663762133192e-05,
+      "loss": 0.7913,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949350183601113e-05,
+      "loss": 0.6947,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949035636325326e-05,
+      "loss": 0.7479,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948720120367394e-05,
+      "loss": 0.7967,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948403635789076e-05,
+      "loss": 0.7744,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948086182652317e-05,
+      "loss": 0.7315,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947767761019252e-05,
+      "loss": 0.7633,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947448370952207e-05,
+      "loss": 0.7398,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947128012513695e-05,
+      "loss": 0.7937,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946806685766425e-05,
+      "loss": 0.7222,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946484390773285e-05,
+      "loss": 0.4355,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946161127597362e-05,
+      "loss": 0.6821,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945836896301927e-05,
+      "loss": 0.7083,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945511696950442e-05,
+      "loss": 0.7099,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945185529606563e-05,
+      "loss": 0.7128,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944858394334125e-05,
+      "loss": 0.7576,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944530291197161e-05,
+      "loss": 0.7887,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944201220259892e-05,
+      "loss": 0.7664,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943871181586728e-05,
+      "loss": 0.7465,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943540175242267e-05,
+      "loss": 0.696,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943208201291296e-05,
+      "loss": 0.7456,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942875259798796e-05,
+      "loss": 0.6868,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942541350829932e-05,
+      "loss": 0.7503,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94220647445006e-05,
+      "loss": 0.718,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941870630724728e-05,
+      "loss": 0.7402,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941533819719668e-05,
+      "loss": 0.7278,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941196041500809e-05,
+      "loss": 0.6762,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94085729613426e-05,
+      "loss": 0.7379,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940517583686327e-05,
+      "loss": 0.7433,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940176904223502e-05,
+      "loss": 0.728,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939835257812468e-05,
+      "loss": 0.7022,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939492644520091e-05,
+      "loss": 0.7884,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93914906441344e-05,
+      "loss": 0.7273,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938804517559755e-05,
+      "loss": 0.6815,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938459004026478e-05,
+      "loss": 0.7868,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938112523881242e-05,
+      "loss": 0.68,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937765077191857e-05,
+      "loss": 0.7787,
+      "step": 577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937416664026333e-05,
+      "loss": 0.7393,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937067284452863e-05,
+      "loss": 0.7365,
+      "step": 579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936716938539834e-05,
+      "loss": 0.6791,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936365626355819e-05,
+      "loss": 0.788,
+      "step": 581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936013347969581e-05,
+      "loss": 0.7829,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935660103450071e-05,
+      "loss": 0.6746,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935305892866431e-05,
+      "loss": 0.7106,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93495071628799e-05,
+      "loss": 0.708,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93459457378427e-05,
+      "loss": 0.7804,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934237465424976e-05,
+      "loss": 0.7015,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933879391280007e-05,
+      "loss": 0.7316,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93352035141945e-05,
+      "loss": 0.7242,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933160345913579e-05,
+      "loss": 0.7304,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932799374832857e-05,
+      "loss": 0.4013,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932437438247942e-05,
+      "loss": 0.7999,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932074536229674e-05,
+      "loss": 0.7656,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931710668849082e-05,
+      "loss": 0.6348,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93134583617739e-05,
+      "loss": 0.7502,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930980038286004e-05,
+      "loss": 0.7751,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930613275246526e-05,
+      "loss": 0.7593,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93024554713074e-05,
+      "loss": 0.735,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929876854010623e-05,
+      "loss": 0.714,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92950719595834e-05,
+      "loss": 0.6931,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929136573046244e-05,
+      "loss": 0.6839,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928764985346878e-05,
+      "loss": 0.7165,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928392432932974e-05,
+      "loss": 0.8065,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928018915877451e-05,
+      "loss": 0.7082,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927644434253419e-05,
+      "loss": 0.6791,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927268988134176e-05,
+      "loss": 0.7419,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926892577593208e-05,
+      "loss": 0.7247,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926515202704192e-05,
+      "loss": 0.8384,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926136863540989e-05,
+      "loss": 0.7004,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925757560177654e-05,
+      "loss": 0.7678,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925377292688429e-05,
+      "loss": 0.648,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924996061147742e-05,
+      "loss": 0.7607,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924613865630216e-05,
+      "loss": 0.6826,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924230706210656e-05,
+      "loss": 0.411,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923846582964056e-05,
+      "loss": 0.755,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923461495965607e-05,
+      "loss": 0.6948,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923075445290677e-05,
+      "loss": 0.762,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922688431014833e-05,
+      "loss": 0.7495,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922300453213822e-05,
+      "loss": 0.7139,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921911511963584e-05,
+      "loss": 0.7015,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921521607340249e-05,
+      "loss": 0.6976,
+      "step": 621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921130739420134e-05,
+      "loss": 0.7372,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920738908279742e-05,
+      "loss": 0.7467,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920346113995766e-05,
+      "loss": 0.7354,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919952356645092e-05,
+      "loss": 0.6972,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919557636304788e-05,
+      "loss": 0.739,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919161953052114e-05,
+      "loss": 0.6915,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918765306964515e-05,
+      "loss": 0.7848,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918367698119632e-05,
+      "loss": 0.724,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917969126595285e-05,
+      "loss": 0.7293,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917569592469492e-05,
+      "loss": 0.7501,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917169095820449e-05,
+      "loss": 0.7061,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91676763672655e-05,
+      "loss": 0.6858,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91636521526637e-05,
+      "loss": 0.7441,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915961831518676e-05,
+      "loss": 0.6853,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915557485562427e-05,
+      "loss": 0.7753,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915152177476761e-05,
+      "loss": 0.4234,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914745907341012e-05,
+      "loss": 0.7798,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9143386752347e-05,
+      "loss": 0.7653,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913930481237532e-05,
+      "loss": 0.7134,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913521325429407e-05,
+      "loss": 0.7438,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913111207890407e-05,
+      "loss": 0.7522,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912700128700805e-05,
+      "loss": 0.7374,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912288087941062e-05,
+      "loss": 0.7948,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91187508569183e-05,
+      "loss": 0.716,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911461122033944e-05,
+      "loss": 0.7133,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91104619704843e-05,
+      "loss": 0.752,
+      "step": 647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910630310816504e-05,
+      "loss": 0.6754,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910213463419565e-05,
+      "loss": 0.7131,
+      "step": 649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909795654939204e-05,
+      "loss": 0.7491,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909376885457202e-05,
+      "loss": 0.7252,
+      "step": 651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908957155055523e-05,
+      "loss": 0.7079,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90853646381632e-05,
+      "loss": 0.7041,
+      "step": 653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908114811821939e-05,
+      "loss": 0.774,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907692199154908e-05,
+      "loss": 0.712,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907268625897947e-05,
+      "loss": 0.7262,
+      "step": 656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906844092133961e-05,
+      "loss": 0.7498,
+      "step": 657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906418597946045e-05,
+      "loss": 0.7191,
+      "step": 658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905992143417484e-05,
+      "loss": 0.6724,
+      "step": 659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905564728631746e-05,
+      "loss": 0.7351,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905136353672492e-05,
+      "loss": 0.6777,
+      "step": 661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904707018623566e-05,
+      "loss": 0.7813,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904276723569002e-05,
+      "loss": 0.7197,
+      "step": 663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903845468593025e-05,
+      "loss": 0.7363,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903413253780042e-05,
+      "loss": 0.711,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902980079214656e-05,
+      "loss": 0.6952,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902545944981647e-05,
+      "loss": 0.7092,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902110851165992e-05,
+      "loss": 0.7169,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901674797852852e-05,
+      "loss": 0.705,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901237785127577e-05,
+      "loss": 0.6674,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900799813075702e-05,
+      "loss": 0.6939,
+      "step": 671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900360881782955e-05,
+      "loss": 0.7543,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899920991335248e-05,
+      "loss": 0.7694,
+      "step": 673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899480141818679e-05,
+      "loss": 0.7015,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899038333319537e-05,
+      "loss": 0.7666,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8985955659243e-05,
+      "loss": 0.7107,
+      "step": 676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89815183971963e-05,
+      "loss": 0.7821,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897707154792377e-05,
+      "loss": 0.7294,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897261511229583e-05,
+      "loss": 0.7809,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89681490911847e-05,
+      "loss": 0.7273,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896367348546458e-05,
+      "loss": 0.7589,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895918829601143e-05,
+      "loss": 0.6767,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895469352370316e-05,
+      "loss": 0.7685,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895018916941956e-05,
+      "loss": 0.7328,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894567523404226e-05,
+      "loss": 0.7546,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894115171845478e-05,
+      "loss": 0.7415,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893661862354249e-05,
+      "loss": 0.7202,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893207595019269e-05,
+      "loss": 0.74,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892752369929454e-05,
+      "loss": 0.6605,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892296187173902e-05,
+      "loss": 0.4152,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891839046841904e-05,
+      "loss": 0.7224,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891380949022934e-05,
+      "loss": 0.7215,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890921893806662e-05,
+      "loss": 0.7513,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890461881282936e-05,
+      "loss": 0.7494,
+      "step": 694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890000911541795e-05,
+      "loss": 0.7304,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889538984673465e-05,
+      "loss": 0.7141,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889076100768362e-05,
+      "loss": 0.7243,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888612259917082e-05,
+      "loss": 0.7158,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88814746221042e-05,
+      "loss": 0.733,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887681707739346e-05,
+      "loss": 0.7795,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887214996595026e-05,
+      "loss": 0.715,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886747328868807e-05,
+      "loss": 0.6828,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88627870465223e-05,
+      "loss": 0.6929,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885809124037018e-05,
+      "loss": 0.7131,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885338587115081e-05,
+      "loss": 0.7719,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884867093978519e-05,
+      "loss": 0.4337,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884394644719618e-05,
+      "loss": 0.7431,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883921239430853e-05,
+      "loss": 0.7296,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883446878204883e-05,
+      "loss": 0.6806,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882971561134556e-05,
+      "loss": 0.7893,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882495288312905e-05,
+      "loss": 0.7283,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882018059833152e-05,
+      "loss": 0.7208,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881539875788706e-05,
+      "loss": 0.796,
+      "step": 713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881060736273164e-05,
+      "loss": 0.7875,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880580641380306e-05,
+      "loss": 0.7477,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880099591204104e-05,
+      "loss": 0.6899,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879617585838713e-05,
+      "loss": 0.7204,
+      "step": 717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879134625378478e-05,
+      "loss": 0.7542,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87865070991793e-05,
+      "loss": 0.7358,
+      "step": 719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878165839551784e-05,
+      "loss": 0.7472,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877680014374947e-05,
+      "loss": 0.698,
+      "step": 721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877193234482508e-05,
+      "loss": 0.7151,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876705499969748e-05,
+      "loss": 0.7955,
+      "step": 723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876216810932129e-05,
+      "loss": 0.7436,
+      "step": 724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875727167465305e-05,
+      "loss": 0.7011,
+      "step": 725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875236569665114e-05,
+      "loss": 0.7104,
+      "step": 726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87474501762758e-05,
+      "loss": 0.7937,
+      "step": 727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874252511448919e-05,
+      "loss": 0.7504,
+      "step": 728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873759051225524e-05,
+      "loss": 0.7172,
+      "step": 729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873264637053986e-05,
+      "loss": 0.7735,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872769269031076e-05,
+      "loss": 0.7342,
+      "step": 731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872272947253752e-05,
+      "loss": 0.7147,
+      "step": 732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871775671819162e-05,
+      "loss": 0.6929,
+      "step": 733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871277442824636e-05,
+      "loss": 0.7254,
+      "step": 734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870778260367693e-05,
+      "loss": 0.6949,
+      "step": 735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870278124546042e-05,
+      "loss": 0.68,
+      "step": 736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869777035457572e-05,
+      "loss": 0.7499,
+      "step": 737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869274993200363e-05,
+      "loss": 0.4259,
+      "step": 738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868771997872681e-05,
+      "loss": 0.681,
+      "step": 739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868268049572977e-05,
+      "loss": 0.7669,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867763148399891e-05,
+      "loss": 0.7329,
+      "step": 741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867257294452245e-05,
+      "loss": 0.7178,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866750487829056e-05,
+      "loss": 0.7063,
+      "step": 743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866242728629518e-05,
+      "loss": 0.7444,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865734016953015e-05,
+      "loss": 0.7292,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865224352899119e-05,
+      "loss": 0.7626,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864713736567589e-05,
+      "loss": 0.6403,
+      "step": 747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864202168058366e-05,
+      "loss": 0.7214,
+      "step": 748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863689647471581e-05,
+      "loss": 0.7737,
+      "step": 749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863176174907552e-05,
+      "loss": 0.7121,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86266175046678e-05,
+      "loss": 0.6942,
+      "step": 751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862146374249954e-05,
+      "loss": 0.7437,
+      "step": 752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86163004635795e-05,
+      "loss": 0.6833,
+      "step": 753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861112766891829e-05,
+      "loss": 0.8154,
+      "step": 754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86059453595284e-05,
+      "loss": 0.7794,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860075353642415e-05,
+      "loss": 0.6958,
+      "step": 756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859555220062178e-05,
+      "loss": 0.7101,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859034135313932e-05,
+      "loss": 0.7315,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85851209949967e-05,
+      "loss": 0.7442,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857989112721574e-05,
+      "loss": 0.6449,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857465175082006e-05,
+      "loss": 0.7249,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856940286683519e-05,
+      "loss": 0.719,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856414447628846e-05,
+      "loss": 0.6793,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855887658020916e-05,
+      "loss": 0.707,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855359917962835e-05,
+      "loss": 0.7441,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854831227557899e-05,
+      "loss": 0.6852,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85430158690959e-05,
+      "loss": 0.7006,
+      "step": 767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853770996121573e-05,
+      "loss": 0.7191,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853239455297705e-05,
+      "loss": 0.6835,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852706964542022e-05,
+      "loss": 0.6644,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852173523958752e-05,
+      "loss": 0.7186,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851639133652305e-05,
+      "loss": 0.7463,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851103793727277e-05,
+      "loss": 0.7916,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85056750428845e-05,
+      "loss": 0.7097,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850030265440796e-05,
+      "loss": 0.7169,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849492077289468e-05,
+      "loss": 0.6766,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848952939939806e-05,
+      "loss": 0.7527,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848412853497337e-05,
+      "loss": 0.7042,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84787181806777e-05,
+      "loss": 0.7455,
+      "step": 779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847329833757008e-05,
+      "loss": 0.7989,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84678690067113e-05,
+      "loss": 0.7001,
+      "step": 781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846243018916407e-05,
+      "loss": 0.654,
+      "step": 782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845698188599293e-05,
+      "loss": 0.7256,
+      "step": 783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845152409826428e-05,
+      "loss": 0.7076,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844605682704639e-05,
+      "loss": 0.4247,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844058007340939e-05,
+      "loss": 0.7085,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843509383842525e-05,
+      "loss": 0.6805,
+      "step": 787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842959812316776e-05,
+      "loss": 0.7554,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842409292871264e-05,
+      "loss": 0.7327,
+      "step": 789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841857825613742e-05,
+      "loss": 0.6487,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84130541065215e-05,
+      "loss": 0.6836,
+      "step": 791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840752048094615e-05,
+      "loss": 0.7542,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840197738049443e-05,
+      "loss": 0.748,
+      "step": 793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839642480625133e-05,
+      "loss": 0.7133,
+      "step": 794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839086275930365e-05,
+      "loss": 0.6809,
+      "step": 795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838529124074006e-05,
+      "loss": 0.6725,
+      "step": 796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837971025165109e-05,
+      "loss": 0.7409,
+      "step": 797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837411979312911e-05,
+      "loss": 0.6986,
+      "step": 798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836851986626835e-05,
+      "loss": 0.7968,
+      "step": 799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83629104721649e-05,
+      "loss": 0.7076,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835729161191667e-05,
+      "loss": 0.7849,
+      "step": 801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835166328662347e-05,
+      "loss": 0.7061,
+      "step": 802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834602549738694e-05,
+      "loss": 0.7573,
+      "step": 803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834037824531056e-05,
+      "loss": 0.7118,
+      "step": 804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833472153149968e-05,
+      "loss": 0.7503,
+      "step": 805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832905535706152e-05,
+      "loss": 0.7379,
+      "step": 806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832337972310508e-05,
+      "loss": 0.7517,
+      "step": 807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831769463074131e-05,
+      "loss": 0.7449,
+      "step": 808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831200008108293e-05,
+      "loss": 0.4291,
+      "step": 809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830629607524457e-05,
+      "loss": 0.7619,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830058261434265e-05,
+      "loss": 0.7386,
+      "step": 811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829485969949549e-05,
+      "loss": 0.7068,
+      "step": 812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828912733182325e-05,
+      "loss": 0.7033,
+      "step": 813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828338551244794e-05,
+      "loss": 0.4238,
+      "step": 814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827763424249339e-05,
+      "loss": 0.6601,
+      "step": 815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827187352308534e-05,
+      "loss": 0.424,
+      "step": 816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82661033553513e-05,
+      "loss": 0.7165,
+      "step": 817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826032374042071e-05,
+      "loss": 0.743,
+      "step": 818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82545346794248e-05,
+      "loss": 0.7189,
+      "step": 819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82487361734967e-05,
+      "loss": 0.7432,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824292822377135e-05,
+      "loss": 0.7082,
+      "step": 821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823711083138552e-05,
+      "loss": 0.7334,
+      "step": 822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823128399747788e-05,
+      "loss": 0.7717,
+      "step": 823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822544772318893e-05,
+      "loss": 0.6896,
+      "step": 824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821960200966101e-05,
+      "loss": 0.7693,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82137468580383e-05,
+      "loss": 0.7576,
+      "step": 826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820788226946684e-05,
+      "loss": 0.6946,
+      "step": 827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820200824509451e-05,
+      "loss": 0.692,
+      "step": 828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819612478607106e-05,
+      "loss": 0.6679,
+      "step": 829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819023189354806e-05,
+      "loss": 0.7342,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818432956867894e-05,
+      "loss": 0.6927,
+      "step": 831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817841781261894e-05,
+      "loss": 0.7397,
+      "step": 832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817249662652522e-05,
+      "loss": 0.7492,
+      "step": 833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816656601155671e-05,
+      "loss": 0.6844,
+      "step": 834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816062596887425e-05,
+      "loss": 0.6898,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815467649964045e-05,
+      "loss": 0.7148,
+      "step": 836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814871760501986e-05,
+      "loss": 0.7377,
+      "step": 837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814274928617878e-05,
+      "loss": 0.7678,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813677154428542e-05,
+      "loss": 0.7282,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81307843805098e-05,
+      "loss": 0.7327,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812478779602381e-05,
+      "loss": 0.7762,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811878179200116e-05,
+      "loss": 0.7268,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811276636961741e-05,
+      "loss": 0.6764,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810674153005e-05,
+      "loss": 0.685,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810070727447813e-05,
+      "loss": 0.7094,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809466360408293e-05,
+      "loss": 0.7672,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808861052004732e-05,
+      "loss": 0.6763,
+      "step": 847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80825480235561e-05,
+      "loss": 0.7181,
+      "step": 848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807647611579586e-05,
+      "loss": 0.7191,
+      "step": 849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80703947979551e-05,
+      "loss": 0.7031,
+      "step": 850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80643040712241e-05,
+      "loss": 0.7585,
+      "step": 851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805820393679501e-05,
+      "loss": 0.6915,
+      "step": 852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805209439586183e-05,
+      "loss": 0.7014,
+      "step": 853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804597544962037e-05,
+      "loss": 0.7211,
+      "step": 854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803984709926832e-05,
+      "loss": 0.7287,
+      "step": 855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803370934600518e-05,
+      "loss": 0.8291,
+      "step": 856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802756219103231e-05,
+      "loss": 0.7802,
+      "step": 857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802140563555289e-05,
+      "loss": 0.7222,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801523968077195e-05,
+      "loss": 0.705,
+      "step": 859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800906432789638e-05,
+      "loss": 0.7424,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800287957813488e-05,
+      "loss": 0.7291,
+      "step": 861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799668543269801e-05,
+      "loss": 0.4063,
+      "step": 862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799048189279814e-05,
+      "loss": 0.7111,
+      "step": 863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798426895964952e-05,
+      "loss": 0.6971,
+      "step": 864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79780466344682e-05,
+      "loss": 0.7168,
+      "step": 865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797181491847209e-05,
+      "loss": 0.7453,
+      "step": 866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796557381288092e-05,
+      "loss": 0.6835,
+      "step": 867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79593233189163e-05,
+      "loss": 0.7241,
+      "step": 868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795306343780163e-05,
+      "loss": 0.7133,
+      "step": 869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794679417076217e-05,
+      "loss": 0.6908,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794051551902501e-05,
+      "loss": 0.6868,
+      "step": 871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793422748381908e-05,
+      "loss": 0.7615,
+      "step": 872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792793006637515e-05,
+      "loss": 0.7071,
+      "step": 873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792162326792582e-05,
+      "loss": 0.7333,
+      "step": 874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791530708970554e-05,
+      "loss": 0.7069,
+      "step": 875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790898153295058e-05,
+      "loss": 0.659,
+      "step": 876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790264659889903e-05,
+      "loss": 0.7759,
+      "step": 877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789630228879085e-05,
+      "loss": 0.6954,
+      "step": 878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788994860386783e-05,
+      "loss": 0.739,
+      "step": 879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788358554537358e-05,
+      "loss": 0.6599,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787721311455355e-05,
+      "loss": 0.7395,
+      "step": 881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787083131265504e-05,
+      "loss": 0.7381,
+      "step": 882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786444014092713e-05,
+      "loss": 0.7231,
+      "step": 883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785803960062081e-05,
+      "loss": 0.4086,
+      "step": 884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785162969298887e-05,
+      "loss": 0.732,
+      "step": 885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78452104192859e-05,
+      "loss": 0.4201,
+      "step": 886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783878178076837e-05,
+      "loss": 0.7276,
+      "step": 887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783234377869459e-05,
+      "loss": 0.6862,
+      "step": 888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782589641432465e-05,
+      "loss": 0.6603,
+      "step": 889
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78194396889205e-05,
+      "loss": 0.6832,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781297360374594e-05,
+      "loss": 0.7196,
+      "step": 891
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780649816006659e-05,
+      "loss": 0.768,
+      "step": 892
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780001335914988e-05,
+      "loss": 0.6615,
+      "step": 893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779351920226509e-05,
+      "loss": 0.743,
+      "step": 894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778701569068336e-05,
+      "loss": 0.7498,
+      "step": 895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778050282567761e-05,
+      "loss": 0.6953,
+      "step": 896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77739806085226e-05,
+      "loss": 0.774,
+      "step": 897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776744904049497e-05,
+      "loss": 0.685,
+      "step": 898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776090812287312e-05,
+      "loss": 0.7603,
+      "step": 899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775435785693733e-05,
+      "loss": 0.7054,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774779824396969e-05,
+      "loss": 0.706,
+      "step": 901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774122928525411e-05,
+      "loss": 0.7972,
+      "step": 902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773465098207637e-05,
+      "loss": 0.7157,
+      "step": 903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772806333572405e-05,
+      "loss": 0.7323,
+      "step": 904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772146634748651e-05,
+      "loss": 0.7602,
+      "step": 905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771486001865504e-05,
+      "loss": 0.8042,
+      "step": 906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770824435052268e-05,
+      "loss": 0.672,
+      "step": 907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770161934438435e-05,
+      "loss": 0.7003,
+      "step": 908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769498500153676e-05,
+      "loss": 0.7121,
+      "step": 909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768834132327844e-05,
+      "loss": 0.7076,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76816883109098e-05,
+      "loss": 0.7008,
+      "step": 911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767502596573301e-05,
+      "loss": 0.683,
+      "step": 912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766835428905213e-05,
+      "loss": 0.7433,
+      "step": 913
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766167328217301e-05,
+      "loss": 0.7303,
+      "step": 914
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765498294640333e-05,
+      "loss": 0.7398,
+      "step": 915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76482832830526e-05,
+      "loss": 0.7099,
+      "step": 916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764157429343214e-05,
+      "loss": 0.675,
+      "step": 917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763485597885513e-05,
+      "loss": 0.7668,
+      "step": 918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762812834063657e-05,
+      "loss": 0.7397,
+      "step": 919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762139138009323e-05,
+      "loss": 0.6898,
+      "step": 920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761464509854377e-05,
+      "loss": 0.6688,
+      "step": 921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760788949730866e-05,
+      "loss": 0.7784,
+      "step": 922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760112457771016e-05,
+      "loss": 0.7772,
+      "step": 923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759435034107238e-05,
+      "loss": 0.7546,
+      "step": 924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758756678872128e-05,
+      "loss": 0.698,
+      "step": 925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758077392198457e-05,
+      "loss": 0.7727,
+      "step": 926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757397174219188e-05,
+      "loss": 0.6917,
+      "step": 927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756716025067456e-05,
+      "loss": 0.6957,
+      "step": 928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756033944876586e-05,
+      "loss": 0.6881,
+      "step": 929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755350933780082e-05,
+      "loss": 0.7457,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754666991911633e-05,
+      "loss": 0.7176,
+      "step": 931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753982119405105e-05,
+      "loss": 0.6934,
+      "step": 932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753296316394549e-05,
+      "loss": 0.7324,
+      "step": 933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7526095830142e-05,
+      "loss": 0.6709,
+      "step": 934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751921919398474e-05,
+      "loss": 0.7412,
+      "step": 935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751233325681965e-05,
+      "loss": 0.6953,
+      "step": 936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750543801999457e-05,
+      "loss": 0.782,
+      "step": 937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749853348485908e-05,
+      "loss": 0.6983,
+      "step": 938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749161965276462e-05,
+      "loss": 0.7878,
+      "step": 939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748469652506446e-05,
+      "loss": 0.6564,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747776410311366e-05,
+      "loss": 0.7747,
+      "step": 941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747082238826912e-05,
+      "loss": 0.7252,
+      "step": 942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746387138188955e-05,
+      "loss": 0.8059,
+      "step": 943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745691108533548e-05,
+      "loss": 0.6778,
+      "step": 944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744994149996925e-05,
+      "loss": 0.6994,
+      "step": 945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744296262715505e-05,
+      "loss": 0.6998,
+      "step": 946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743597446825886e-05,
+      "loss": 0.7301,
+      "step": 947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742897702464846e-05,
+      "loss": 0.7313,
+      "step": 948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74219702976935e-05,
+      "loss": 0.707,
+      "step": 949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741495428876539e-05,
+      "loss": 0.6589,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740792899923741e-05,
+      "loss": 0.7453,
+      "step": 951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740089443048462e-05,
+      "loss": 0.7068,
+      "step": 952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739385058388389e-05,
+      "loss": 0.7297,
+      "step": 953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738679746081396e-05,
+      "loss": 0.6268,
+      "step": 954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737973506265529e-05,
+      "loss": 0.7597,
+      "step": 955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737266339079027e-05,
+      "loss": 0.6508,
+      "step": 956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736558244660303e-05,
+      "loss": 0.6645,
+      "step": 957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735849223147952e-05,
+      "loss": 0.7485,
+      "step": 958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735139274680755e-05,
+      "loss": 0.697,
+      "step": 959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734428399397669e-05,
+      "loss": 0.7122,
+      "step": 960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733716597437832e-05,
+      "loss": 0.689,
+      "step": 961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733003868940572e-05,
+      "loss": 0.7712,
+      "step": 962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732290214045388e-05,
+      "loss": 0.7225,
+      "step": 963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731575632891967e-05,
+      "loss": 0.6566,
+      "step": 964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730860125620174e-05,
+      "loss": 0.7227,
+      "step": 965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730143692370056e-05,
+      "loss": 0.7081,
+      "step": 966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729426333281841e-05,
+      "loss": 0.7138,
+      "step": 967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728708048495943e-05,
+      "loss": 0.43,
+      "step": 968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727988838152949e-05,
+      "loss": 0.4248,
+      "step": 969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727268702393629e-05,
+      "loss": 0.6736,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726547641358942e-05,
+      "loss": 0.6535,
+      "step": 971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725825655190019e-05,
+      "loss": 0.7225,
+      "step": 972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725102744028174e-05,
+      "loss": 0.7666,
+      "step": 973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724378908014909e-05,
+      "loss": 0.7343,
+      "step": 974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723654147291894e-05,
+      "loss": 0.714,
+      "step": 975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722928462000995e-05,
+      "loss": 0.7524,
+      "step": 976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722201852284245e-05,
+      "loss": 0.7537,
+      "step": 977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72147431828387e-05,
+      "loss": 0.7026,
+      "step": 978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720745860142266e-05,
+      "loss": 0.7571,
+      "step": 979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720016478002019e-05,
+      "loss": 0.6692,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71928617200589e-05,
+      "loss": 0.7804,
+      "step": 981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718554942296826e-05,
+      "loss": 0.7113,
+      "step": 982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717822789017948e-05,
+      "loss": 0.7236,
+      "step": 983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717089712312563e-05,
+      "loss": 0.6256,
+      "step": 984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716355712324156e-05,
+      "loss": 0.746,
+      "step": 985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715620789196397e-05,
+      "loss": 0.7257,
+      "step": 986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71488494307313e-05,
+      "loss": 0.6982,
+      "step": 987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714148174098387e-05,
+      "loss": 0.731,
+      "step": 988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713410482416374e-05,
+      "loss": 0.6932,
+      "step": 989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712671868171478e-05,
+      "loss": 0.7469,
+      "step": 990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711932331508277e-05,
+      "loss": 0.7301,
+      "step": 991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711191872571515e-05,
+      "loss": 0.7138,
+      "step": 992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710450491506124e-05,
+      "loss": 0.6998,
+      "step": 993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709708188457218e-05,
+      "loss": 0.6764,
+      "step": 994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708964963570088e-05,
+      "loss": 0.7895,
+      "step": 995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708220816990206e-05,
+      "loss": 0.6979,
+      "step": 996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707475748863226e-05,
+      "loss": 0.6669,
+      "step": 997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706729759334978e-05,
+      "loss": 0.7303,
+      "step": 998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70598284855148e-05,
+      "loss": 0.6528,
+      "step": 999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705235016658925e-05,
+      "loss": 0.6903,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704486263803685e-05,
+      "loss": 0.7475,
+      "step": 1001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703736590132317e-05,
+      "loss": 0.6454,
+      "step": 1002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702985995791554e-05,
+      "loss": 0.6413,
+      "step": 1003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702234480928312e-05,
+      "loss": 0.6798,
+      "step": 1004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701482045689684e-05,
+      "loss": 0.6844,
+      "step": 1005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700728690222949e-05,
+      "loss": 0.7043,
+      "step": 1006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699974414675559e-05,
+      "loss": 0.7109,
+      "step": 1007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699219219195153e-05,
+      "loss": 0.7138,
+      "step": 1008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698463103929542e-05,
+      "loss": 0.7144,
+      "step": 1009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697706069026727e-05,
+      "loss": 0.6856,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696948114634877e-05,
+      "loss": 0.7954,
+      "step": 1011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696189240902353e-05,
+      "loss": 0.6976,
+      "step": 1012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695429447977689e-05,
+      "loss": 0.6617,
+      "step": 1013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6946687360096e-05,
+      "loss": 0.7755,
+      "step": 1014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69390710514698e-05,
+      "loss": 0.6961,
+      "step": 1015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693144555538906e-05,
+      "loss": 0.7675,
+      "step": 1016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692381087334633e-05,
+      "loss": 0.7608,
+      "step": 1017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691616700683595e-05,
+      "loss": 0.7428,
+      "step": 1018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690851395735407e-05,
+      "loss": 0.7429,
+      "step": 1019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690085172639862e-05,
+      "loss": 0.6933,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689318031546935e-05,
+      "loss": 0.7229,
+      "step": 1021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688549972606782e-05,
+      "loss": 0.7328,
+      "step": 1022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687780995969732e-05,
+      "loss": 0.7097,
+      "step": 1023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6870111017863e-05,
+      "loss": 0.7584,
+      "step": 1024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68624029020718e-05,
+      "loss": 0.3882,
+      "step": 1025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685468561383241e-05,
+      "loss": 0.771,
+      "step": 1026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684695915465536e-05,
+      "loss": 0.75,
+      "step": 1027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683922352605297e-05,
+      "loss": 0.7289,
+      "step": 1028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683147872953935e-05,
+      "loss": 0.7355,
+      "step": 1029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682372476663037e-05,
+      "loss": 0.6926,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681596163884376e-05,
+      "loss": 0.7669,
+      "step": 1031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680818934769899e-05,
+      "loss": 0.78,
+      "step": 1032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680040789471736e-05,
+      "loss": 0.7307,
+      "step": 1033
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67926172814219e-05,
+      "loss": 0.7229,
+      "step": 1034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678481750933754e-05,
+      "loss": 0.7284,
+      "step": 1035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67770085799909e-05,
+      "loss": 0.704,
+      "step": 1036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676919049491045e-05,
+      "loss": 0.7679,
+      "step": 1037
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676136325562642e-05,
+      "loss": 0.7013,
+      "step": 1038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675352686367086e-05,
+      "loss": 0.7309,
+      "step": 1039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67456813205776e-05,
+      "loss": 0.6003,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673782662788224e-05,
+      "loss": 0.7653,
+      "step": 1041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672996278712223e-05,
+      "loss": 0.7098,
+      "step": 1042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672208979983673e-05,
+      "loss": 0.457,
+      "step": 1043
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671420766756676e-05,
+      "loss": 0.7078,
+      "step": 1044
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670631639185509e-05,
+      "loss": 0.7253,
+      "step": 1045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669841597424629e-05,
+      "loss": 0.6788,
+      "step": 1046
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669050641628672e-05,
+      "loss": 0.6965,
+      "step": 1047
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668258771952453e-05,
+      "loss": 0.7001,
+      "step": 1048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667465988550967e-05,
+      "loss": 0.7454,
+      "step": 1049
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666672291579385e-05,
+      "loss": 0.6952,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66587768119306e-05,
+      "loss": 0.6643,
+      "step": 1051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66508215754752e-05,
+      "loss": 0.7553,
+      "step": 1052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664285720798477e-05,
+      "loss": 0.6699,
+      "step": 1053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663488371101816e-05,
+      "loss": 0.7537,
+      "step": 1054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662690108613605e-05,
+      "loss": 0.6615,
+      "step": 1055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661890933490089e-05,
+      "loss": 0.7562,
+      "step": 1056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661090845887693e-05,
+      "loss": 0.7124,
+      "step": 1057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660289845963017e-05,
+      "loss": 0.6835,
+      "step": 1058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659487933872844e-05,
+      "loss": 0.6285,
+      "step": 1059
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65868510977413e-05,
+      "loss": 0.6921,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657881373824016e-05,
+      "loss": 0.6896,
+      "step": 1061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65707672617982e-05,
+      "loss": 0.6871,
+      "step": 1062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656271166999031e-05,
+      "loss": 0.7574,
+      "step": 1063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655464696439328e-05,
+      "loss": 0.755,
+      "step": 1064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654657314658559e-05,
+      "loss": 0.7042,
+      "step": 1065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653849021814759e-05,
+      "loss": 0.7318,
+      "step": 1066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65303981806613e-05,
+      "loss": 0.6659,
+      "step": 1067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652229703571061e-05,
+      "loss": 0.704,
+      "step": 1068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65141867848812e-05,
+      "loss": 0.6438,
+      "step": 1069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650606742976047e-05,
+      "loss": 0.7509,
+      "step": 1070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649793897193763e-05,
+      "loss": 0.7409,
+      "step": 1071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64898014130037e-05,
+      "loss": 0.6791,
+      "step": 1072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648165475455143e-05,
+      "loss": 0.6218,
+      "step": 1073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647349899817539e-05,
+      "loss": 0.7473,
+      "step": 1074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646533414547193e-05,
+      "loss": 0.7053,
+      "step": 1075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645716019803915e-05,
+      "loss": 0.6458,
+      "step": 1076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644897715747696e-05,
+      "loss": 0.7308,
+      "step": 1077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644078502538704e-05,
+      "loss": 0.6922,
+      "step": 1078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643258380337283e-05,
+      "loss": 0.694,
+      "step": 1079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64243734930396e-05,
+      "loss": 0.6809,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641615409599433e-05,
+      "loss": 0.7462,
+      "step": 1081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640792561384582e-05,
+      "loss": 0.7582,
+      "step": 1082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639968804820468e-05,
+      "loss": 0.6681,
+      "step": 1083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639144140068324e-05,
+      "loss": 0.723,
+      "step": 1084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63831856728956e-05,
+      "loss": 0.741,
+      "step": 1085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637492086645771e-05,
+      "loss": 0.67,
+      "step": 1086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636664698298722e-05,
+      "loss": 0.6943,
+      "step": 1087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635836402410361e-05,
+      "loss": 0.7803,
+      "step": 1088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635007199142812e-05,
+      "loss": 0.7826,
+      "step": 1089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634177088658374e-05,
+      "loss": 0.4081,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633346071119527e-05,
+      "loss": 0.743,
+      "step": 1091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632514146688927e-05,
+      "loss": 0.6736,
+      "step": 1092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631681315529408e-05,
+      "loss": 0.7132,
+      "step": 1093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630847577803981e-05,
+      "loss": 0.7336,
+      "step": 1094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630012933675835e-05,
+      "loss": 0.7508,
+      "step": 1095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629177383308336e-05,
+      "loss": 0.7312,
+      "step": 1096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628340926865029e-05,
+      "loss": 0.3959,
+      "step": 1097
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627503564509633e-05,
+      "loss": 0.7248,
+      "step": 1098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626665296406047e-05,
+      "loss": 0.6941,
+      "step": 1099
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625826122718345e-05,
+      "loss": 0.7679,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624986043610781e-05,
+      "loss": 0.7157,
+      "step": 1101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624145059247787e-05,
+      "loss": 0.6872,
+      "step": 1102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623303169793965e-05,
+      "loss": 0.7451,
+      "step": 1103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622460375414103e-05,
+      "loss": 0.7569,
+      "step": 1104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621616676273164e-05,
+      "loss": 0.6924,
+      "step": 1105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620772072536282e-05,
+      "loss": 0.6652,
+      "step": 1106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619926564368776e-05,
+      "loss": 0.7476,
+      "step": 1107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619080151936138e-05,
+      "loss": 0.6523,
+      "step": 1108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618232835404034e-05,
+      "loss": 0.6756,
+      "step": 1109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617384614938315e-05,
+      "loss": 0.7173,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616535490705004e-05,
+      "loss": 0.7164,
+      "step": 1111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615685462870298e-05,
+      "loss": 0.4016,
+      "step": 1112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614834531600577e-05,
+      "loss": 0.6903,
+      "step": 1113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613982697062395e-05,
+      "loss": 0.7578,
+      "step": 1114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613129959422483e-05,
+      "loss": 0.7279,
+      "step": 1115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612276318847746e-05,
+      "loss": 0.738,
+      "step": 1116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611421775505271e-05,
+      "loss": 0.6834,
+      "step": 1117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61056632956232e-05,
+      "loss": 0.7288,
+      "step": 1118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609709981186328e-05,
+      "loss": 0.7814,
+      "step": 1119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60885273054491e-05,
+      "loss": 0.6648,
+      "step": 1120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607994577805857e-05,
+      "loss": 0.7507,
+      "step": 1121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607135523137137e-05,
+      "loss": 0.7193,
+      "step": 1122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606275566706895e-05,
+      "loss": 0.7423,
+      "step": 1123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605414708683449e-05,
+      "loss": 0.686,
+      "step": 1124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604552949235299e-05,
+      "loss": 0.7036,
+      "step": 1125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603690288531117e-05,
+      "loss": 0.6648,
+      "step": 1126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602826726739751e-05,
+      "loss": 0.7344,
+      "step": 1127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60196226403023e-05,
+      "loss": 0.6793,
+      "step": 1128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601096900571757e-05,
+      "loss": 0.6649,
+      "step": 1129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600230636533708e-05,
+      "loss": 0.7319,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59936347208564e-05,
+      "loss": 0.7092,
+      "step": 1131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598495407397284e-05,
+      "loss": 0.6538,
+      "step": 1132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597626442638548e-05,
+      "loss": 0.7206,
+      "step": 1133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596756577979515e-05,
+      "loss": 0.6588,
+      "step": 1134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595885813590446e-05,
+      "loss": 0.7667,
+      "step": 1135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595014149641776e-05,
+      "loss": 0.6273,
+      "step": 1136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594141586304117e-05,
+      "loss": 0.6916,
+      "step": 1137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593268123748259e-05,
+      "loss": 0.4178,
+      "step": 1138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592393762145164e-05,
+      "loss": 0.6366,
+      "step": 1139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591518501665973e-05,
+      "loss": 0.689,
+      "step": 1140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590642342482001e-05,
+      "loss": 0.7294,
+      "step": 1141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589765284764742e-05,
+      "loss": 0.6701,
+      "step": 1142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588887328685863e-05,
+      "loss": 0.6739,
+      "step": 1143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588008474417204e-05,
+      "loss": 0.7724,
+      "step": 1144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587128722130791e-05,
+      "loss": 0.7175,
+      "step": 1145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586248071998816e-05,
+      "loss": 0.7653,
+      "step": 1146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585366524193648e-05,
+      "loss": 0.6815,
+      "step": 1147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584484078887838e-05,
+      "loss": 0.389,
+      "step": 1148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583600736254103e-05,
+      "loss": 0.6872,
+      "step": 1149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582716496465347e-05,
+      "loss": 0.7135,
+      "step": 1150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58183135969464e-05,
+      "loss": 0.7319,
+      "step": 1151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580945326115229e-05,
+      "loss": 0.6694,
+      "step": 1152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580058395900544e-05,
+      "loss": 0.7061,
+      "step": 1153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57917056922418e-05,
+      "loss": 0.7409,
+      "step": 1154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578281846259918e-05,
+      "loss": 0.7148,
+      "step": 1155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577392227181706e-05,
+      "loss": 0.6512,
+      "step": 1156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576501712163668e-05,
+      "loss": 0.6481,
+      "step": 1157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575610301380111e-05,
+      "loss": 0.7107,
+      "step": 1158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574717995005509e-05,
+      "loss": 0.647,
+      "step": 1159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573824793214515e-05,
+      "loss": 0.7515,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572930696181956e-05,
+      "loss": 0.7512,
+      "step": 1161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572035704082838e-05,
+      "loss": 0.7233,
+      "step": 1162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571139817092336e-05,
+      "loss": 0.7238,
+      "step": 1163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570243035385805e-05,
+      "loss": 0.4136,
+      "step": 1164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569345359138771e-05,
+      "loss": 0.7304,
+      "step": 1165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568446788526939e-05,
+      "loss": 0.3955,
+      "step": 1166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567547323726189e-05,
+      "loss": 0.3947,
+      "step": 1167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566646964912573e-05,
+      "loss": 0.6824,
+      "step": 1168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565745712262321e-05,
+      "loss": 0.7022,
+      "step": 1169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564843565951834e-05,
+      "loss": 0.7821,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563940526157692e-05,
+      "loss": 0.7411,
+      "step": 1171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563036593056649e-05,
+      "loss": 0.7371,
+      "step": 1172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562131766825632e-05,
+      "loss": 0.7561,
+      "step": 1173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561226047641745e-05,
+      "loss": 0.635,
+      "step": 1174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560319435682265e-05,
+      "loss": 0.602,
+      "step": 1175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559411931124645e-05,
+      "loss": 0.7855,
+      "step": 1176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55850353414651e-05,
+      "loss": 0.7848,
+      "step": 1177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557594244925665e-05,
+      "loss": 0.7566,
+      "step": 1178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556684063640086e-05,
+      "loss": 0.7518,
+      "step": 1179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555772990467924e-05,
+      "loss": 0.652,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554861025587502e-05,
+      "loss": 0.6889,
+      "step": 1181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553948169177324e-05,
+      "loss": 0.7476,
+      "step": 1182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553034421416062e-05,
+      "loss": 0.4225,
+      "step": 1183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552119782482569e-05,
+      "loss": 0.731,
+      "step": 1184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551204252555864e-05,
+      "loss": 0.7455,
+      "step": 1185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550287831815148e-05,
+      "loss": 0.3803,
+      "step": 1186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549370520439792e-05,
+      "loss": 0.6641,
+      "step": 1187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548452318609343e-05,
+      "loss": 0.7406,
+      "step": 1188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547533226503522e-05,
+      "loss": 0.7065,
+      "step": 1189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546613244302226e-05,
+      "loss": 0.7455,
+      "step": 1190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545692372185522e-05,
+      "loss": 0.7354,
+      "step": 1191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544770610333655e-05,
+      "loss": 0.6425,
+      "step": 1192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543847958927043e-05,
+      "loss": 0.7305,
+      "step": 1193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542924418146278e-05,
+      "loss": 0.7168,
+      "step": 1194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541999988172123e-05,
+      "loss": 0.7198,
+      "step": 1195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541074669185521e-05,
+      "loss": 0.7128,
+      "step": 1196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540148461367588e-05,
+      "loss": 0.7428,
+      "step": 1197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539221364899608e-05,
+      "loss": 0.6734,
+      "step": 1198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538293379963045e-05,
+      "loss": 0.723,
+      "step": 1199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537364506739534e-05,
+      "loss": 0.7019,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536434745410885e-05,
+      "loss": 0.7311,
+      "step": 1201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535504096159082e-05,
+      "loss": 0.7338,
+      "step": 1202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534572559166283e-05,
+      "loss": 0.7065,
+      "step": 1203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533640134614818e-05,
+      "loss": 0.7171,
+      "step": 1204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532706822687192e-05,
+      "loss": 0.6332,
+      "step": 1205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531772623566084e-05,
+      "loss": 0.686,
+      "step": 1206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530837537434347e-05,
+      "loss": 0.4334,
+      "step": 1207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.529901564475003e-05,
+      "loss": 0.7173,
+      "step": 1208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528964704871257e-05,
+      "loss": 0.7206,
+      "step": 1209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528026958806477e-05,
+      "loss": 0.768,
+      "step": 1210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527088326464213e-05,
+      "loss": 0.7096,
+      "step": 1211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526148808028182e-05,
+      "loss": 0.6407,
+      "step": 1212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525208403682279e-05,
+      "loss": 0.6901,
+      "step": 1213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524267113610571e-05,
+      "loss": 0.7236,
+      "step": 1214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523324937997297e-05,
+      "loss": 0.7301,
+      "step": 1215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522381877026871e-05,
+      "loss": 0.6916,
+      "step": 1216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521437930883881e-05,
+      "loss": 0.691,
+      "step": 1217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520493099753085e-05,
+      "loss": 0.7385,
+      "step": 1218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519547383819416e-05,
+      "loss": 0.6979,
+      "step": 1219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518600783267982e-05,
+      "loss": 0.7502,
+      "step": 1220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517653298284063e-05,
+      "loss": 0.6978,
+      "step": 1221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51670492905311e-05,
+      "loss": 0.7399,
+      "step": 1222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515755675760749e-05,
+      "loss": 0.7499,
+      "step": 1223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51480553859278e-05,
+      "loss": 0.7299,
+      "step": 1224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513854517735172e-05,
+      "loss": 0.7167,
+      "step": 1225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512902613374072e-05,
+      "loss": 0.6959,
+      "step": 1226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511949825695797e-05,
+      "loss": 0.7055,
+      "step": 1227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510996154886839e-05,
+      "loss": 0.6832,
+      "step": 1228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510041601133859e-05,
+      "loss": 0.6791,
+      "step": 1229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509086164623694e-05,
+      "loss": 0.6384,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508129845543355e-05,
+      "loss": 0.7613,
+      "step": 1231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507172644080023e-05,
+      "loss": 0.6971,
+      "step": 1232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50621456042105e-05,
+      "loss": 0.7692,
+      "step": 1233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505255594753964e-05,
+      "loss": 0.6712,
+      "step": 1234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504295747266467e-05,
+      "loss": 0.7233,
+      "step": 1235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50333501814643e-05,
+      "loss": 0.6604,
+      "step": 1236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502373407581897e-05,
+      "loss": 0.7044,
+      "step": 1237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501410915761087e-05,
+      "loss": 0.6418,
+      "step": 1238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500447542872391e-05,
+      "loss": 0.6891,
+      "step": 1239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499483289104367e-05,
+      "loss": 0.6643,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498518154645754e-05,
+      "loss": 0.7065,
+      "step": 1241
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497552139685457e-05,
+      "loss": 0.6695,
+      "step": 1242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496585244412556e-05,
+      "loss": 0.7557,
+      "step": 1243
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495617469016305e-05,
+      "loss": 0.6869,
+      "step": 1244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494648813686126e-05,
+      "loss": 0.6993,
+      "step": 1245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493679278611616e-05,
+      "loss": 0.6712,
+      "step": 1246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492708863982543e-05,
+      "loss": 0.7374,
+      "step": 1247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491737569988845e-05,
+      "loss": 0.6101,
+      "step": 1248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490765396820643e-05,
+      "loss": 0.7307,
+      "step": 1249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.489792344668213e-05,
+      "loss": 0.6771,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488818413722016e-05,
+      "loss": 0.7731,
+      "step": 1251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487843604172681e-05,
+      "loss": 0.724,
+      "step": 1252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48686791621101e-05,
+      "loss": 0.7098,
+      "step": 1253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485891350027974e-05,
+      "loss": 0.6772,
+      "step": 1254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484913905814716e-05,
+      "loss": 0.7656,
+      "step": 1255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483935583762555e-05,
+      "loss": 0.7117,
+      "step": 1256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482956384062982e-05,
+      "loss": 0.7354,
+      "step": 1257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481976306907651e-05,
+      "loss": 0.6704,
+      "step": 1258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4809953524884e-05,
+      "loss": 0.6335,
+      "step": 1259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48001352099723e-05,
+      "loss": 0.6764,
+      "step": 1260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479030812626316e-05,
+      "loss": 0.6896,
+      "step": 1261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478047227568005e-05,
+      "loss": 0.6917,
+      "step": 1262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477062766014817e-05,
+      "loss": 0.7164,
+      "step": 1263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47607742815944e-05,
+      "loss": 0.7352,
+      "step": 1264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475091214194737e-05,
+      "loss": 0.7042,
+      "step": 1265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474104124313741e-05,
+      "loss": 0.4522,
+      "step": 1266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473116158709657e-05,
+      "loss": 0.7107,
+      "step": 1267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472127317575859e-05,
+      "loss": 0.6877,
+      "step": 1268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471137601105897e-05,
+      "loss": 0.7438,
+      "step": 1269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470147009493488e-05,
+      "loss": 0.7042,
+      "step": 1270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469155542932522e-05,
+      "loss": 0.7327,
+      "step": 1271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468163201617062e-05,
+      "loss": 0.682,
+      "step": 1272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467169985741337e-05,
+      "loss": 0.6973,
+      "step": 1273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466175895499756e-05,
+      "loss": 0.7074,
+      "step": 1274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465180931086889e-05,
+      "loss": 0.7292,
+      "step": 1275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464185092697483e-05,
+      "loss": 0.6501,
+      "step": 1276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463188380526455e-05,
+      "loss": 0.7043,
+      "step": 1277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462190794768894e-05,
+      "loss": 0.6935,
+      "step": 1278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461192335620059e-05,
+      "loss": 0.7332,
+      "step": 1279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460193003275378e-05,
+      "loss": 0.6977,
+      "step": 1280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459192797930454e-05,
+      "loss": 0.6899,
+      "step": 1281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458191719781056e-05,
+      "loss": 0.6745,
+      "step": 1282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45718976902313e-05,
+      "loss": 0.7529,
+      "step": 1283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456186945852787e-05,
+      "loss": 0.6775,
+      "step": 1284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45518325046631e-05,
+      "loss": 0.7003,
+      "step": 1285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454178683060157e-05,
+      "loss": 0.6874,
+      "step": 1286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453173243830953e-05,
+      "loss": 0.7514,
+      "step": 1287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452166932975492e-05,
+      "loss": 0.6877,
+      "step": 1288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451159750690741e-05,
+      "loss": 0.6798,
+      "step": 1289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450151697173839e-05,
+      "loss": 0.7114,
+      "step": 1290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449142772622092e-05,
+      "loss": 0.7105,
+      "step": 1291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44813297723298e-05,
+      "loss": 0.6615,
+      "step": 1292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447122311204151e-05,
+      "loss": 0.6917,
+      "step": 1293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446110774733424e-05,
+      "loss": 0.7058,
+      "step": 1294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445098368018786e-05,
+      "loss": 0.6711,
+      "step": 1295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444085091258402e-05,
+      "loss": 0.7223,
+      "step": 1296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4430709446506e-05,
+      "loss": 0.7162,
+      "step": 1297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442055928393879e-05,
+      "loss": 0.6537,
+      "step": 1298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441040042686911e-05,
+      "loss": 0.7399,
+      "step": 1299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440023287728537e-05,
+      "loss": 0.6281,
+      "step": 1300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439005663717768e-05,
+      "loss": 0.6642,
+      "step": 1301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437987170853783e-05,
+      "loss": 0.6756,
+      "step": 1302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436967809335937e-05,
+      "loss": 0.7091,
+      "step": 1303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435947579363748e-05,
+      "loss": 0.6764,
+      "step": 1304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434926481136908e-05,
+      "loss": 0.735,
+      "step": 1305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433904514855278e-05,
+      "loss": 0.6857,
+      "step": 1306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432881680718888e-05,
+      "loss": 0.7095,
+      "step": 1307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431857978927941e-05,
+      "loss": 0.7233,
+      "step": 1308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430833409682806e-05,
+      "loss": 0.7002,
+      "step": 1309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429807973184025e-05,
+      "loss": 0.7125,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428781669632308e-05,
+      "loss": 0.606,
+      "step": 1311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427754499228534e-05,
+      "loss": 0.7219,
+      "step": 1312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.426726462173751e-05,
+      "loss": 0.6986,
+      "step": 1313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425697558669182e-05,
+      "loss": 0.6493,
+      "step": 1314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424667788916215e-05,
+      "loss": 0.7354,
+      "step": 1315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423637153116405e-05,
+      "loss": 0.7707,
+      "step": 1316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422605651471484e-05,
+      "loss": 0.7205,
+      "step": 1317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421573284183345e-05,
+      "loss": 0.7156,
+      "step": 1318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420540051454058e-05,
+      "loss": 0.6986,
+      "step": 1319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419505953485861e-05,
+      "loss": 0.7442,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418470990481155e-05,
+      "loss": 0.7501,
+      "step": 1321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417435162642517e-05,
+      "loss": 0.7408,
+      "step": 1322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416398470172691e-05,
+      "loss": 0.6723,
+      "step": 1323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415360913274592e-05,
+      "loss": 0.737,
+      "step": 1324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4143224921513e-05,
+      "loss": 0.6039,
+      "step": 1325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413283207006068e-05,
+      "loss": 0.7253,
+      "step": 1326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412243058042315e-05,
+      "loss": 0.6835,
+      "step": 1327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411202045463634e-05,
+      "loss": 0.6782,
+      "step": 1328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410160169473782e-05,
+      "loss": 0.6667,
+      "step": 1329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409117430276687e-05,
+      "loss": 0.6388,
+      "step": 1330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408073828076447e-05,
+      "loss": 0.6482,
+      "step": 1331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407029363077326e-05,
+      "loss": 0.7514,
+      "step": 1332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405984035483761e-05,
+      "loss": 0.7264,
+      "step": 1333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404937845500354e-05,
+      "loss": 0.624,
+      "step": 1334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403890793331876e-05,
+      "loss": 0.7762,
+      "step": 1335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402842879183272e-05,
+      "loss": 0.6978,
+      "step": 1336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40179410325965e-05,
+      "loss": 0.6798,
+      "step": 1337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400744465766285e-05,
+      "loss": 0.659,
+      "step": 1338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399693966908627e-05,
+      "loss": 0.7165,
+      "step": 1339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398642606892295e-05,
+      "loss": 0.6961,
+      "step": 1340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397590385923069e-05,
+      "loss": 0.6973,
+      "step": 1341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396537304206902e-05,
+      "loss": 0.4127,
+      "step": 1342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395483361949916e-05,
+      "loss": 0.7521,
+      "step": 1343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394428559358401e-05,
+      "loss": 0.7093,
+      "step": 1344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393372896638816e-05,
+      "loss": 0.7205,
+      "step": 1345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392316373997785e-05,
+      "loss": 0.3753,
+      "step": 1346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391258991642104e-05,
+      "loss": 0.6854,
+      "step": 1347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390200749778738e-05,
+      "loss": 0.7313,
+      "step": 1348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.389141648614815e-05,
+      "loss": 0.6539,
+      "step": 1349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388081688357637e-05,
+      "loss": 0.6734,
+      "step": 1350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387020869214669e-05,
+      "loss": 0.675,
+      "step": 1351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385959191393549e-05,
+      "loss": 0.7699,
+      "step": 1352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38489665510208e-05,
+      "loss": 0.7158,
+      "step": 1353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.383833260548233e-05,
+      "loss": 0.7815,
+      "step": 1354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.382769007940148e-05,
+      "loss": 0.6894,
+      "step": 1355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381703897486136e-05,
+      "loss": 0.6964,
+      "step": 1356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380637929394668e-05,
+      "loss": 0.6928,
+      "step": 1357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37957110387439e-05,
+      "loss": 0.4318,
+      "step": 1358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378503421134113e-05,
+      "loss": 0.7266,
+      "step": 1359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377434881382816e-05,
+      "loss": 0.7327,
+      "step": 1360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376365484829646e-05,
+      "loss": 0.7896,
+      "step": 1361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375295231683918e-05,
+      "loss": 0.6743,
+      "step": 1362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374224122155111e-05,
+      "loss": 0.6271,
+      "step": 1363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37315215645288e-05,
+      "loss": 0.7264,
+      "step": 1364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372079334787041e-05,
+      "loss": 0.7688,
+      "step": 1365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371005657367577e-05,
+      "loss": 0.6553,
+      "step": 1366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36993112440464e-05,
+      "loss": 0.6954,
+      "step": 1367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368855736108552e-05,
+      "loss": 0.7148,
+      "step": 1368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3677794926898e-05,
+      "loss": 0.7479,
+      "step": 1369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366702394359042e-05,
+      "loss": 0.7034,
+      "step": 1370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365624441327093e-05,
+      "loss": 0.672,
+      "step": 1371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364545633804946e-05,
+      "loss": 0.7047,
+      "step": 1372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363465972003758e-05,
+      "loss": 0.6577,
+      "step": 1373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362385456134855e-05,
+      "loss": 0.712,
+      "step": 1374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361304086409721e-05,
+      "loss": 0.6017,
+      "step": 1375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360221863040022e-05,
+      "loss": 0.7196,
+      "step": 1376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.359138786237578e-05,
+      "loss": 0.7697,
+      "step": 1377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358054856214384e-05,
+      "loss": 0.6506,
+      "step": 1378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356970073182597e-05,
+      "loss": 0.6749,
+      "step": 1379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355884437354545e-05,
+      "loss": 0.6939,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354797948942719e-05,
+      "loss": 0.4266,
+      "step": 1381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353710608159783e-05,
+      "loss": 0.6858,
+      "step": 1382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352622415218562e-05,
+      "loss": 0.6735,
+      "step": 1383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351533370332046e-05,
+      "loss": 0.744,
+      "step": 1384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350443473713399e-05,
+      "loss": 0.6474,
+      "step": 1385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34935272557595e-05,
+      "loss": 0.6682,
+      "step": 1386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348261126133189e-05,
+      "loss": 0.6884,
+      "step": 1387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347168675598776e-05,
+      "loss": 0.7302,
+      "step": 1388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.346075374186541e-05,
+      "loss": 0.7227,
+      "step": 1389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344981222110476e-05,
+      "loss": 0.7217,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343886219584741e-05,
+      "loss": 0.6683,
+      "step": 1391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342790366823663e-05,
+      "loss": 0.6499,
+      "step": 1392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341693664041734e-05,
+      "loss": 0.7189,
+      "step": 1393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340596111453615e-05,
+      "loss": 0.7043,
+      "step": 1394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.339497709274128e-05,
+      "loss": 0.7076,
+      "step": 1395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338398457718269e-05,
+      "loss": 0.6546,
+      "step": 1396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337298357001194e-05,
+      "loss": 0.5689,
+      "step": 1397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336197407338229e-05,
+      "loss": 0.6259,
+      "step": 1398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335095608944861e-05,
+      "loss": 0.6271,
+      "step": 1399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333992962036751e-05,
+      "loss": 0.688,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33288946682972e-05,
+      "loss": 0.6869,
+      "step": 1401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331785123539756e-05,
+      "loss": 0.7216,
+      "step": 1402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330679932383012e-05,
+      "loss": 0.6714,
+      "step": 1403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329573893575814e-05,
+      "loss": 0.7356,
+      "step": 1404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328467007334644e-05,
+      "loss": 0.6504,
+      "step": 1405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327359273876156e-05,
+      "loss": 0.6773,
+      "step": 1406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326250693417169e-05,
+      "loss": 0.7309,
+      "step": 1407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325141266174666e-05,
+      "loss": 0.4118,
+      "step": 1408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324030992365798e-05,
+      "loss": 0.7303,
+      "step": 1409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32291987220788e-05,
+      "loss": 0.7371,
+      "step": 1410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321807905918394e-05,
+      "loss": 0.7132,
+      "step": 1411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320695093714984e-05,
+      "loss": 0.7294,
+      "step": 1412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319581435815466e-05,
+      "loss": 0.6677,
+      "step": 1413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318466932437814e-05,
+      "loss": 0.6685,
+      "step": 1414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317351583800176e-05,
+      "loss": 0.755,
+      "step": 1415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316235390120858e-05,
+      "loss": 0.6854,
+      "step": 1416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315118351618335e-05,
+      "loss": 0.705,
+      "step": 1417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314000468511245e-05,
+      "loss": 0.6602,
+      "step": 1418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312881741018395e-05,
+      "loss": 0.6953,
+      "step": 1419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311762169358756e-05,
+      "loss": 0.7495,
+      "step": 1420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310641753751461e-05,
+      "loss": 0.7348,
+      "step": 1421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309520494415811e-05,
+      "loss": 0.7456,
+      "step": 1422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308398391571274e-05,
+      "loss": 0.7066,
+      "step": 1423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307275445437478e-05,
+      "loss": 0.6571,
+      "step": 1424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30615165623422e-05,
+      "loss": 0.7642,
+      "step": 1425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305027024181462e-05,
+      "loss": 0.7136,
+      "step": 1426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.303901549499327e-05,
+      "loss": 0.6796,
+      "step": 1427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302775232408111e-05,
+      "loss": 0.7148,
+      "step": 1428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301648073128265e-05,
+      "loss": 0.6292,
+      "step": 1429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300520071880412e-05,
+      "loss": 0.6582,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299391228885334e-05,
+      "loss": 0.7556,
+      "step": 1431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298261544363987e-05,
+      "loss": 0.7191,
+      "step": 1432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297131018537481e-05,
+      "loss": 0.6804,
+      "step": 1433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295999651627097e-05,
+      "loss": 0.7137,
+      "step": 1434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294867443854278e-05,
+      "loss": 0.6953,
+      "step": 1435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293734395440636e-05,
+      "loss": 0.6868,
+      "step": 1436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292600506607942e-05,
+      "loss": 0.7336,
+      "step": 1437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291465777578132e-05,
+      "loss": 0.7882,
+      "step": 1438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290330208573313e-05,
+      "loss": 0.6836,
+      "step": 1439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289193799815745e-05,
+      "loss": 0.6567,
+      "step": 1440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288056551527865e-05,
+      "loss": 0.7686,
+      "step": 1441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286918463932267e-05,
+      "loss": 0.7191,
+      "step": 1442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285779537251707e-05,
+      "loss": 0.7165,
+      "step": 1443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284639771709112e-05,
+      "loss": 0.656,
+      "step": 1444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283499167527571e-05,
+      "loss": 0.754,
+      "step": 1445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282357724930333e-05,
+      "loss": 0.6576,
+      "step": 1446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281215444140817e-05,
+      "loss": 0.6976,
+      "step": 1447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280072325382602e-05,
+      "loss": 0.6565,
+      "step": 1448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278928368879432e-05,
+      "loss": 0.7043,
+      "step": 1449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277783574855216e-05,
+      "loss": 0.7021,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.276637943534027e-05,
+      "loss": 0.7155,
+      "step": 1451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2754914751401e-05,
+      "loss": 0.6464,
+      "step": 1452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274344169897834e-05,
+      "loss": 0.7331,
+      "step": 1453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273196028031794e-05,
+      "loss": 0.7128,
+      "step": 1454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272047049766709e-05,
+      "loss": 0.7017,
+      "step": 1455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270897235327469e-05,
+      "loss": 0.6521,
+      "step": 1456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269746584939126e-05,
+      "loss": 0.686,
+      "step": 1457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268595098826903e-05,
+      "loss": 0.4026,
+      "step": 1458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26744277721618e-05,
+      "loss": 0.7014,
+      "step": 1459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266289620332503e-05,
+      "loss": 0.6263,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265135628401582e-05,
+      "loss": 0.6651,
+      "step": 1461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263980801649286e-05,
+      "loss": 0.6763,
+      "step": 1462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262825140301655e-05,
+      "loss": 0.6577,
+      "step": 1463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261668644584886e-05,
+      "loss": 0.4083,
+      "step": 1464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260511314725345e-05,
+      "loss": 0.7137,
+      "step": 1465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259353150949553e-05,
+      "loss": 0.6884,
+      "step": 1466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.258194153484202e-05,
+      "loss": 0.4003,
+      "step": 1467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257034322556145e-05,
+      "loss": 0.7204,
+      "step": 1468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.255873658392394e-05,
+      "loss": 0.7437,
+      "step": 1469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254712161220131e-05,
+      "loss": 0.6066,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253549831266696e-05,
+      "loss": 0.6908,
+      "step": 1471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252386668759597e-05,
+      "loss": 0.7086,
+      "step": 1472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251222673926493e-05,
+      "loss": 0.6697,
+      "step": 1473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250057846995224e-05,
+      "loss": 0.7503,
+      "step": 1474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248892188193778e-05,
+      "loss": 0.6497,
+      "step": 1475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247725697750314e-05,
+      "loss": 0.7638,
+      "step": 1476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.246558375893146e-05,
+      "loss": 0.6751,
+      "step": 1477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245390222850762e-05,
+      "loss": 0.7679,
+      "step": 1478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244221238851802e-05,
+      "loss": 0.695,
+      "step": 1479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243051424125075e-05,
+      "loss": 0.6328,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241880778899549e-05,
+      "loss": 0.6998,
+      "step": 1481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240709303404357e-05,
+      "loss": 0.7203,
+      "step": 1482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239536997868794e-05,
+      "loss": 0.6958,
+      "step": 1483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238363862522317e-05,
+      "loss": 0.7502,
+      "step": 1484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237189897594544e-05,
+      "loss": 0.7163,
+      "step": 1485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236015103315261e-05,
+      "loss": 0.7443,
+      "step": 1486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234839479914406e-05,
+      "loss": 0.6603,
+      "step": 1487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233663027622092e-05,
+      "loss": 0.6703,
+      "step": 1488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.232485746668584e-05,
+      "loss": 0.7352,
+      "step": 1489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231307637284312e-05,
+      "loss": 0.7078,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230128699699874e-05,
+      "loss": 0.6738,
+      "step": 1491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228948934146019e-05,
+      "loss": 0.68,
+      "step": 1492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227768340853668e-05,
+      "loss": 0.67,
+      "step": 1493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226586920053902e-05,
+      "loss": 0.7349,
+      "step": 1494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225404671977958e-05,
+      "loss": 0.717,
+      "step": 1495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224221596857241e-05,
+      "loss": 0.6786,
+      "step": 1496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223037694923316e-05,
+      "loss": 0.7612,
+      "step": 1497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221852966407909e-05,
+      "loss": 0.7895,
+      "step": 1498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220667411542912e-05,
+      "loss": 0.6515,
+      "step": 1499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219481030560371e-05,
+      "loss": 0.4085,
+      "step": 1500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.218293823692501e-05,
+      "loss": 0.7102,
+      "step": 1501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217105791171676e-05,
+      "loss": 0.6851,
+      "step": 1502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215916933230429e-05,
+      "loss": 0.7342,
+      "step": 1503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21472725010146e-05,
+      "loss": 0.3716,
+      "step": 1504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213536742017626e-05,
+      "loss": 0.6787,
+      "step": 1505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212345409211944e-05,
+      "loss": 0.6845,
+      "step": 1506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.2111532519176e-05,
+      "loss": 0.6882,
+      "step": 1507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209960270367935e-05,
+      "loss": 0.4002,
+      "step": 1508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208766464796453e-05,
+      "loss": 0.7325,
+      "step": 1509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207571835436817e-05,
+      "loss": 0.6963,
+      "step": 1510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206376382522857e-05,
+      "loss": 0.6614,
+      "step": 1511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205180106288561e-05,
+      "loss": 0.6762,
+      "step": 1512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203983006968076e-05,
+      "loss": 0.6556,
+      "step": 1513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20278508479571e-05,
+      "loss": 0.7286,
+      "step": 1514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201586340005941e-05,
+      "loss": 0.6769,
+      "step": 1515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200386772833394e-05,
+      "loss": 0.6779,
+      "step": 1516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199186383512865e-05,
+      "loss": 0.6714,
+      "step": 1517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197985172279309e-05,
+      "loss": 0.617,
+      "step": 1518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19678313936784e-05,
+      "loss": 0.7192,
+      "step": 1519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195580285013732e-05,
+      "loss": 0.7068,
+      "step": 1520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194376609452423e-05,
+      "loss": 0.6387,
+      "step": 1521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19317211291951e-05,
+      "loss": 0.6783,
+      "step": 1522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191966795650753e-05,
+      "loss": 0.7117,
+      "step": 1523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190760657882066e-05,
+      "loss": 0.7117,
+      "step": 1524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189553699849531e-05,
+      "loss": 0.634,
+      "step": 1525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18834592178939e-05,
+      "loss": 0.688,
+      "step": 1526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187137323938038e-05,
+      "loss": 0.6613,
+      "step": 1527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185927906532039e-05,
+      "loss": 0.7124,
+      "step": 1528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184717669808114e-05,
+      "loss": 0.753,
+      "step": 1529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183506614003143e-05,
+      "loss": 0.6709,
+      "step": 1530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182294739354169e-05,
+      "loss": 0.7371,
+      "step": 1531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181082046098393e-05,
+      "loss": 0.7461,
+      "step": 1532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179868534473179e-05,
+      "loss": 0.679,
+      "step": 1533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.178654204716048e-05,
+      "loss": 0.7263,
+      "step": 1534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177439057064683e-05,
+      "loss": 0.6354,
+      "step": 1535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176223091756925e-05,
+      "loss": 0.7134,
+      "step": 1536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17500630903078e-05,
+      "loss": 0.6312,
+      "step": 1537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17378870912441e-05,
+      "loss": 0.6557,
+      "step": 1538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172570292276135e-05,
+      "loss": 0.669,
+      "step": 1539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171351058724438e-05,
+      "loss": 0.6882,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170131008707964e-05,
+      "loss": 0.5773,
+      "step": 1541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168910142465514e-05,
+      "loss": 0.6984,
+      "step": 1542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167688460236049e-05,
+      "loss": 0.7257,
+      "step": 1543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166465962258692e-05,
+      "loss": 0.6142,
+      "step": 1544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165242648772724e-05,
+      "loss": 0.7073,
+      "step": 1545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164018520017586e-05,
+      "loss": 0.7148,
+      "step": 1546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162793576232877e-05,
+      "loss": 0.697,
+      "step": 1547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16156781765836e-05,
+      "loss": 0.7034,
+      "step": 1548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160341244533952e-05,
+      "loss": 0.4003,
+      "step": 1549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159113857099734e-05,
+      "loss": 0.7019,
+      "step": 1550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157885655595942e-05,
+      "loss": 0.7613,
+      "step": 1551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156656640262975e-05,
+      "loss": 0.7606,
+      "step": 1552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155426811341391e-05,
+      "loss": 0.6888,
+      "step": 1553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154196169071905e-05,
+      "loss": 0.4353,
+      "step": 1554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152964713695394e-05,
+      "loss": 0.7084,
+      "step": 1555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15173244545289e-05,
+      "loss": 0.6396,
+      "step": 1556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150499364585589e-05,
+      "loss": 0.7357,
+      "step": 1557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149265471334842e-05,
+      "loss": 0.7271,
+      "step": 1558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148030765942162e-05,
+      "loss": 0.6904,
+      "step": 1559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.146795248649218e-05,
+      "loss": 0.657,
+      "step": 1560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145558919697844e-05,
+      "loss": 0.7811,
+      "step": 1561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144321779330023e-05,
+      "loss": 0.6525,
+      "step": 1562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143083827787906e-05,
+      "loss": 0.7011,
+      "step": 1563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.141845065313796e-05,
+      "loss": 0.7119,
+      "step": 1564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140605492150163e-05,
+      "loss": 0.6388,
+      "step": 1565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139365108539625e-05,
+      "loss": 0.74,
+      "step": 1566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.138123914724968e-05,
+      "loss": 0.7238,
+      "step": 1567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136881910949129e-05,
+      "loss": 0.6408,
+      "step": 1568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135639097455212e-05,
+      "loss": 0.7372,
+      "step": 1569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13439547448647e-05,
+      "loss": 0.7069,
+      "step": 1570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13315104228632e-05,
+      "loss": 0.4282,
+      "step": 1571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131905801098341e-05,
+      "loss": 0.6777,
+      "step": 1572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130659751166262e-05,
+      "loss": 0.3793,
+      "step": 1573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.129412892733975e-05,
+      "loss": 0.6882,
+      "step": 1574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128165226045529e-05,
+      "loss": 0.3997,
+      "step": 1575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.126916751345132e-05,
+      "loss": 0.7314,
+      "step": 1576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125667468877152e-05,
+      "loss": 0.6946,
+      "step": 1577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124417378886109e-05,
+      "loss": 0.6858,
+      "step": 1578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123166481616687e-05,
+      "loss": 0.6966,
+      "step": 1579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.121914777313727e-05,
+      "loss": 0.7165,
+      "step": 1580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120662266222225e-05,
+      "loss": 0.7286,
+      "step": 1581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119408948587338e-05,
+      "loss": 0.6873,
+      "step": 1582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118154824654379e-05,
+      "loss": 0.6907,
+      "step": 1583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11689989466882e-05,
+      "loss": 0.6643,
+      "step": 1584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11564415887629e-05,
+      "loss": 0.6332,
+      "step": 1585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114387617522577e-05,
+      "loss": 0.697,
+      "step": 1586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.113130270853625e-05,
+      "loss": 0.6464,
+      "step": 1587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111872119115537e-05,
+      "loss": 0.7166,
+      "step": 1588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110613162554573e-05,
+      "loss": 0.7111,
+      "step": 1589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.109353401417147e-05,
+      "loss": 0.6922,
+      "step": 1590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10809283594984e-05,
+      "loss": 0.6755,
+      "step": 1591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106831466399378e-05,
+      "loss": 0.7062,
+      "step": 1592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105569293012656e-05,
+      "loss": 0.6211,
+      "step": 1593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.104306316036718e-05,
+      "loss": 0.6868,
+      "step": 1594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103042535718769e-05,
+      "loss": 0.7087,
+      "step": 1595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101777952306171e-05,
+      "loss": 0.4073,
+      "step": 1596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100512566046444e-05,
+      "loss": 0.709,
+      "step": 1597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099246377187262e-05,
+      "loss": 0.6963,
+      "step": 1598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.097979385976458e-05,
+      "loss": 0.3842,
+      "step": 1599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.096711592662024e-05,
+      "loss": 0.6863,
+      "step": 1600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095442997492105e-05,
+      "loss": 0.6909,
+      "step": 1601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094173600715005e-05,
+      "loss": 0.742,
+      "step": 1602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092903402579186e-05,
+      "loss": 0.7048,
+      "step": 1603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091632403333267e-05,
+      "loss": 0.7288,
+      "step": 1604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090360603226022e-05,
+      "loss": 0.6855,
+      "step": 1605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089088002506379e-05,
+      "loss": 0.6568,
+      "step": 1606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08781460142343e-05,
+      "loss": 0.6736,
+      "step": 1607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086540400226417e-05,
+      "loss": 0.3827,
+      "step": 1608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085265399164743e-05,
+      "loss": 0.3686,
+      "step": 1609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.083989598487968e-05,
+      "loss": 0.34,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082712998445801e-05,
+      "loss": 0.7507,
+      "step": 1611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081435599288116e-05,
+      "loss": 0.7076,
+      "step": 1612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080157401264941e-05,
+      "loss": 0.6564,
+      "step": 1613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.078878404626456e-05,
+      "loss": 0.6471,
+      "step": 1614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077598609623006e-05,
+      "loss": 0.7255,
+      "step": 1615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076318016505083e-05,
+      "loss": 0.7115,
+      "step": 1616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075036625523343e-05,
+      "loss": 0.6709,
+      "step": 1617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07375443692859e-05,
+      "loss": 0.6467,
+      "step": 1618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072471450971793e-05,
+      "loss": 0.7754,
+      "step": 1619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07118766790407e-05,
+      "loss": 0.6955,
+      "step": 1620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0699030879767e-05,
+      "loss": 0.6979,
+      "step": 1621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068617711441114e-05,
+      "loss": 0.7516,
+      "step": 1622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.067331538548901e-05,
+      "loss": 0.6934,
+      "step": 1623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066044569551807e-05,
+      "loss": 0.7401,
+      "step": 1624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064756804701729e-05,
+      "loss": 0.7017,
+      "step": 1625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063468244250726e-05,
+      "loss": 0.6411,
+      "step": 1626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06217888845101e-05,
+      "loss": 0.7185,
+      "step": 1627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060888737554946e-05,
+      "loss": 0.7031,
+      "step": 1628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.059597791815061e-05,
+      "loss": 0.7239,
+      "step": 1629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05830605148403e-05,
+      "loss": 0.6959,
+      "step": 1630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05701351681469e-05,
+      "loss": 0.6769,
+      "step": 1631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055720188060028e-05,
+      "loss": 0.6554,
+      "step": 1632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054426065473192e-05,
+      "loss": 0.6277,
+      "step": 1633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053131149307481e-05,
+      "loss": 0.7485,
+      "step": 1634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.051835439816352e-05,
+      "loss": 0.6817,
+      "step": 1635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050538937253415e-05,
+      "loss": 0.6573,
+      "step": 1636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049241641872435e-05,
+      "loss": 0.6576,
+      "step": 1637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047943553927336e-05,
+      "loss": 0.7272,
+      "step": 1638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.046644673672195e-05,
+      "loss": 0.6092,
+      "step": 1639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045345001361243e-05,
+      "loss": 0.7214,
+      "step": 1640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044044537248865e-05,
+      "loss": 0.6299,
+      "step": 1641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042743281589605e-05,
+      "loss": 0.6748,
+      "step": 1642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04144123463816e-05,
+      "loss": 0.7082,
+      "step": 1643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040138396649379e-05,
+      "loss": 0.7667,
+      "step": 1644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038834767878271e-05,
+      "loss": 0.6785,
+      "step": 1645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037530348579997e-05,
+      "loss": 0.7763,
+      "step": 1646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.03622513900987e-05,
+      "loss": 0.664,
+      "step": 1647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034919139423366e-05,
+      "loss": 0.6855,
+      "step": 1648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033612350076103e-05,
+      "loss": 0.7312,
+      "step": 1649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03230477122387e-05,
+      "loss": 0.6469,
+      "step": 1650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.030996403122592e-05,
+      "loss": 0.7119,
+      "step": 1651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029687246028364e-05,
+      "loss": 0.7163,
+      "step": 1652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028377300197428e-05,
+      "loss": 0.656,
+      "step": 1653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02706656588618e-05,
+      "loss": 0.7054,
+      "step": 1654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025755043351174e-05,
+      "loss": 0.6484,
+      "step": 1655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024442732849114e-05,
+      "loss": 0.4329,
+      "step": 1656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023129634636862e-05,
+      "loss": 0.6945,
+      "step": 1657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021815748971434e-05,
+      "loss": 0.7306,
+      "step": 1658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020501076109995e-05,
+      "loss": 0.6195,
+      "step": 1659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019185616309871e-05,
+      "loss": 0.7124,
+      "step": 1660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01786936982854e-05,
+      "loss": 0.6631,
+      "step": 1661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01655233692363e-05,
+      "loss": 0.6903,
+      "step": 1662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015234517852926e-05,
+      "loss": 0.6839,
+      "step": 1663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013915912874368e-05,
+      "loss": 0.6925,
+      "step": 1664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.012596522246048e-05,
+      "loss": 0.6538,
+      "step": 1665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011276346226211e-05,
+      "loss": 0.721,
+      "step": 1666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.009955385073262e-05,
+      "loss": 0.5917,
+      "step": 1667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008633639045747e-05,
+      "loss": 0.6468,
+      "step": 1668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007311108402379e-05,
+      "loss": 0.6794,
+      "step": 1669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005987793402017e-05,
+      "loss": 0.6842,
+      "step": 1670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004663694303676e-05,
+      "loss": 0.7015,
+      "step": 1671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003338811366524e-05,
+      "loss": 0.7806,
+      "step": 1672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002013144849881e-05,
+      "loss": 0.6981,
+      "step": 1673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000686695013223e-05,
+      "loss": 0.6844,
+      "step": 1674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999359462116179e-05,
+      "loss": 0.7241,
+      "step": 1675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998031446418527e-05,
+      "loss": 0.6826,
+      "step": 1676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996702648180204e-05,
+      "loss": 0.7805,
+      "step": 1677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.995373067661296e-05,
+      "loss": 0.6903,
+      "step": 1678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.994042705122047e-05,
+      "loss": 0.7019,
+      "step": 1679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.992711560822848e-05,
+      "loss": 0.6918,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991379635024247e-05,
+      "loss": 0.6782,
+      "step": 1681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990046927986942e-05,
+      "loss": 0.6737,
+      "step": 1682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988713439971788e-05,
+      "loss": 0.6019,
+      "step": 1683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.98737917123979e-05,
+      "loss": 0.6957,
+      "step": 1684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986044122052104e-05,
+      "loss": 0.6386,
+      "step": 1685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984708292670045e-05,
+      "loss": 0.7459,
+      "step": 1686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983371683355075e-05,
+      "loss": 0.693,
+      "step": 1687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.982034294368811e-05,
+      "loss": 0.6753,
+      "step": 1688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980696125973022e-05,
+      "loss": 0.712,
+      "step": 1689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979357178429629e-05,
+      "loss": 0.6734,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978017452000707e-05,
+      "loss": 0.6285,
+      "step": 1691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.976676946948484e-05,
+      "loss": 0.7234,
+      "step": 1692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975335663535336e-05,
+      "loss": 0.7229,
+      "step": 1693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973993602023796e-05,
+      "loss": 0.668,
+      "step": 1694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.972650762676547e-05,
+      "loss": 0.6619,
+      "step": 1695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971307145756426e-05,
+      "loss": 0.6526,
+      "step": 1696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969962751526421e-05,
+      "loss": 0.4118,
+      "step": 1697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968617580249671e-05,
+      "loss": 0.6774,
+      "step": 1698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967271632189471e-05,
+      "loss": 0.6806,
+      "step": 1699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965924907609263e-05,
+      "loss": 0.4141,
+      "step": 1700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964577406772643e-05,
+      "loss": 0.7114,
+      "step": 1701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.963229129943363e-05,
+      "loss": 0.7059,
+      "step": 1702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96188007738532e-05,
+      "loss": 0.7393,
+      "step": 1703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960530249362567e-05,
+      "loss": 0.6885,
+      "step": 1704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.959179646139306e-05,
+      "loss": 0.7102,
+      "step": 1705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957828267979895e-05,
+      "loss": 0.4152,
+      "step": 1706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956476115148842e-05,
+      "loss": 0.6698,
+      "step": 1707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955123187910805e-05,
+      "loss": 0.7095,
+      "step": 1708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.953769486530593e-05,
+      "loss": 0.7272,
+      "step": 1709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952415011273167e-05,
+      "loss": 0.687,
+      "step": 1710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951059762403644e-05,
+      "loss": 0.7106,
+      "step": 1711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949703740187289e-05,
+      "loss": 0.6502,
+      "step": 1712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948346944889515e-05,
+      "loss": 0.7295,
+      "step": 1713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94698937677589e-05,
+      "loss": 0.6966,
+      "step": 1714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945631036112137e-05,
+      "loss": 0.7041,
+      "step": 1715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94427192316412e-05,
+      "loss": 0.6958,
+      "step": 1716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942912038197865e-05,
+      "loss": 0.7027,
+      "step": 1717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941551381479542e-05,
+      "loss": 0.6353,
+      "step": 1718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940189953275475e-05,
+      "loss": 0.6499,
+      "step": 1719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.938827753852139e-05,
+      "loss": 0.6664,
+      "step": 1720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937464783476157e-05,
+      "loss": 0.7392,
+      "step": 1721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936101042414309e-05,
+      "loss": 0.6892,
+      "step": 1722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934736530933519e-05,
+      "loss": 0.732,
+      "step": 1723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.933371249300865e-05,
+      "loss": 0.7041,
+      "step": 1724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932005197783578e-05,
+      "loss": 0.7147,
+      "step": 1725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930638376649035e-05,
+      "loss": 0.7205,
+      "step": 1726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929270786164765e-05,
+      "loss": 0.3839,
+      "step": 1727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927902426598451e-05,
+      "loss": 0.6805,
+      "step": 1728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926533298217924e-05,
+      "loss": 0.7571,
+      "step": 1729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.925163401291166e-05,
+      "loss": 0.6635,
+      "step": 1730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923792736086305e-05,
+      "loss": 0.7334,
+      "step": 1731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.922421302871629e-05,
+      "loss": 0.6718,
+      "step": 1732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.921049101915567e-05,
+      "loss": 0.7056,
+      "step": 1733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.919676133486702e-05,
+      "loss": 0.6969,
+      "step": 1734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918302397853768e-05,
+      "loss": 0.768,
+      "step": 1735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916927895285649e-05,
+      "loss": 0.7086,
+      "step": 1736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915552626051376e-05,
+      "loss": 0.3861,
+      "step": 1737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914176590420135e-05,
+      "loss": 0.6861,
+      "step": 1738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91279978866126e-05,
+      "loss": 0.6921,
+      "step": 1739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911422221044232e-05,
+      "loss": 0.6679,
+      "step": 1740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910043887838686e-05,
+      "loss": 0.7339,
+      "step": 1741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.908664789314405e-05,
+      "loss": 0.7817,
+      "step": 1742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907284925741322e-05,
+      "loss": 0.677,
+      "step": 1743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.905904297389518e-05,
+      "loss": 0.7152,
+      "step": 1744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90452290452923e-05,
+      "loss": 0.7484,
+      "step": 1745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903140747430833e-05,
+      "loss": 0.716,
+      "step": 1746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901757826364865e-05,
+      "loss": 0.6421,
+      "step": 1747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.900374141602006e-05,
+      "loss": 0.6841,
+      "step": 1748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898989693413083e-05,
+      "loss": 0.6969,
+      "step": 1749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89760448206908e-05,
+      "loss": 0.6562,
+      "step": 1750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.896218507841126e-05,
+      "loss": 0.6832,
+      "step": 1751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.894831771000498e-05,
+      "loss": 0.6761,
+      "step": 1752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893444271818627e-05,
+      "loss": 0.6421,
+      "step": 1753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.892056010567086e-05,
+      "loss": 0.6522,
+      "step": 1754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.890666987517605e-05,
+      "loss": 0.668,
+      "step": 1755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889277202942059e-05,
+      "loss": 0.7688,
+      "step": 1756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.887886657112473e-05,
+      "loss": 0.6533,
+      "step": 1757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886495350301016e-05,
+      "loss": 0.7098,
+      "step": 1758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885103282780016e-05,
+      "loss": 0.6882,
+      "step": 1759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883710454821942e-05,
+      "loss": 0.6558,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.882316866699419e-05,
+      "loss": 0.674,
+      "step": 1761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880922518685207e-05,
+      "loss": 0.6764,
+      "step": 1762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87952741105223e-05,
+      "loss": 0.7065,
+      "step": 1763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.878131544073556e-05,
+      "loss": 0.6845,
+      "step": 1764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876734918022396e-05,
+      "loss": 0.7034,
+      "step": 1765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875337533172114e-05,
+      "loss": 0.6606,
+      "step": 1766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873939389796225e-05,
+      "loss": 0.7135,
+      "step": 1767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.872540488168389e-05,
+      "loss": 0.6767,
+      "step": 1768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871140828562413e-05,
+      "loss": 0.7307,
+      "step": 1769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869740411252255e-05,
+      "loss": 0.641,
+      "step": 1770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868339236512022e-05,
+      "loss": 0.6745,
+      "step": 1771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.866937304615968e-05,
+      "loss": 0.733,
+      "step": 1772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865534615838494e-05,
+      "loss": 0.7049,
+      "step": 1773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864131170454151e-05,
+      "loss": 0.6877,
+      "step": 1774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.862726968737638e-05,
+      "loss": 0.681,
+      "step": 1775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.861322010963801e-05,
+      "loss": 0.6495,
+      "step": 1776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.859916297407636e-05,
+      "loss": 0.7085,
+      "step": 1777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858509828344282e-05,
+      "loss": 0.6854,
+      "step": 1778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.857102604049031e-05,
+      "loss": 0.6735,
+      "step": 1779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855694624797323e-05,
+      "loss": 0.6332,
+      "step": 1780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85428589086474e-05,
+      "loss": 0.7044,
+      "step": 1781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85287640252702e-05,
+      "loss": 0.4019,
+      "step": 1782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851466160060042e-05,
+      "loss": 0.6987,
+      "step": 1783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850055163739834e-05,
+      "loss": 0.6886,
+      "step": 1784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848643413842575e-05,
+      "loss": 0.6678,
+      "step": 1785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.847230910644586e-05,
+      "loss": 0.7011,
+      "step": 1786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84581765442234e-05,
+      "loss": 0.704,
+      "step": 1787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844403645452455e-05,
+      "loss": 0.7254,
+      "step": 1788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.842988884011698e-05,
+      "loss": 0.662,
+      "step": 1789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841573370376982e-05,
+      "loss": 0.7531,
+      "step": 1790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.840157104825368e-05,
+      "loss": 0.6777,
+      "step": 1791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838740087634063e-05,
+      "loss": 0.6452,
+      "step": 1792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.83732231908042e-05,
+      "loss": 0.6482,
+      "step": 1793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.835903799441947e-05,
+      "loss": 0.7129,
+      "step": 1794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.834484528996287e-05,
+      "loss": 0.7112,
+      "step": 1795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.833064508021238e-05,
+      "loss": 0.682,
+      "step": 1796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.831643736794743e-05,
+      "loss": 0.6504,
+      "step": 1797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83022221559489e-05,
+      "loss": 0.7402,
+      "step": 1798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828799944699919e-05,
+      "loss": 0.6853,
+      "step": 1799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827376924388211e-05,
+      "loss": 0.6333,
+      "step": 1800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825953154938296e-05,
+      "loss": 0.6481,
+      "step": 1801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82452863662885e-05,
+      "loss": 0.6956,
+      "step": 1802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.823103369738697e-05,
+      "loss": 0.7229,
+      "step": 1803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.821677354546807e-05,
+      "loss": 0.7195,
+      "step": 1804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820250591332295e-05,
+      "loss": 0.6767,
+      "step": 1805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818823080374423e-05,
+      "loss": 0.6884,
+      "step": 1806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817394821952602e-05,
+      "loss": 0.6983,
+      "step": 1807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815965816346385e-05,
+      "loss": 0.7099,
+      "step": 1808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.814536063835474e-05,
+      "loss": 0.6825,
+      "step": 1809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81310556469972e-05,
+      "loss": 0.6558,
+      "step": 1810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81167431921911e-05,
+      "loss": 0.6766,
+      "step": 1811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81024232767379e-05,
+      "loss": 0.705,
+      "step": 1812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808809590344042e-05,
+      "loss": 0.7149,
+      "step": 1813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.807376107510299e-05,
+      "loss": 0.7117,
+      "step": 1814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.805941879453141e-05,
+      "loss": 0.7022,
+      "step": 1815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.804506906453288e-05,
+      "loss": 0.7141,
+      "step": 1816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80307118879161e-05,
+      "loss": 0.6893,
+      "step": 1817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801634726749124e-05,
+      "loss": 0.7168,
+      "step": 1818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800197520606988e-05,
+      "loss": 0.6937,
+      "step": 1819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798759570646514e-05,
+      "loss": 0.7055,
+      "step": 1820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797320877149147e-05,
+      "loss": 0.6759,
+      "step": 1821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795881440396491e-05,
+      "loss": 0.6291,
+      "step": 1822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.794441260670286e-05,
+      "loss": 0.7095,
+      "step": 1823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793000338252418e-05,
+      "loss": 0.6895,
+      "step": 1824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791558673424927e-05,
+      "loss": 0.6565,
+      "step": 1825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.790116266469988e-05,
+      "loss": 0.714,
+      "step": 1826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788673117669924e-05,
+      "loss": 0.6668,
+      "step": 1827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.787229227307209e-05,
+      "loss": 0.6993,
+      "step": 1828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.785784595664457e-05,
+      "loss": 0.7331,
+      "step": 1829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784339223024425e-05,
+      "loss": 0.6478,
+      "step": 1830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78289310967002e-05,
+      "loss": 0.6686,
+      "step": 1831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781446255884292e-05,
+      "loss": 0.7606,
+      "step": 1832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779998661950435e-05,
+      "loss": 0.7257,
+      "step": 1833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77855032815179e-05,
+      "loss": 0.6824,
+      "step": 1834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77710125477184e-05,
+      "loss": 0.6554,
+      "step": 1835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775651442094215e-05,
+      "loss": 0.7412,
+      "step": 1836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77420089040269e-05,
+      "loss": 0.75,
+      "step": 1837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772749599981183e-05,
+      "loss": 0.7032,
+      "step": 1838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771297571113757e-05,
+      "loss": 0.6718,
+      "step": 1839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769844804084619e-05,
+      "loss": 0.7532,
+      "step": 1840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768391299178121e-05,
+      "loss": 0.7066,
+      "step": 1841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766937056678763e-05,
+      "loss": 0.6954,
+      "step": 1842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765482076871182e-05,
+      "loss": 0.7211,
+      "step": 1843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764026360040165e-05,
+      "loss": 0.393,
+      "step": 1844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762569906470643e-05,
+      "loss": 0.7139,
+      "step": 1845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761112716447688e-05,
+      "loss": 0.646,
+      "step": 1846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759654790256519e-05,
+      "loss": 0.6244,
+      "step": 1847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758196128182496e-05,
+      "loss": 0.7071,
+      "step": 1848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.756736730511128e-05,
+      "loss": 0.6901,
+      "step": 1849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755276597528064e-05,
+      "loss": 0.6926,
+      "step": 1850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753815729519099e-05,
+      "loss": 0.6937,
+      "step": 1851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752354126770168e-05,
+      "loss": 0.7202,
+      "step": 1852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750891789567355e-05,
+      "loss": 0.7012,
+      "step": 1853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.749428718196885e-05,
+      "loss": 0.7329,
+      "step": 1854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.747964912945126e-05,
+      "loss": 0.6423,
+      "step": 1855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746500374098593e-05,
+      "loss": 0.683,
+      "step": 1856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745035101943941e-05,
+      "loss": 0.6688,
+      "step": 1857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74356909676797e-05,
+      "loss": 0.678,
+      "step": 1858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742102358857623e-05,
+      "loss": 0.6932,
+      "step": 1859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740634888499988e-05,
+      "loss": 0.6994,
+      "step": 1860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739166685982294e-05,
+      "loss": 0.7102,
+      "step": 1861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.737697751591916e-05,
+      "loss": 0.6944,
+      "step": 1862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736228085616366e-05,
+      "loss": 0.6871,
+      "step": 1863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.734757688343312e-05,
+      "loss": 0.6096,
+      "step": 1864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733286560060551e-05,
+      "loss": 0.3926,
+      "step": 1865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.731814701056031e-05,
+      "loss": 0.615,
+      "step": 1866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.73034211161784e-05,
+      "loss": 0.6786,
+      "step": 1867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728868792034212e-05,
+      "loss": 0.7295,
+      "step": 1868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.727394742593519e-05,
+      "loss": 0.6998,
+      "step": 1869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725919963584284e-05,
+      "loss": 0.7101,
+      "step": 1870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.724444455295163e-05,
+      "loss": 0.6714,
+      "step": 1871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.722968218014962e-05,
+      "loss": 0.788,
+      "step": 1872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.721491252032625e-05,
+      "loss": 0.6976,
+      "step": 1873
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720013557637243e-05,
+      "loss": 0.7114,
+      "step": 1874
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.718535135118046e-05,
+      "loss": 0.6333,
+      "step": 1875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.717055984764411e-05,
+      "loss": 0.6959,
+      "step": 1876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715576106865848e-05,
+      "loss": 0.6979,
+      "step": 1877
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71409550171202e-05,
+      "loss": 0.674,
+      "step": 1878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.712614169592728e-05,
+      "loss": 0.7572,
+      "step": 1879
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711132110797916e-05,
+      "loss": 0.6546,
+      "step": 1880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709649325617665e-05,
+      "loss": 0.7135,
+      "step": 1881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.708165814342207e-05,
+      "loss": 0.6713,
+      "step": 1882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706681577261913e-05,
+      "loss": 0.6858,
+      "step": 1883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.705196614667291e-05,
+      "loss": 0.7028,
+      "step": 1884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.703710926848995e-05,
+      "loss": 0.6451,
+      "step": 1885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.702224514097826e-05,
+      "loss": 0.7298,
+      "step": 1886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700737376704717e-05,
+      "loss": 0.6895,
+      "step": 1887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699249514960748e-05,
+      "loss": 0.716,
+      "step": 1888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697760929157144e-05,
+      "loss": 0.6615,
+      "step": 1889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696271619585264e-05,
+      "loss": 0.6803,
+      "step": 1890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.694781586536614e-05,
+      "loss": 0.6682,
+      "step": 1891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.693290830302843e-05,
+      "loss": 0.3697,
+      "step": 1892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691799351175733e-05,
+      "loss": 0.6671,
+      "step": 1893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.690307149447221e-05,
+      "loss": 0.6601,
+      "step": 1894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688814225409372e-05,
+      "loss": 0.6769,
+      "step": 1895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687320579354401e-05,
+      "loss": 0.7039,
+      "step": 1896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68582621157466e-05,
+      "loss": 0.6554,
+      "step": 1897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.684331122362646e-05,
+      "loss": 0.6366,
+      "step": 1898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.682835312010992e-05,
+      "loss": 0.6764,
+      "step": 1899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681338780812478e-05,
+      "loss": 0.743,
+      "step": 1900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.679841529060023e-05,
+      "loss": 0.653,
+      "step": 1901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.678343557046682e-05,
+      "loss": 0.7116,
+      "step": 1902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676844865065659e-05,
+      "loss": 0.6686,
+      "step": 1903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675345453410293e-05,
+      "loss": 0.6214,
+      "step": 1904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673845322374068e-05,
+      "loss": 0.3972,
+      "step": 1905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672344472250607e-05,
+      "loss": 0.7067,
+      "step": 1906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670842903333671e-05,
+      "loss": 0.6961,
+      "step": 1907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669340615917168e-05,
+      "loss": 0.6829,
+      "step": 1908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66783761029514e-05,
+      "loss": 0.71,
+      "step": 1909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666333886761773e-05,
+      "loss": 0.7178,
+      "step": 1910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664829445611394e-05,
+      "loss": 0.6287,
+      "step": 1911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.663324287138469e-05,
+      "loss": 0.6771,
+      "step": 1912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.661818411637605e-05,
+      "loss": 0.6611,
+      "step": 1913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66031181940355e-05,
+      "loss": 0.704,
+      "step": 1914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65880451073119e-05,
+      "loss": 0.6702,
+      "step": 1915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657296485915554e-05,
+      "loss": 0.6904,
+      "step": 1916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655787745251809e-05,
+      "loss": 0.7284,
+      "step": 1917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654278289035264e-05,
+      "loss": 0.6919,
+      "step": 1918
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.652768117561367e-05,
+      "loss": 0.6753,
+      "step": 1919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.651257231125706e-05,
+      "loss": 0.6948,
+      "step": 1920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64974563002401e-05,
+      "loss": 0.6891,
+      "step": 1921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.648233314552144e-05,
+      "loss": 0.7437,
+      "step": 1922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646720285006118e-05,
+      "loss": 0.6598,
+      "step": 1923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645206541682081e-05,
+      "loss": 0.6584,
+      "step": 1924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.643692084876316e-05,
+      "loss": 0.6981,
+      "step": 1925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642176914885254e-05,
+      "loss": 0.6223,
+      "step": 1926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.640661032005459e-05,
+      "loss": 0.7365,
+      "step": 1927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639144436533638e-05,
+      "loss": 0.6737,
+      "step": 1928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.637627128766635e-05,
+      "loss": 0.6855,
+      "step": 1929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636109109001438e-05,
+      "loss": 0.6943,
+      "step": 1930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634590377535169e-05,
+      "loss": 0.6442,
+      "step": 1931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.633070934665091e-05,
+      "loss": 0.6877,
+      "step": 1932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63155078068861e-05,
+      "loss": 0.6811,
+      "step": 1933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.630029915903265e-05,
+      "loss": 0.6727,
+      "step": 1934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.628508340606738e-05,
+      "loss": 0.6101,
+      "step": 1935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626986055096848e-05,
+      "loss": 0.7358,
+      "step": 1936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62546305967156e-05,
+      "loss": 0.6742,
+      "step": 1937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623939354628963e-05,
+      "loss": 0.697,
+      "step": 1938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6224149402673e-05,
+      "loss": 0.6376,
+      "step": 1939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.620889816884946e-05,
+      "loss": 0.6455,
+      "step": 1940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.619363984780415e-05,
+      "loss": 0.6621,
+      "step": 1941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617837444252362e-05,
+      "loss": 0.6752,
+      "step": 1942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.616310195599576e-05,
+      "loss": 0.6627,
+      "step": 1943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614782239120991e-05,
+      "loss": 0.6842,
+      "step": 1944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.613253575115674e-05,
+      "loss": 0.7096,
+      "step": 1945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.611724203882833e-05,
+      "loss": 0.6334,
+      "step": 1946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.610194125721814e-05,
+      "loss": 0.6795,
+      "step": 1947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.608663340932104e-05,
+      "loss": 0.6938,
+      "step": 1948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607131849813322e-05,
+      "loss": 0.7114,
+      "step": 1949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.605599652665231e-05,
+      "loss": 0.6923,
+      "step": 1950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60406674978773e-05,
+      "loss": 0.7168,
+      "step": 1951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602533141480856e-05,
+      "loss": 0.7394,
+      "step": 1952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.600998828044786e-05,
+      "loss": 0.727,
+      "step": 1953
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599463809779832e-05,
+      "loss": 0.3928,
+      "step": 1954
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597928086986446e-05,
+      "loss": 0.6834,
+      "step": 1955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.596391659965217e-05,
+      "loss": 0.6838,
+      "step": 1956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.594854529016872e-05,
+      "loss": 0.7263,
+      "step": 1957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.593316694442276e-05,
+      "loss": 0.6784,
+      "step": 1958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59177815654243e-05,
+      "loss": 0.7027,
+      "step": 1959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.590238915618478e-05,
+      "loss": 0.691,
+      "step": 1960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588698971971696e-05,
+      "loss": 0.6399,
+      "step": 1961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.587158325903499e-05,
+      "loss": 0.6578,
+      "step": 1962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585616977715441e-05,
+      "loss": 0.6922,
+      "step": 1963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.584074927709212e-05,
+      "loss": 0.6002,
+      "step": 1964
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582532176186639e-05,
+      "loss": 0.673,
+      "step": 1965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.580988723449688e-05,
+      "loss": 0.7344,
+      "step": 1966
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.579444569800461e-05,
+      "loss": 0.6896,
+      "step": 1967
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577899715541198e-05,
+      "loss": 0.6434,
+      "step": 1968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576354160974274e-05,
+      "loss": 0.6756,
+      "step": 1969
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574807906402205e-05,
+      "loss": 0.6049,
+      "step": 1970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57326095212764e-05,
+      "loss": 0.683,
+      "step": 1971
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571713298453368e-05,
+      "loss": 0.6644,
+      "step": 1972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57016494568231e-05,
+      "loss": 0.7093,
+      "step": 1973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.568615894117533e-05,
+      "loss": 0.7039,
+      "step": 1974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567066144062232e-05,
+      "loss": 0.7158,
+      "step": 1975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.565515695819741e-05,
+      "loss": 0.6898,
+      "step": 1976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563964549693535e-05,
+      "loss": 0.6788,
+      "step": 1977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562412705987219e-05,
+      "loss": 0.3661,
+      "step": 1978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.560860165004539e-05,
+      "loss": 0.6466,
+      "step": 1979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559306927049374e-05,
+      "loss": 0.7025,
+      "step": 1980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557752992425745e-05,
+      "loss": 0.6901,
+      "step": 1981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556198361437803e-05,
+      "loss": 0.6144,
+      "step": 1982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554643034389843e-05,
+      "loss": 0.7467,
+      "step": 1983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553087011586284e-05,
+      "loss": 0.6173,
+      "step": 1984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551530293331694e-05,
+      "loss": 0.631,
+      "step": 1985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549972879930772e-05,
+      "loss": 0.6929,
+      "step": 1986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54841477168835e-05,
+      "loss": 0.6342,
+      "step": 1987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.546855968909401e-05,
+      "loss": 0.6865,
+      "step": 1988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54529647189903e-05,
+      "loss": 0.6416,
+      "step": 1989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543736280962482e-05,
+      "loss": 0.7031,
+      "step": 1990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.542175396405134e-05,
+      "loss": 0.705,
+      "step": 1991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540613818532499e-05,
+      "loss": 0.6983,
+      "step": 1992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.53905154765023e-05,
+      "loss": 0.7317,
+      "step": 1993
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.537488584064112e-05,
+      "loss": 0.6266,
+      "step": 1994
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535924928080064e-05,
+      "loss": 0.6589,
+      "step": 1995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534360580004144e-05,
+      "loss": 0.7043,
+      "step": 1996
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532795540142544e-05,
+      "loss": 0.6694,
+      "step": 1997
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531229808801593e-05,
+      "loss": 0.6663,
+      "step": 1998
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.529663386287751e-05,
+      "loss": 0.6215,
+      "step": 1999
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52809627290762e-05,
+      "loss": 0.7072,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.526528468967929e-05,
+      "loss": 0.7498,
+      "step": 2001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.524959974775551e-05,
+      "loss": 0.3858,
+      "step": 2002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523390790637487e-05,
+      "loss": 0.6194,
+      "step": 2003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.521820916860877e-05,
+      "loss": 0.7351,
+      "step": 2004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.520250353752994e-05,
+      "loss": 0.672,
+      "step": 2005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.518679101621246e-05,
+      "loss": 0.6389,
+      "step": 2006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.517107160773178e-05,
+      "loss": 0.6869,
+      "step": 2007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.515534531516467e-05,
+      "loss": 0.4173,
+      "step": 2008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.513961214158928e-05,
+      "loss": 0.6925,
+      "step": 2009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512387209008507e-05,
+      "loss": 0.7129,
+      "step": 2010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510812516373288e-05,
+      "loss": 0.6816,
+      "step": 2011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.509237136561486e-05,
+      "loss": 0.6934,
+      "step": 2012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.507661069881453e-05,
+      "loss": 0.7006,
+      "step": 2013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.506084316641676e-05,
+      "loss": 0.6646,
+      "step": 2014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.504506877150773e-05,
+      "loss": 0.7168,
+      "step": 2015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.502928751717503e-05,
+      "loss": 0.6713,
+      "step": 2016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501349940650749e-05,
+      "loss": 0.7591,
+      "step": 2017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499770444259539e-05,
+      "loss": 0.694,
+      "step": 2018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498190262853027e-05,
+      "loss": 0.6656,
+      "step": 2019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496609396740506e-05,
+      "loss": 0.6858,
+      "step": 2020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4950278462314e-05,
+      "loss": 0.6448,
+      "step": 2021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49344561163527e-05,
+      "loss": 0.639,
+      "step": 2022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491862693261809e-05,
+      "loss": 0.681,
+      "step": 2023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49027909142084e-05,
+      "loss": 0.6914,
+      "step": 2024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488694806422329e-05,
+      "loss": 0.6855,
+      "step": 2025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.487109838576368e-05,
+      "loss": 0.7224,
+      "step": 2026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.485524188193186e-05,
+      "loss": 0.6325,
+      "step": 2027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.483937855583144e-05,
+      "loss": 0.6967,
+      "step": 2028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.482350841056737e-05,
+      "loss": 0.699,
+      "step": 2029
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.480763144924595e-05,
+      "loss": 0.3771,
+      "step": 2030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479174767497478e-05,
+      "loss": 0.7191,
+      "step": 2031
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477585709086284e-05,
+      "loss": 0.7161,
+      "step": 2032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.47599597000204e-05,
+      "loss": 0.7194,
+      "step": 2033
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.474405550555908e-05,
+      "loss": 0.6981,
+      "step": 2034
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.472814451059185e-05,
+      "loss": 0.6692,
+      "step": 2035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471222671823298e-05,
+      "loss": 0.7411,
+      "step": 2036
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469630213159808e-05,
+      "loss": 0.7599,
+      "step": 2037
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.468037075380408e-05,
+      "loss": 0.5894,
+      "step": 2038
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466443258796929e-05,
+      "loss": 0.7713,
+      "step": 2039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.464848763721329e-05,
+      "loss": 0.7109,
+      "step": 2040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4632535904657e-05,
+      "loss": 0.6583,
+      "step": 2041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46165773934227e-05,
+      "loss": 0.6795,
+      "step": 2042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.460061210663393e-05,
+      "loss": 0.6808,
+      "step": 2043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.458464004741565e-05,
+      "loss": 0.6589,
+      "step": 2044
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456866121889407e-05,
+      "loss": 0.6884,
+      "step": 2045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.455267562419675e-05,
+      "loss": 0.71,
+      "step": 2046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453668326645259e-05,
+      "loss": 0.6721,
+      "step": 2047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.452068414879177e-05,
+      "loss": 0.7013,
+      "step": 2048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.450467827434584e-05,
+      "loss": 0.7267,
+      "step": 2049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448866564624767e-05,
+      "loss": 0.701,
+      "step": 2050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44726462676314e-05,
+      "loss": 0.6749,
+      "step": 2051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.445662014163254e-05,
+      "loss": 0.663,
+      "step": 2052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.444058727138792e-05,
+      "loss": 0.7209,
+      "step": 2053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442454766003565e-05,
+      "loss": 0.6983,
+      "step": 2054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.440850131071523e-05,
+      "loss": 0.631,
+      "step": 2055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43924482265674e-05,
+      "loss": 0.7198,
+      "step": 2056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437638841073427e-05,
+      "loss": 0.6788,
+      "step": 2057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.436032186635925e-05,
+      "loss": 0.6297,
+      "step": 2058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.434424859658705e-05,
+      "loss": 0.6609,
+      "step": 2059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.432816860456375e-05,
+      "loss": 0.6567,
+      "step": 2060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43120818934367e-05,
+      "loss": 0.7553,
+      "step": 2061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.429598846635454e-05,
+      "loss": 0.7038,
+      "step": 2062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42798883264673e-05,
+      "loss": 0.6413,
+      "step": 2063
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.426378147692629e-05,
+      "loss": 0.6948,
+      "step": 2064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424766792088408e-05,
+      "loss": 0.6599,
+      "step": 2065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423154766149466e-05,
+      "loss": 0.7356,
+      "step": 2066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.421542070191324e-05,
+      "loss": 0.589,
+      "step": 2067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419928704529636e-05,
+      "loss": 0.6901,
+      "step": 2068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418314669480191e-05,
+      "loss": 0.6598,
+      "step": 2069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.416699965358908e-05,
+      "loss": 0.7207,
+      "step": 2070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41508459248183e-05,
+      "loss": 0.6552,
+      "step": 2071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41346855116514e-05,
+      "loss": 0.622,
+      "step": 2072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.411851841725149e-05,
+      "loss": 0.6144,
+      "step": 2073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410234464478297e-05,
+      "loss": 0.7014,
+      "step": 2074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.408616419741155e-05,
+      "loss": 0.6972,
+      "step": 2075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406997707830424e-05,
+      "loss": 0.7143,
+      "step": 2076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405378329062941e-05,
+      "loss": 0.6654,
+      "step": 2077
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.403758283755665e-05,
+      "loss": 0.4029,
+      "step": 2078
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.402137572225692e-05,
+      "loss": 0.6878,
+      "step": 2079
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400516194790248e-05,
+      "loss": 0.7293,
+      "step": 2080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.398894151766686e-05,
+      "loss": 0.3711,
+      "step": 2081
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.397271443472487e-05,
+      "loss": 0.5885,
+      "step": 2082
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.395648070225272e-05,
+      "loss": 0.6769,
+      "step": 2083
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.394024032342786e-05,
+      "loss": 0.6289,
+      "step": 2084
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.392399330142901e-05,
+      "loss": 0.6076,
+      "step": 2085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.390773963943623e-05,
+      "loss": 0.6818,
+      "step": 2086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38914793406309e-05,
+      "loss": 0.6769,
+      "step": 2087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.387521240819565e-05,
+      "loss": 0.6859,
+      "step": 2088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.385893884531442e-05,
+      "loss": 0.6782,
+      "step": 2089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.384265865517249e-05,
+      "loss": 0.3587,
+      "step": 2090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.382637184095639e-05,
+      "loss": 0.7036,
+      "step": 2091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381007840585395e-05,
+      "loss": 0.708,
+      "step": 2092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.379377835305433e-05,
+      "loss": 0.6803,
+      "step": 2093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377747168574796e-05,
+      "loss": 0.6777,
+      "step": 2094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.376115840712655e-05,
+      "loss": 0.657,
+      "step": 2095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374483852038315e-05,
+      "loss": 0.6542,
+      "step": 2096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372851202871205e-05,
+      "loss": 0.6922,
+      "step": 2097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37121789353089e-05,
+      "loss": 0.7073,
+      "step": 2098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.369583924337055e-05,
+      "loss": 0.6196,
+      "step": 2099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367949295609521e-05,
+      "loss": 0.7329,
+      "step": 2100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.366314007668235e-05,
+      "loss": 0.715,
+      "step": 2101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.364678060833278e-05,
+      "loss": 0.6611,
+      "step": 2102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.363041455424855e-05,
+      "loss": 0.7077,
+      "step": 2103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3614041917633e-05,
+      "loss": 0.7145,
+      "step": 2104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.359766270169078e-05,
+      "loss": 0.6766,
+      "step": 2105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.358127690962781e-05,
+      "loss": 0.6419,
+      "step": 2106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35648845446513e-05,
+      "loss": 0.7157,
+      "step": 2107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354848560996978e-05,
+      "loss": 0.6537,
+      "step": 2108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.353208010879301e-05,
+      "loss": 0.7567,
+      "step": 2109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.351566804433207e-05,
+      "loss": 0.6206,
+      "step": 2110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.349924941979933e-05,
+      "loss": 0.7212,
+      "step": 2111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348282423840841e-05,
+      "loss": 0.6515,
+      "step": 2112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.346639250337427e-05,
+      "loss": 0.7094,
+      "step": 2113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.344995421791308e-05,
+      "loss": 0.6434,
+      "step": 2114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.343350938524237e-05,
+      "loss": 0.7126,
+      "step": 2115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.341705800858087e-05,
+      "loss": 0.643,
+      "step": 2116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340060009114866e-05,
+      "loss": 0.7338,
+      "step": 2117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.338413563616705e-05,
+      "loss": 0.7207,
+      "step": 2118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.336766464685869e-05,
+      "loss": 0.6736,
+      "step": 2119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335118712644742e-05,
+      "loss": 0.6472,
+      "step": 2120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.333470307815845e-05,
+      "loss": 0.6188,
+      "step": 2121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33182125052182e-05,
+      "loss": 0.698,
+      "step": 2122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330171541085443e-05,
+      "loss": 0.6477,
+      "step": 2123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.328521179829613e-05,
+      "loss": 0.7146,
+      "step": 2124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326870167077354e-05,
+      "loss": 0.6227,
+      "step": 2125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.325218503151826e-05,
+      "loss": 0.6692,
+      "step": 2126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323566188376309e-05,
+      "loss": 0.6712,
+      "step": 2127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321913223074212e-05,
+      "loss": 0.6548,
+      "step": 2128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320259607569077e-05,
+      "loss": 0.7641,
+      "step": 2129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.318605342184563e-05,
+      "loss": 0.6787,
+      "step": 2130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.316950427244466e-05,
+      "loss": 0.6182,
+      "step": 2131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315294863072702e-05,
+      "loss": 0.6687,
+      "step": 2132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313638649993318e-05,
+      "loss": 0.6819,
+      "step": 2133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31198178833049e-05,
+      "loss": 0.4073,
+      "step": 2134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.310324278408515e-05,
+      "loss": 0.6681,
+      "step": 2135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.308666120551819e-05,
+      "loss": 0.7054,
+      "step": 2136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.307007315084958e-05,
+      "loss": 0.7071,
+      "step": 2137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30534786233261e-05,
+      "loss": 0.6877,
+      "step": 2138
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.303687762619586e-05,
+      "loss": 0.6638,
+      "step": 2139
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.302027016270816e-05,
+      "loss": 0.5892,
+      "step": 2140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300365623611363e-05,
+      "loss": 0.7371,
+      "step": 2141
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298703584966414e-05,
+      "loss": 0.7339,
+      "step": 2142
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29704090066128e-05,
+      "loss": 0.7081,
+      "step": 2143
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.295377571021402e-05,
+      "loss": 0.6896,
+      "step": 2144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.293713596372343e-05,
+      "loss": 0.6652,
+      "step": 2145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.292048977039801e-05,
+      "loss": 0.6845,
+      "step": 2146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.290383713349592e-05,
+      "loss": 0.6803,
+      "step": 2147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.288717805627659e-05,
+      "loss": 0.7712,
+      "step": 2148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.287051254200073e-05,
+      "loss": 0.7104,
+      "step": 2149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.285384059393032e-05,
+      "loss": 0.7045,
+      "step": 2150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.283716221532858e-05,
+      "loss": 0.6398,
+      "step": 2151
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.282047740945997e-05,
+      "loss": 0.6461,
+      "step": 2152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280378617959029e-05,
+      "loss": 0.6899,
+      "step": 2153
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.278708852898647e-05,
+      "loss": 0.6608,
+      "step": 2154
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.277038446091683e-05,
+      "loss": 0.7556,
+      "step": 2155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.275367397865084e-05,
+      "loss": 0.6486,
+      "step": 2156
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.273695708545927e-05,
+      "loss": 0.6586,
+      "step": 2157
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272023378461415e-05,
+      "loss": 0.6589,
+      "step": 2158
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270350407938877e-05,
+      "loss": 0.6645,
+      "step": 2159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268676797305764e-05,
+      "loss": 0.5961,
+      "step": 2160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.267002546889656e-05,
+      "loss": 0.719,
+      "step": 2161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.265327657018256e-05,
+      "loss": 0.6837,
+      "step": 2162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.26365212801939e-05,
+      "loss": 0.6309,
+      "step": 2163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.261975960221017e-05,
+      "loss": 0.6457,
+      "step": 2164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.260299153951212e-05,
+      "loss": 0.6954,
+      "step": 2165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.258621709538179e-05,
+      "loss": 0.6983,
+      "step": 2166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25694362731025e-05,
+      "loss": 0.6773,
+      "step": 2167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.255264907595873e-05,
+      "loss": 0.6458,
+      "step": 2168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.253585550723631e-05,
+      "loss": 0.6907,
+      "step": 2169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.251905557022225e-05,
+      "loss": 0.7242,
+      "step": 2170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250224926820482e-05,
+      "loss": 0.6958,
+      "step": 2171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.248543660447358e-05,
+      "loss": 0.6317,
+      "step": 2172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.246861758231925e-05,
+      "loss": 0.6981,
+      "step": 2173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245179220503387e-05,
+      "loss": 0.6833,
+      "step": 2174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.243496047591071e-05,
+      "loss": 0.7166,
+      "step": 2175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241812239824422e-05,
+      "loss": 0.6256,
+      "step": 2176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.240127797533019e-05,
+      "loss": 0.6706,
+      "step": 2177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.238442721046557e-05,
+      "loss": 0.3947,
+      "step": 2178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236757010694861e-05,
+      "loss": 0.6743,
+      "step": 2179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.235070666807877e-05,
+      "loss": 0.6332,
+      "step": 2180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233383689715674e-05,
+      "loss": 0.6481,
+      "step": 2181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23169607974845e-05,
+      "loss": 0.7377,
+      "step": 2182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.230007837236519e-05,
+      "loss": 0.6635,
+      "step": 2183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.228318962510327e-05,
+      "loss": 0.6325,
+      "step": 2184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.226629455900439e-05,
+      "loss": 0.3611,
+      "step": 2185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.224939317737543e-05,
+      "loss": 0.5948,
+      "step": 2186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.223248548352454e-05,
+      "loss": 0.7252,
+      "step": 2187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.221557148076107e-05,
+      "loss": 0.666,
+      "step": 2188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219865117239564e-05,
+      "loss": 0.6712,
+      "step": 2189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.218172456174009e-05,
+      "loss": 0.6459,
+      "step": 2190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.216479165210748e-05,
+      "loss": 0.6468,
+      "step": 2191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.214785244681214e-05,
+      "loss": 0.6537,
+      "step": 2192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.213090694916955e-05,
+      "loss": 0.3773,
+      "step": 2193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.211395516249652e-05,
+      "loss": 0.6432,
+      "step": 2194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.209699709011105e-05,
+      "loss": 0.7328,
+      "step": 2195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.208003273533235e-05,
+      "loss": 0.6262,
+      "step": 2196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.206306210148089e-05,
+      "loss": 0.7182,
+      "step": 2197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.204608519187835e-05,
+      "loss": 0.7004,
+      "step": 2198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.202910200984764e-05,
+      "loss": 0.6774,
+      "step": 2199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201211255871293e-05,
+      "loss": 0.6972,
+      "step": 2200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199511684179957e-05,
+      "loss": 0.5966,
+      "step": 2201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197811486243416e-05,
+      "loss": 0.6504,
+      "step": 2202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.196110662394453e-05,
+      "loss": 0.6369,
+      "step": 2203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194409212965972e-05,
+      "loss": 0.6876,
+      "step": 2204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.192707138291e-05,
+      "loss": 0.6539,
+      "step": 2205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191004438702687e-05,
+      "loss": 0.6425,
+      "step": 2206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189301114534305e-05,
+      "loss": 0.717,
+      "step": 2207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18759716611925e-05,
+      "loss": 0.7149,
+      "step": 2208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.185892593791034e-05,
+      "loss": 0.6387,
+      "step": 2209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184187397883301e-05,
+      "loss": 0.6272,
+      "step": 2210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.182481578729806e-05,
+      "loss": 0.7282,
+      "step": 2211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.180775136664436e-05,
+      "loss": 0.6474,
+      "step": 2212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179068072021192e-05,
+      "loss": 0.7094,
+      "step": 2213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177360385134204e-05,
+      "loss": 0.7029,
+      "step": 2214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.175652076337716e-05,
+      "loss": 0.7342,
+      "step": 2215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.173943145966099e-05,
+      "loss": 0.6775,
+      "step": 2216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.172233594353846e-05,
+      "loss": 0.3957,
+      "step": 2217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170523421835572e-05,
+      "loss": 0.68,
+      "step": 2218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.168812628746007e-05,
+      "loss": 0.6566,
+      "step": 2219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16710121542001e-05,
+      "loss": 0.7219,
+      "step": 2220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.165389182192558e-05,
+      "loss": 0.6193,
+      "step": 2221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163676529398747e-05,
+      "loss": 0.6748,
+      "step": 2222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161963257373801e-05,
+      "loss": 0.6651,
+      "step": 2223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.160249366453061e-05,
+      "loss": 0.7521,
+      "step": 2224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.158534856971987e-05,
+      "loss": 0.6779,
+      "step": 2225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.156819729266165e-05,
+      "loss": 0.6879,
+      "step": 2226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.155103983671297e-05,
+      "loss": 0.6205,
+      "step": 2227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15338762052321e-05,
+      "loss": 0.7059,
+      "step": 2228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.151670640157852e-05,
+      "loss": 0.7277,
+      "step": 2229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149953042911285e-05,
+      "loss": 0.6458,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.148234829119702e-05,
+      "loss": 0.7091,
+      "step": 2231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.14651599911941e-05,
+      "loss": 0.7051,
+      "step": 2232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.144796553246836e-05,
+      "loss": 0.6561,
+      "step": 2233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.143076491838533e-05,
+      "loss": 0.6613,
+      "step": 2234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.14135581523117e-05,
+      "loss": 0.6674,
+      "step": 2235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139634523761537e-05,
+      "loss": 0.7236,
+      "step": 2236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.137912617766544e-05,
+      "loss": 0.5967,
+      "step": 2237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.136190097583225e-05,
+      "loss": 0.645,
+      "step": 2238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.134466963548729e-05,
+      "loss": 0.6686,
+      "step": 2239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.132743216000329e-05,
+      "loss": 0.6959,
+      "step": 2240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.131018855275416e-05,
+      "loss": 0.6591,
+      "step": 2241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.129293881711501e-05,
+      "loss": 0.6168,
+      "step": 2242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12756829564622e-05,
+      "loss": 0.6462,
+      "step": 2243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.125842097417318e-05,
+      "loss": 0.7092,
+      "step": 2244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12411528736267e-05,
+      "loss": 0.7074,
+      "step": 2245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.122387865820268e-05,
+      "loss": 0.6327,
+      "step": 2246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12065983312822e-05,
+      "loss": 0.6852,
+      "step": 2247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.118931189624758e-05,
+      "loss": 0.6186,
+      "step": 2248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.117201935648233e-05,
+      "loss": 0.6977,
+      "step": 2249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115472071537111e-05,
+      "loss": 0.6665,
+      "step": 2250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.113741597629985e-05,
+      "loss": 0.6918,
+      "step": 2251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.11201051426556e-05,
+      "loss": 0.7235,
+      "step": 2252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110278821782666e-05,
+      "loss": 0.6934,
+      "step": 2253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10854652052025e-05,
+      "loss": 0.6726,
+      "step": 2254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.106813610817374e-05,
+      "loss": 0.6197,
+      "step": 2255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10508009301323e-05,
+      "loss": 0.6148,
+      "step": 2256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.103345967447113e-05,
+      "loss": 0.4166,
+      "step": 2257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101611234458452e-05,
+      "loss": 0.6679,
+      "step": 2258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.099875894386787e-05,
+      "loss": 0.6882,
+      "step": 2259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.098139947571781e-05,
+      "loss": 0.6527,
+      "step": 2260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.096403394353209e-05,
+      "loss": 0.6533,
+      "step": 2261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.094666235070974e-05,
+      "loss": 0.7149,
+      "step": 2262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.092928470065091e-05,
+      "loss": 0.677,
+      "step": 2263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.091190099675694e-05,
+      "loss": 0.6879,
+      "step": 2264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089451124243038e-05,
+      "loss": 0.7052,
+      "step": 2265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087711544107494e-05,
+      "loss": 0.6721,
+      "step": 2266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.085971359609554e-05,
+      "loss": 0.4016,
+      "step": 2267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.084230571089828e-05,
+      "loss": 0.661,
+      "step": 2268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.082489178889041e-05,
+      "loss": 0.6933,
+      "step": 2269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08074718334804e-05,
+      "loss": 0.6363,
+      "step": 2270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.079004584807786e-05,
+      "loss": 0.6441,
+      "step": 2271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.077261383609363e-05,
+      "loss": 0.6994,
+      "step": 2272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.075517580093968e-05,
+      "loss": 0.6442,
+      "step": 2273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.073773174602919e-05,
+      "loss": 0.6844,
+      "step": 2274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.072028167477653e-05,
+      "loss": 0.6955,
+      "step": 2275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07028255905972e-05,
+      "loss": 0.6235,
+      "step": 2276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068536349690794e-05,
+      "loss": 0.7476,
+      "step": 2277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06678953971266e-05,
+      "loss": 0.6991,
+      "step": 2278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065042129467227e-05,
+      "loss": 0.6959,
+      "step": 2279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063294119296514e-05,
+      "loss": 0.6229,
+      "step": 2280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061545509542663e-05,
+      "loss": 0.6664,
+      "step": 2281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.059796300547934e-05,
+      "loss": 0.6776,
+      "step": 2282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.0580464926547e-05,
+      "loss": 0.6255,
+      "step": 2283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056296086205457e-05,
+      "loss": 0.6377,
+      "step": 2284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05454508154281e-05,
+      "loss": 0.5976,
+      "step": 2285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052793479009488e-05,
+      "loss": 0.6503,
+      "step": 2286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.051041278948337e-05,
+      "loss": 0.7365,
+      "step": 2287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.049288481702313e-05,
+      "loss": 0.6561,
+      "step": 2288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.047535087614497e-05,
+      "loss": 0.3943,
+      "step": 2289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.045781097028083e-05,
+      "loss": 0.6721,
+      "step": 2290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.044026510286382e-05,
+      "loss": 0.6985,
+      "step": 2291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.042271327732822e-05,
+      "loss": 0.7114,
+      "step": 2292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.040515549710945e-05,
+      "loss": 0.6596,
+      "step": 2293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.038759176564416e-05,
+      "loss": 0.6631,
+      "step": 2294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03700220863701e-05,
+      "loss": 0.7086,
+      "step": 2295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.035244646272621e-05,
+      "loss": 0.7082,
+      "step": 2296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.033486489815263e-05,
+      "loss": 0.6598,
+      "step": 2297
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031727739609058e-05,
+      "loss": 0.702,
+      "step": 2298
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02996839599825e-05,
+      "loss": 0.7055,
+      "step": 2299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.028208459327198e-05,
+      "loss": 0.632,
+      "step": 2300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.026447929940378e-05,
+      "loss": 0.3765,
+      "step": 2301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024686808182381e-05,
+      "loss": 0.6656,
+      "step": 2302
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022925094397913e-05,
+      "loss": 0.7271,
+      "step": 2303
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.021162788931798e-05,
+      "loss": 0.7161,
+      "step": 2304
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.019399892128973e-05,
+      "loss": 0.7315,
+      "step": 2305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.017636404334495e-05,
+      "loss": 0.6799,
+      "step": 2306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01587232589353e-05,
+      "loss": 0.6926,
+      "step": 2307
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.014107657151369e-05,
+      "loss": 0.6477,
+      "step": 2308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.012342398453408e-05,
+      "loss": 0.6738,
+      "step": 2309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010576550145168e-05,
+      "loss": 0.6747,
+      "step": 2310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.008810112572278e-05,
+      "loss": 0.637,
+      "step": 2311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.007043086080488e-05,
+      "loss": 0.7107,
+      "step": 2312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.005275471015659e-05,
+      "loss": 0.667,
+      "step": 2313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00350726772377e-05,
+      "loss": 0.6284,
+      "step": 2314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.001738476550913e-05,
+      "loss": 0.6705,
+      "step": 2315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999969097843298e-05,
+      "loss": 0.653,
+      "step": 2316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.998199131947247e-05,
+      "loss": 0.6858,
+      "step": 2317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.996428579209195e-05,
+      "loss": 0.6899,
+      "step": 2318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.9946574399757e-05,
+      "loss": 0.6573,
+      "step": 2319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.992885714593428e-05,
+      "loss": 0.7336,
+      "step": 2320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99111340340916e-05,
+      "loss": 0.7165,
+      "step": 2321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989340506769795e-05,
+      "loss": 0.7242,
+      "step": 2322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987567025022342e-05,
+      "loss": 0.6416,
+      "step": 2323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.985792958513931e-05,
+      "loss": 0.6617,
+      "step": 2324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984018307591801e-05,
+      "loss": 0.7009,
+      "step": 2325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.982243072603306e-05,
+      "loss": 0.3918,
+      "step": 2326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.980467253895917e-05,
+      "loss": 0.3815,
+      "step": 2327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.978690851817216e-05,
+      "loss": 0.7243,
+      "step": 2328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.976913866714904e-05,
+      "loss": 0.6769,
+      "step": 2329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.975136298936787e-05,
+      "loss": 0.6346,
+      "step": 2330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973358148830799e-05,
+      "loss": 0.7274,
+      "step": 2331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.971579416744972e-05,
+      "loss": 0.6627,
+      "step": 2332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.969800103027466e-05,
+      "loss": 0.6449,
+      "step": 2333
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.968020208026545e-05,
+      "loss": 0.6786,
+      "step": 2334
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966239732090592e-05,
+      "loss": 0.7672,
+      "step": 2335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.964458675568103e-05,
+      "loss": 0.6953,
+      "step": 2336
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.962677038807684e-05,
+      "loss": 0.7323,
+      "step": 2337
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96089482215806e-05,
+      "loss": 0.687,
+      "step": 2338
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.959112025968068e-05,
+      "loss": 0.6684,
+      "step": 2339
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.957328650586654e-05,
+      "loss": 0.6353,
+      "step": 2340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955544696362882e-05,
+      "loss": 0.6801,
+      "step": 2341
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.953760163645929e-05,
+      "loss": 0.7114,
+      "step": 2342
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.951975052785084e-05,
+      "loss": 0.6586,
+      "step": 2343
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95018936412975e-05,
+      "loss": 0.663,
+      "step": 2344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.948403098029439e-05,
+      "loss": 0.6486,
+      "step": 2345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.946616254833784e-05,
+      "loss": 0.6504,
+      "step": 2346
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944828834892524e-05,
+      "loss": 0.7031,
+      "step": 2347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.943040838555514e-05,
+      "loss": 0.6486,
+      "step": 2348
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.94125226617272e-05,
+      "loss": 0.7219,
+      "step": 2349
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939463118094226e-05,
+      "loss": 0.699,
+      "step": 2350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93767339467022e-05,
+      "loss": 0.7171,
+      "step": 2351
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.935883096251009e-05,
+      "loss": 0.7306,
+      "step": 2352
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93409222318701e-05,
+      "loss": 0.6432,
+      "step": 2353
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932300775828756e-05,
+      "loss": 0.6777,
+      "step": 2354
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.930508754526886e-05,
+      "loss": 0.5838,
+      "step": 2355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.928716159632157e-05,
+      "loss": 0.6953,
+      "step": 2356
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.926922991495437e-05,
+      "loss": 0.7704,
+      "step": 2357
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925129250467703e-05,
+      "loss": 0.7114,
+      "step": 2358
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923334936900049e-05,
+      "loss": 0.6435,
+      "step": 2359
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.921540051143678e-05,
+      "loss": 0.6792,
+      "step": 2360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.919744593549905e-05,
+      "loss": 0.621,
+      "step": 2361
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.917948564470157e-05,
+      "loss": 0.3913,
+      "step": 2362
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.916151964255976e-05,
+      "loss": 0.6695,
+      "step": 2363
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914354793259011e-05,
+      "loss": 0.7042,
+      "step": 2364
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.912557051831026e-05,
+      "loss": 0.694,
+      "step": 2365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.910758740323896e-05,
+      "loss": 0.4068,
+      "step": 2366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.908959859089605e-05,
+      "loss": 0.7049,
+      "step": 2367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.907160408480251e-05,
+      "loss": 0.6462,
+      "step": 2368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905360388848047e-05,
+      "loss": 0.38,
+      "step": 2369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.903559800545312e-05,
+      "loss": 0.7054,
+      "step": 2370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.901758643924475e-05,
+      "loss": 0.6773,
+      "step": 2371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89995691933808e-05,
+      "loss": 0.6745,
+      "step": 2372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898154627138783e-05,
+      "loss": 0.6651,
+      "step": 2373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.896351767679349e-05,
+      "loss": 0.693,
+      "step": 2374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894548341312652e-05,
+      "loss": 0.6927,
+      "step": 2375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892744348391682e-05,
+      "loss": 0.6883,
+      "step": 2376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.890939789269537e-05,
+      "loss": 0.6476,
+      "step": 2377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889134664299424e-05,
+      "loss": 0.6553,
+      "step": 2378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.887328973834667e-05,
+      "loss": 0.6795,
+      "step": 2379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88552271822869e-05,
+      "loss": 0.6427,
+      "step": 2380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88371589783504e-05,
+      "loss": 0.6298,
+      "step": 2381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881908513007368e-05,
+      "loss": 0.7011,
+      "step": 2382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.880100564099432e-05,
+      "loss": 0.7219,
+      "step": 2383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.878292051465108e-05,
+      "loss": 0.7035,
+      "step": 2384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.876482975458377e-05,
+      "loss": 0.6659,
+      "step": 2385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874673336433335e-05,
+      "loss": 0.7419,
+      "step": 2386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.872863134744184e-05,
+      "loss": 0.688,
+      "step": 2387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.871052370745236e-05,
+      "loss": 0.5909,
+      "step": 2388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.869241044790915e-05,
+      "loss": 0.6503,
+      "step": 2389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867429157235755e-05,
+      "loss": 0.7081,
+      "step": 2390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8656167084344e-05,
+      "loss": 0.6523,
+      "step": 2391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.863803698741603e-05,
+      "loss": 0.6478,
+      "step": 2392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.861990128512227e-05,
+      "loss": 0.6769,
+      "step": 2393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.860175998101246e-05,
+      "loss": 0.6165,
+      "step": 2394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85836130786374e-05,
+      "loss": 0.6571,
+      "step": 2395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.856546058154902e-05,
+      "loss": 0.6254,
+      "step": 2396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854730249330035e-05,
+      "loss": 0.4041,
+      "step": 2397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.852913881744547e-05,
+      "loss": 0.6786,
+      "step": 2398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.851096955753962e-05,
+      "loss": 0.6528,
+      "step": 2399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.849279471713907e-05,
+      "loss": 0.6826,
+      "step": 2400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847461429980121e-05,
+      "loss": 0.7321,
+      "step": 2401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.845642830908452e-05,
+      "loss": 0.6565,
+      "step": 2402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84382367485486e-05,
+      "loss": 0.6684,
+      "step": 2403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.842003962175408e-05,
+      "loss": 0.7107,
+      "step": 2404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.840183693226273e-05,
+      "loss": 0.6729,
+      "step": 2405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.838362868363736e-05,
+      "loss": 0.6322,
+      "step": 2406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.836541487944193e-05,
+      "loss": 0.6699,
+      "step": 2407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834719552324144e-05,
+      "loss": 0.7381,
+      "step": 2408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8328970618602e-05,
+      "loss": 0.699,
+      "step": 2409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.831074016909079e-05,
+      "loss": 0.6347,
+      "step": 2410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.829250417827608e-05,
+      "loss": 0.7127,
+      "step": 2411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.827426264972723e-05,
+      "loss": 0.3937,
+      "step": 2412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82560155870147e-05,
+      "loss": 0.6845,
+      "step": 2413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.823776299370998e-05,
+      "loss": 0.7296,
+      "step": 2414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.821950487338572e-05,
+      "loss": 0.6327,
+      "step": 2415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820124122961557e-05,
+      "loss": 0.6376,
+      "step": 2416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.818297206597431e-05,
+      "loss": 0.7023,
+      "step": 2417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.816469738603783e-05,
+      "loss": 0.6902,
+      "step": 2418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.814641719338298e-05,
+      "loss": 0.6753,
+      "step": 2419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.812813149158785e-05,
+      "loss": 0.7345,
+      "step": 2420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81098402842315e-05,
+      "loss": 0.6692,
+      "step": 2421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.809154357489406e-05,
+      "loss": 0.6564,
+      "step": 2422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.807324136715682e-05,
+      "loss": 0.623,
+      "step": 2423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.805493366460207e-05,
+      "loss": 0.7094,
+      "step": 2424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.803662047081323e-05,
+      "loss": 0.6524,
+      "step": 2425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.801830178937476e-05,
+      "loss": 0.7023,
+      "step": 2426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.799997762387219e-05,
+      "loss": 0.6325,
+      "step": 2427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.798164797789214e-05,
+      "loss": 0.7176,
+      "step": 2428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79633128550223e-05,
+      "loss": 0.7466,
+      "step": 2429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.794497225885143e-05,
+      "loss": 0.6711,
+      "step": 2430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792662619296937e-05,
+      "loss": 0.7007,
+      "step": 2431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790827466096701e-05,
+      "loss": 0.6756,
+      "step": 2432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.788991766643634e-05,
+      "loss": 0.71,
+      "step": 2433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78715552129704e-05,
+      "loss": 0.6027,
+      "step": 2434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.785318730416328e-05,
+      "loss": 0.676,
+      "step": 2435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783481394361017e-05,
+      "loss": 0.7066,
+      "step": 2436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.781643513490733e-05,
+      "loss": 0.6428,
+      "step": 2437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.779805088165206e-05,
+      "loss": 0.7085,
+      "step": 2438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.777966118744272e-05,
+      "loss": 0.676,
+      "step": 2439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.776126605587877e-05,
+      "loss": 0.6387,
+      "step": 2440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.774286549056072e-05,
+      "loss": 0.7075,
+      "step": 2441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.772445949509015e-05,
+      "loss": 0.6283,
+      "step": 2442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.770604807306966e-05,
+      "loss": 0.4131,
+      "step": 2443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768763122810297e-05,
+      "loss": 0.7202,
+      "step": 2444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766920896379481e-05,
+      "loss": 0.5979,
+      "step": 2445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.765078128375105e-05,
+      "loss": 0.7,
+      "step": 2446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.763234819157852e-05,
+      "loss": 0.6697,
+      "step": 2447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761390969088516e-05,
+      "loss": 0.6727,
+      "step": 2448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759546578528e-05,
+      "loss": 0.6343,
+      "step": 2449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757701647837306e-05,
+      "loss": 0.685,
+      "step": 2450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.755856177377545e-05,
+      "loss": 0.6613,
+      "step": 2451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.754010167509935e-05,
+      "loss": 0.3816,
+      "step": 2452
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.752163618595798e-05,
+      "loss": 0.6042,
+      "step": 2453
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.750316530996563e-05,
+      "loss": 0.7069,
+      "step": 2454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748468905073762e-05,
+      "loss": 0.741,
+      "step": 2455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.746620741189032e-05,
+      "loss": 0.6645,
+      "step": 2456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74477203970412e-05,
+      "loss": 0.66,
+      "step": 2457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.742922800980871e-05,
+      "loss": 0.7125,
+      "step": 2458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.741073025381245e-05,
+      "loss": 0.6509,
+      "step": 2459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.739222713267296e-05,
+      "loss": 0.7138,
+      "step": 2460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73737186500119e-05,
+      "loss": 0.6765,
+      "step": 2461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.7355204809452e-05,
+      "loss": 0.6644,
+      "step": 2462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.733668561461694e-05,
+      "loss": 0.6936,
+      "step": 2463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.731816106913155e-05,
+      "loss": 0.7158,
+      "step": 2464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.729963117662166e-05,
+      "loss": 0.6899,
+      "step": 2465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.728109594071415e-05,
+      "loss": 0.6922,
+      "step": 2466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.726255536503694e-05,
+      "loss": 0.6859,
+      "step": 2467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.724400945321902e-05,
+      "loss": 0.6708,
+      "step": 2468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72254582088904e-05,
+      "loss": 0.6293,
+      "step": 2469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.720690163568214e-05,
+      "loss": 0.6994,
+      "step": 2470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.718833973722637e-05,
+      "loss": 0.7021,
+      "step": 2471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.716977251715621e-05,
+      "loss": 0.6948,
+      "step": 2472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.715119997910585e-05,
+      "loss": 0.6616,
+      "step": 2473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713262212671055e-05,
+      "loss": 0.6782,
+      "step": 2474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711403896360653e-05,
+      "loss": 0.6909,
+      "step": 2475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.709545049343117e-05,
+      "loss": 0.6594,
+      "step": 2476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.707685671982275e-05,
+      "loss": 0.6748,
+      "step": 2477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.705825764642071e-05,
+      "loss": 0.6207,
+      "step": 2478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.703965327686544e-05,
+      "loss": 0.6094,
+      "step": 2479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.702104361479841e-05,
+      "loss": 0.6789,
+      "step": 2480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.700242866386212e-05,
+      "loss": 0.6045,
+      "step": 2481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.698380842770011e-05,
+      "loss": 0.6463,
+      "step": 2482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.696518290995693e-05,
+      "loss": 0.6811,
+      "step": 2483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69465521142782e-05,
+      "loss": 0.6876,
+      "step": 2484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.692791604431053e-05,
+      "loss": 0.7248,
+      "step": 2485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.690927470370162e-05,
+      "loss": 0.6242,
+      "step": 2486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.689062809610013e-05,
+      "loss": 0.6371,
+      "step": 2487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68719762251558e-05,
+      "loss": 0.6183,
+      "step": 2488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.685331909451941e-05,
+      "loss": 0.7143,
+      "step": 2489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.683465670784272e-05,
+      "loss": 0.6624,
+      "step": 2490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681598906877854e-05,
+      "loss": 0.7364,
+      "step": 2491
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679731618098076e-05,
+      "loss": 0.7011,
+      "step": 2492
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.677863804810421e-05,
+      "loss": 0.6273,
+      "step": 2493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675995467380481e-05,
+      "loss": 0.666,
+      "step": 2494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.674126606173948e-05,
+      "loss": 0.6827,
+      "step": 2495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.672257221556616e-05,
+      "loss": 0.7078,
+      "step": 2496
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.670387313894384e-05,
+      "loss": 0.6751,
+      "step": 2497
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.668516883553251e-05,
+      "loss": 0.5988,
+      "step": 2498
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66664593089932e-05,
+      "loss": 0.3601,
+      "step": 2499
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.664774456298794e-05,
+      "loss": 0.7377,
+      "step": 2500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66290246011798e-05,
+      "loss": 0.6677,
+      "step": 2501
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.661029942723287e-05,
+      "loss": 0.6543,
+      "step": 2502
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.659156904481223e-05,
+      "loss": 0.6421,
+      "step": 2503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.657283345758407e-05,
+      "loss": 0.6233,
+      "step": 2504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.655409266921547e-05,
+      "loss": 0.7082,
+      "step": 2505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.653534668337463e-05,
+      "loss": 0.6647,
+      "step": 2506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65165955037307e-05,
+      "loss": 0.6628,
+      "step": 2507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64978391339539e-05,
+      "loss": 0.6834,
+      "step": 2508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.647907757771545e-05,
+      "loss": 0.7101,
+      "step": 2509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.646031083868755e-05,
+      "loss": 0.6249,
+      "step": 2510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644153892054344e-05,
+      "loss": 0.6609,
+      "step": 2511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.642276182695743e-05,
+      "loss": 0.6907,
+      "step": 2512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64039795616047e-05,
+      "loss": 0.6345,
+      "step": 2513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63851921281616e-05,
+      "loss": 0.6784,
+      "step": 2514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.636639953030541e-05,
+      "loss": 0.6746,
+      "step": 2515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634760177171442e-05,
+      "loss": 0.74,
+      "step": 2516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.632879885606794e-05,
+      "loss": 0.6825,
+      "step": 2517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630999078704631e-05,
+      "loss": 0.6253,
+      "step": 2518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.629117756833085e-05,
+      "loss": 0.3778,
+      "step": 2519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.627235920360392e-05,
+      "loss": 0.6367,
+      "step": 2520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.625353569654882e-05,
+      "loss": 0.6996,
+      "step": 2521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.623470705084995e-05,
+      "loss": 0.6559,
+      "step": 2522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621587327019264e-05,
+      "loss": 0.6616,
+      "step": 2523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.619703435826328e-05,
+      "loss": 0.651,
+      "step": 2524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.617819031874919e-05,
+      "loss": 0.6641,
+      "step": 2525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.615934115533883e-05,
+      "loss": 0.6772,
+      "step": 2526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.614048687172149e-05,
+      "loss": 0.6968,
+      "step": 2527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612162747158757e-05,
+      "loss": 0.6199,
+      "step": 2528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610276295862846e-05,
+      "loss": 0.6694,
+      "step": 2529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608389333653654e-05,
+      "loss": 0.6484,
+      "step": 2530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.606501860900519e-05,
+      "loss": 0.6495,
+      "step": 2531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.604613877972877e-05,
+      "loss": 0.6951,
+      "step": 2532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.602725385240268e-05,
+      "loss": 0.6663,
+      "step": 2533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.600836383072329e-05,
+      "loss": 0.6595,
+      "step": 2534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.598946871838795e-05,
+      "loss": 0.6546,
+      "step": 2535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.597056851909505e-05,
+      "loss": 0.7269,
+      "step": 2536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595166323654396e-05,
+      "loss": 0.3979,
+      "step": 2537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.593275287443504e-05,
+      "loss": 0.6622,
+      "step": 2538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.591383743646962e-05,
+      "loss": 0.6438,
+      "step": 2539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.589491692635006e-05,
+      "loss": 0.7059,
+      "step": 2540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.587599134777972e-05,
+      "loss": 0.6526,
+      "step": 2541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.585706070446288e-05,
+      "loss": 0.5808,
+      "step": 2542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.583812500010493e-05,
+      "loss": 0.7046,
+      "step": 2543
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.581918423841212e-05,
+      "loss": 0.6423,
+      "step": 2544
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.580023842309179e-05,
+      "loss": 0.6417,
+      "step": 2545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.578128755785222e-05,
+      "loss": 0.6821,
+      "step": 2546
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.576233164640271e-05,
+      "loss": 0.712,
+      "step": 2547
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.57433706924535e-05,
+      "loss": 0.6116,
+      "step": 2548
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.572440469971587e-05,
+      "loss": 0.6054,
+      "step": 2549
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.570543367190205e-05,
+      "loss": 0.548,
+      "step": 2550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568645761272527e-05,
+      "loss": 0.6618,
+      "step": 2551
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.566747652589975e-05,
+      "loss": 0.6058,
+      "step": 2552
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564849041514069e-05,
+      "loss": 0.6402,
+      "step": 2553
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.562949928416423e-05,
+      "loss": 0.6702,
+      "step": 2554
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.561050313668759e-05,
+      "loss": 0.6579,
+      "step": 2555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.559150197642886e-05,
+      "loss": 0.6532,
+      "step": 2556
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55724958071072e-05,
+      "loss": 0.6371,
+      "step": 2557
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.555348463244271e-05,
+      "loss": 0.642,
+      "step": 2558
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.553446845615649e-05,
+      "loss": 0.749,
+      "step": 2559
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.551544728197057e-05,
+      "loss": 0.6625,
+      "step": 2560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.549642111360801e-05,
+      "loss": 0.6503,
+      "step": 2561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.547738995479284e-05,
+      "loss": 0.6374,
+      "step": 2562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.545835380925005e-05,
+      "loss": 0.6339,
+      "step": 2563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.54393126807056e-05,
+      "loss": 0.6616,
+      "step": 2564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.542026657288645e-05,
+      "loss": 0.652,
+      "step": 2565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.540121548952053e-05,
+      "loss": 0.6543,
+      "step": 2566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.538215943433674e-05,
+      "loss": 0.3449,
+      "step": 2567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.536309841106493e-05,
+      "loss": 0.5676,
+      "step": 2568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.534403242343595e-05,
+      "loss": 0.6647,
+      "step": 2569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532496147518162e-05,
+      "loss": 0.6858,
+      "step": 2570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.530588557003473e-05,
+      "loss": 0.6637,
+      "step": 2571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.528680471172903e-05,
+      "loss": 0.6288,
+      "step": 2572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.526771890399922e-05,
+      "loss": 0.6731,
+      "step": 2573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.524862815058102e-05,
+      "loss": 0.7033,
+      "step": 2574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.522953245521111e-05,
+      "loss": 0.6109,
+      "step": 2575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.521043182162709e-05,
+      "loss": 0.6544,
+      "step": 2576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.519132625356756e-05,
+      "loss": 0.6776,
+      "step": 2577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.517221575477209e-05,
+      "loss": 0.6595,
+      "step": 2578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.515310032898117e-05,
+      "loss": 0.6888,
+      "step": 2579
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.513397997993637e-05,
+      "loss": 0.6845,
+      "step": 2580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.511485471138007e-05,
+      "loss": 0.6923,
+      "step": 2581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.509572452705571e-05,
+      "loss": 0.6874,
+      "step": 2582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.507658943070768e-05,
+      "loss": 0.6851,
+      "step": 2583
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505744942608131e-05,
+      "loss": 0.6673,
+      "step": 2584
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50383045169229e-05,
+      "loss": 0.6445,
+      "step": 2585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.501915470697974e-05,
+      "loss": 0.6437,
+      "step": 2586
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.6377,
+      "step": 2587
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.498084039973291e-05,
+      "loss": 0.6754,
+      "step": 2588
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.496167590992857e-05,
+      "loss": 0.6652,
+      "step": 2589
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49425065343381e-05,
+      "loss": 0.6844,
+      "step": 2590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492333227671354e-05,
+      "loss": 0.7201,
+      "step": 2591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49041531408079e-05,
+      "loss": 0.6396,
+      "step": 2592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488496913037512e-05,
+      "loss": 0.6662,
+      "step": 2593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.486578024917015e-05,
+      "loss": 0.6405,
+      "step": 2594
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.484658650094882e-05,
+      "loss": 0.6539,
+      "step": 2595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.482738788946799e-05,
+      "loss": 0.656,
+      "step": 2596
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.48081844184854e-05,
+      "loss": 0.6982,
+      "step": 2597
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.47889760917598e-05,
+      "loss": 0.6716,
+      "step": 2598
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.476976291305085e-05,
+      "loss": 0.693,
+      "step": 2599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.475054488611916e-05,
+      "loss": 0.6686,
+      "step": 2600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.473132201472634e-05,
+      "loss": 0.6168,
+      "step": 2601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.47120943026349e-05,
+      "loss": 0.7108,
+      "step": 2602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.469286175360827e-05,
+      "loss": 0.6743,
+      "step": 2603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.467362437141093e-05,
+      "loss": 0.6738,
+      "step": 2604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.465438215980819e-05,
+      "loss": 0.6564,
+      "step": 2605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.463513512256637e-05,
+      "loss": 0.6917,
+      "step": 2606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.461588326345275e-05,
+      "loss": 0.6197,
+      "step": 2607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.459662658623547e-05,
+      "loss": 0.6928,
+      "step": 2608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.457736509468372e-05,
+      "loss": 0.6226,
+      "step": 2609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.455809879256754e-05,
+      "loss": 0.7016,
+      "step": 2610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.453882768365799e-05,
+      "loss": 0.6885,
+      "step": 2611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4519551771727e-05,
+      "loss": 0.7075,
+      "step": 2612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.45002710605475e-05,
+      "loss": 0.671,
+      "step": 2613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.448098555389333e-05,
+      "loss": 0.6977,
+      "step": 2614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.446169525553923e-05,
+      "loss": 0.6567,
+      "step": 2615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.444240016926097e-05,
+      "loss": 0.6554,
+      "step": 2616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.442310029883518e-05,
+      "loss": 0.748,
+      "step": 2617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.440379564803946e-05,
+      "loss": 0.7087,
+      "step": 2618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.438448622065235e-05,
+      "loss": 0.6601,
+      "step": 2619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43651720204533e-05,
+      "loss": 0.6727,
+      "step": 2620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434585305122268e-05,
+      "loss": 0.6566,
+      "step": 2621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.432652931674187e-05,
+      "loss": 0.6773,
+      "step": 2622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43072008207931e-05,
+      "loss": 0.644,
+      "step": 2623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.428786756715958e-05,
+      "loss": 0.6611,
+      "step": 2624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.426852955962544e-05,
+      "loss": 0.6986,
+      "step": 2625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.424918680197574e-05,
+      "loss": 0.7148,
+      "step": 2626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.422983929799644e-05,
+      "loss": 0.7409,
+      "step": 2627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.421048705147448e-05,
+      "loss": 0.6726,
+      "step": 2628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.41911300661977e-05,
+      "loss": 0.7039,
+      "step": 2629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.417176834595485e-05,
+      "loss": 0.6762,
+      "step": 2630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415240189453567e-05,
+      "loss": 0.684,
+      "step": 2631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.413303071573077e-05,
+      "loss": 0.6745,
+      "step": 2632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411365481333165e-05,
+      "loss": 0.706,
+      "step": 2633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.409427419113086e-05,
+      "loss": 0.6521,
+      "step": 2634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407488885292176e-05,
+      "loss": 0.7497,
+      "step": 2635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405549880249869e-05,
+      "loss": 0.6491,
+      "step": 2636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.403610404365685e-05,
+      "loss": 0.6376,
+      "step": 2637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401670458019245e-05,
+      "loss": 0.6853,
+      "step": 2638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.399730041590257e-05,
+      "loss": 0.617,
+      "step": 2639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.397789155458518e-05,
+      "loss": 0.7123,
+      "step": 2640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.395847800003925e-05,
+      "loss": 0.6621,
+      "step": 2641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.39390597560646e-05,
+      "loss": 0.6457,
+      "step": 2642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.391963682646199e-05,
+      "loss": 0.6548,
+      "step": 2643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390020921503312e-05,
+      "loss": 0.6903,
+      "step": 2644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388077692558057e-05,
+      "loss": 0.6833,
+      "step": 2645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.386133996190784e-05,
+      "loss": 0.6092,
+      "step": 2646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.384189832781939e-05,
+      "loss": 0.6734,
+      "step": 2647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.382245202712054e-05,
+      "loss": 0.6835,
+      "step": 2648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.380300106361752e-05,
+      "loss": 0.6898,
+      "step": 2649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.378354544111755e-05,
+      "loss": 0.6586,
+      "step": 2650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.376408516342866e-05,
+      "loss": 0.5846,
+      "step": 2651
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374462023435983e-05,
+      "loss": 0.6831,
+      "step": 2652
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.372515065772102e-05,
+      "loss": 0.7363,
+      "step": 2653
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3705676437323e-05,
+      "loss": 0.637,
+      "step": 2654
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.368619757697748e-05,
+      "loss": 0.7464,
+      "step": 2655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.366671408049712e-05,
+      "loss": 0.6567,
+      "step": 2656
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.364722595169543e-05,
+      "loss": 0.6593,
+      "step": 2657
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.362773319438685e-05,
+      "loss": 0.6689,
+      "step": 2658
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.360823581238672e-05,
+      "loss": 0.649,
+      "step": 2659
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.358873380951132e-05,
+      "loss": 0.6836,
+      "step": 2660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356922718957778e-05,
+      "loss": 0.7169,
+      "step": 2661
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.354971595640417e-05,
+      "loss": 0.6908,
+      "step": 2662
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.353020011380946e-05,
+      "loss": 0.6286,
+      "step": 2663
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.35106796656135e-05,
+      "loss": 0.6371,
+      "step": 2664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.349115461563706e-05,
+      "loss": 0.6309,
+      "step": 2665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.34716249677018e-05,
+      "loss": 0.3852,
+      "step": 2666
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.345209072563032e-05,
+      "loss": 0.6625,
+      "step": 2667
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.343255189324605e-05,
+      "loss": 0.6567,
+      "step": 2668
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.341300847437338e-05,
+      "loss": 0.6563,
+      "step": 2669
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.339346047283755e-05,
+      "loss": 0.6332,
+      "step": 2670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.337390789246474e-05,
+      "loss": 0.596,
+      "step": 2671
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.335435073708199e-05,
+      "loss": 0.6941,
+      "step": 2672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.333478901051727e-05,
+      "loss": 0.6848,
+      "step": 2673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.33152227165994e-05,
+      "loss": 0.6637,
+      "step": 2674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.329565185915815e-05,
+      "loss": 0.6517,
+      "step": 2675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327607644202412e-05,
+      "loss": 0.6613,
+      "step": 2676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.325649646902887e-05,
+      "loss": 0.7082,
+      "step": 2677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.323691194400478e-05,
+      "loss": 0.6863,
+      "step": 2678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.321732287078519e-05,
+      "loss": 0.6963,
+      "step": 2679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.319772925320429e-05,
+      "loss": 0.638,
+      "step": 2680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.317813109509716e-05,
+      "loss": 0.6538,
+      "step": 2681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315852840029979e-05,
+      "loss": 0.6831,
+      "step": 2682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.313892117264903e-05,
+      "loss": 0.7134,
+      "step": 2683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.311930941598262e-05,
+      "loss": 0.6462,
+      "step": 2684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.309969313413925e-05,
+      "loss": 0.7077,
+      "step": 2685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30800723309584e-05,
+      "loss": 0.687,
+      "step": 2686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30604470102805e-05,
+      "loss": 0.6383,
+      "step": 2687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.304081717594682e-05,
+      "loss": 0.6898,
+      "step": 2688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.302118283179955e-05,
+      "loss": 0.6263,
+      "step": 2689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.300154398168176e-05,
+      "loss": 0.6939,
+      "step": 2690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.298190062943737e-05,
+      "loss": 0.6825,
+      "step": 2691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.296225277891122e-05,
+      "loss": 0.6636,
+      "step": 2692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.294260043394901e-05,
+      "loss": 0.6054,
+      "step": 2693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.292294359839733e-05,
+      "loss": 0.583,
+      "step": 2694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.290328227610362e-05,
+      "loss": 0.6385,
+      "step": 2695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.288361647091623e-05,
+      "loss": 0.668,
+      "step": 2696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.286394618668438e-05,
+      "loss": 0.6968,
+      "step": 2697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.284427142725817e-05,
+      "loss": 0.7026,
+      "step": 2698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.282459219648856e-05,
+      "loss": 0.6179,
+      "step": 2699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.28049084982274e-05,
+      "loss": 0.6564,
+      "step": 2700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27852203363274e-05,
+      "loss": 0.6104,
+      "step": 2701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.276552771464218e-05,
+      "loss": 0.6898,
+      "step": 2702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.274583063702618e-05,
+      "loss": 0.6195,
+      "step": 2703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.272612910733475e-05,
+      "loss": 0.6638,
+      "step": 2704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27064231294241e-05,
+      "loss": 0.3832,
+      "step": 2705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.268671270715129e-05,
+      "loss": 0.6382,
+      "step": 2706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.266699784437429e-05,
+      "loss": 0.6237,
+      "step": 2707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264727854495195e-05,
+      "loss": 0.6423,
+      "step": 2708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.26275548127439e-05,
+      "loss": 0.6388,
+      "step": 2709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.260782665161071e-05,
+      "loss": 0.6256,
+      "step": 2710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.258809406541384e-05,
+      "loss": 0.682,
+      "step": 2711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.256835705801553e-05,
+      "loss": 0.6834,
+      "step": 2712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.254861563327896e-05,
+      "loss": 0.6873,
+      "step": 2713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.252886979506816e-05,
+      "loss": 0.5911,
+      "step": 2714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.250911954724799e-05,
+      "loss": 0.6521,
+      "step": 2715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.248936489368419e-05,
+      "loss": 0.6548,
+      "step": 2716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.246960583824338e-05,
+      "loss": 0.5946,
+      "step": 2717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.244984238479305e-05,
+      "loss": 0.7005,
+      "step": 2718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.24300745372015e-05,
+      "loss": 0.6496,
+      "step": 2719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.241030229933791e-05,
+      "loss": 0.6824,
+      "step": 2720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.239052567507239e-05,
+      "loss": 0.7367,
+      "step": 2721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.237074466827579e-05,
+      "loss": 0.6083,
+      "step": 2722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.235095928281987e-05,
+      "loss": 0.6437,
+      "step": 2723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233116952257731e-05,
+      "loss": 0.6826,
+      "step": 2724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.231137539142154e-05,
+      "loss": 0.7176,
+      "step": 2725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.229157689322692e-05,
+      "loss": 0.6595,
+      "step": 2726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.227177403186863e-05,
+      "loss": 0.7166,
+      "step": 2727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.225196681122271e-05,
+      "loss": 0.6827,
+      "step": 2728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223215523516607e-05,
+      "loss": 0.6838,
+      "step": 2729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.221233930757645e-05,
+      "loss": 0.6909,
+      "step": 2730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.219251903233246e-05,
+      "loss": 0.6536,
+      "step": 2731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.217269441331351e-05,
+      "loss": 0.7084,
+      "step": 2732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.215286545439997e-05,
+      "loss": 0.6071,
+      "step": 2733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.213303215947296e-05,
+      "loss": 0.7054,
+      "step": 2734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.211319453241445e-05,
+      "loss": 0.6909,
+      "step": 2735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.209335257710734e-05,
+      "loss": 0.663,
+      "step": 2736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.207350629743532e-05,
+      "loss": 0.6799,
+      "step": 2737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.205365569728287e-05,
+      "loss": 0.6809,
+      "step": 2738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.203380078053545e-05,
+      "loss": 0.7454,
+      "step": 2739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.201394155107928e-05,
+      "loss": 0.7002,
+      "step": 2740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.199407801280141e-05,
+      "loss": 0.3842,
+      "step": 2741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197421016958975e-05,
+      "loss": 0.6557,
+      "step": 2742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19543380253331e-05,
+      "loss": 0.7269,
+      "step": 2743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.193446158392105e-05,
+      "loss": 0.6841,
+      "step": 2744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.191458084924404e-05,
+      "loss": 0.7117,
+      "step": 2745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189469582519338e-05,
+      "loss": 0.7032,
+      "step": 2746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.187480651566116e-05,
+      "loss": 0.5283,
+      "step": 2747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.185491292454035e-05,
+      "loss": 0.6816,
+      "step": 2748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.183501505572478e-05,
+      "loss": 0.7051,
+      "step": 2749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.181511291310907e-05,
+      "loss": 0.6608,
+      "step": 2750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.179520650058868e-05,
+      "loss": 0.683,
+      "step": 2751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.177529582205997e-05,
+      "loss": 0.6531,
+      "step": 2752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.175538088142004e-05,
+      "loss": 0.6865,
+      "step": 2753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.173546168256688e-05,
+      "loss": 0.6917,
+      "step": 2754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.171553822939934e-05,
+      "loss": 0.6973,
+      "step": 2755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169561052581702e-05,
+      "loss": 0.6992,
+      "step": 2756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167567857572042e-05,
+      "loss": 0.6619,
+      "step": 2757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.165574238301085e-05,
+      "loss": 0.6342,
+      "step": 2758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.163580195159046e-05,
+      "loss": 0.6515,
+      "step": 2759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.16158572853622e-05,
+      "loss": 0.6296,
+      "step": 2760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.159590838822989e-05,
+      "loss": 0.6865,
+      "step": 2761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.157595526409815e-05,
+      "loss": 0.6558,
+      "step": 2762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.155599791687242e-05,
+      "loss": 0.7028,
+      "step": 2763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1536036350459e-05,
+      "loss": 0.6979,
+      "step": 2764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1516070568765e-05,
+      "loss": 0.7188,
+      "step": 2765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.149610057569832e-05,
+      "loss": 0.7296,
+      "step": 2766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.147612637516775e-05,
+      "loss": 0.6573,
+      "step": 2767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.145614797108287e-05,
+      "loss": 0.6772,
+      "step": 2768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.143616536735407e-05,
+      "loss": 0.6233,
+      "step": 2769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.141617856789256e-05,
+      "loss": 0.6792,
+      "step": 2770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139618757661043e-05,
+      "loss": 0.6498,
+      "step": 2771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137619239742052e-05,
+      "loss": 0.6456,
+      "step": 2772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.135619303423649e-05,
+      "loss": 0.6708,
+      "step": 2773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.13361894909729e-05,
+      "loss": 0.3941,
+      "step": 2774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.131618177154505e-05,
+      "loss": 0.6788,
+      "step": 2775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129616987986905e-05,
+      "loss": 0.6944,
+      "step": 2776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12761538198619e-05,
+      "loss": 0.7074,
+      "step": 2777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.125613359544136e-05,
+      "loss": 0.6286,
+      "step": 2778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.123610921052599e-05,
+      "loss": 0.6263,
+      "step": 2779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.121608066903524e-05,
+      "loss": 0.6412,
+      "step": 2780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.119604797488931e-05,
+      "loss": 0.7317,
+      "step": 2781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.117601113200921e-05,
+      "loss": 0.669,
+      "step": 2782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11559701443168e-05,
+      "loss": 0.6893,
+      "step": 2783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.113592501573473e-05,
+      "loss": 0.6716,
+      "step": 2784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.111587575018648e-05,
+      "loss": 0.6519,
+      "step": 2785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.109582235159628e-05,
+      "loss": 0.6498,
+      "step": 2786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.107576482388925e-05,
+      "loss": 0.617,
+      "step": 2787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.105570317099125e-05,
+      "loss": 0.6972,
+      "step": 2788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.103563739682899e-05,
+      "loss": 0.7471,
+      "step": 2789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.101556750533e-05,
+      "loss": 0.6796,
+      "step": 2790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.099549350042255e-05,
+      "loss": 0.6539,
+      "step": 2791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.097541538603577e-05,
+      "loss": 0.6605,
+      "step": 2792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.095533316609958e-05,
+      "loss": 0.6656,
+      "step": 2793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.093524684454471e-05,
+      "loss": 0.6255,
+      "step": 2794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091515642530266e-05,
+      "loss": 0.6387,
+      "step": 2795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.08950619123058e-05,
+      "loss": 0.6306,
+      "step": 2796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.087496330948723e-05,
+      "loss": 0.7053,
+      "step": 2797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.085486062078088e-05,
+      "loss": 0.6888,
+      "step": 2798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.08347538501215e-05,
+      "loss": 0.6211,
+      "step": 2799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.081464300144459e-05,
+      "loss": 0.6587,
+      "step": 2800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.079452807868649e-05,
+      "loss": 0.6829,
+      "step": 2801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.077440908578431e-05,
+      "loss": 0.6542,
+      "step": 2802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.075428602667602e-05,
+      "loss": 0.7063,
+      "step": 2803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073415890530029e-05,
+      "loss": 0.6332,
+      "step": 2804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071402772559663e-05,
+      "loss": 0.6615,
+      "step": 2805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.069389249150536e-05,
+      "loss": 0.6718,
+      "step": 2806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.067375320696759e-05,
+      "loss": 0.3892,
+      "step": 2807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.065360987592517e-05,
+      "loss": 0.6741,
+      "step": 2808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063346250232085e-05,
+      "loss": 0.6874,
+      "step": 2809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061331109009805e-05,
+      "loss": 0.7299,
+      "step": 2810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.059315564320105e-05,
+      "loss": 0.6584,
+      "step": 2811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.057299616557493e-05,
+      "loss": 0.7242,
+      "step": 2812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.055283266116552e-05,
+      "loss": 0.6906,
+      "step": 2813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053266513391943e-05,
+      "loss": 0.7001,
+      "step": 2814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05124935877841e-05,
+      "loss": 0.6775,
+      "step": 2815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.049231802670775e-05,
+      "loss": 0.676,
+      "step": 2816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.047213845463935e-05,
+      "loss": 0.6781,
+      "step": 2817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.04519548755287e-05,
+      "loss": 0.673,
+      "step": 2818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.043176729332632e-05,
+      "loss": 0.6555,
+      "step": 2819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.04115757119836e-05,
+      "loss": 0.6639,
+      "step": 2820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.039138013545265e-05,
+      "loss": 0.6811,
+      "step": 2821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.037118056768641e-05,
+      "loss": 0.7101,
+      "step": 2822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035097701263852e-05,
+      "loss": 0.6573,
+      "step": 2823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.033076947426347e-05,
+      "loss": 0.6571,
+      "step": 2824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.031055795651653e-05,
+      "loss": 0.6025,
+      "step": 2825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.029034246335373e-05,
+      "loss": 0.6937,
+      "step": 2826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.027012299873185e-05,
+      "loss": 0.7266,
+      "step": 2827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02498995666085e-05,
+      "loss": 0.6762,
+      "step": 2828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.022967217094204e-05,
+      "loss": 0.6791,
+      "step": 2829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.020944081569158e-05,
+      "loss": 0.6226,
+      "step": 2830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.018920550481709e-05,
+      "loss": 0.6903,
+      "step": 2831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.016896624227922e-05,
+      "loss": 0.6769,
+      "step": 2832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.014872303203943e-05,
+      "loss": 0.3737,
+      "step": 2833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.012847587805996e-05,
+      "loss": 0.692,
+      "step": 2834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.010822478430383e-05,
+      "loss": 0.6643,
+      "step": 2835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.008796975473479e-05,
+      "loss": 0.3905,
+      "step": 2836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.00677107933174e-05,
+      "loss": 0.7112,
+      "step": 2837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.004744790401699e-05,
+      "loss": 0.6495,
+      "step": 2838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.002718109079964e-05,
+      "loss": 0.6777,
+      "step": 2839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.000691035763217e-05,
+      "loss": 0.6231,
+      "step": 2840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.998663570848224e-05,
+      "loss": 0.6991,
+      "step": 2841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.996635714731823e-05,
+      "loss": 0.6381,
+      "step": 2842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.994607467810927e-05,
+      "loss": 0.6193,
+      "step": 2843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.992578830482531e-05,
+      "loss": 0.6415,
+      "step": 2844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.990549803143702e-05,
+      "loss": 0.6758,
+      "step": 2845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.988520386191583e-05,
+      "loss": 0.6683,
+      "step": 2846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.986490580023398e-05,
+      "loss": 0.6426,
+      "step": 2847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.984460385036442e-05,
+      "loss": 0.6773,
+      "step": 2848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.982429801628087e-05,
+      "loss": 0.7022,
+      "step": 2849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.980398830195785e-05,
+      "loss": 0.3892,
+      "step": 2850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978367471137059e-05,
+      "loss": 0.6229,
+      "step": 2851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.97633572484951e-05,
+      "loss": 0.7303,
+      "step": 2852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.974303591730815e-05,
+      "loss": 0.6735,
+      "step": 2853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.972271072178727e-05,
+      "loss": 0.6537,
+      "step": 2854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.970238166591073e-05,
+      "loss": 0.6635,
+      "step": 2855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.968204875365759e-05,
+      "loss": 0.7196,
+      "step": 2856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.966171198900761e-05,
+      "loss": 0.6425,
+      "step": 2857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.964137137594136e-05,
+      "loss": 0.7215,
+      "step": 2858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.962102691844011e-05,
+      "loss": 0.6452,
+      "step": 2859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.960067862048596e-05,
+      "loss": 0.6268,
+      "step": 2860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.958032648606166e-05,
+      "loss": 0.6331,
+      "step": 2861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.955997051915079e-05,
+      "loss": 0.6824,
+      "step": 2862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.953961072373767e-05,
+      "loss": 0.6901,
+      "step": 2863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951924710380732e-05,
+      "loss": 0.678,
+      "step": 2864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.949887966334553e-05,
+      "loss": 0.718,
+      "step": 2865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.947850840633892e-05,
+      "loss": 0.6797,
+      "step": 2866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.945813333677472e-05,
+      "loss": 0.6791,
+      "step": 2867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.943775445864101e-05,
+      "loss": 0.7401,
+      "step": 2868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.941737177592653e-05,
+      "loss": 0.6034,
+      "step": 2869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.939698529262086e-05,
+      "loss": 0.6648,
+      "step": 2870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.937659501271428e-05,
+      "loss": 0.6581,
+      "step": 2871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.935620094019777e-05,
+      "loss": 0.618,
+      "step": 2872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933580307906313e-05,
+      "loss": 0.7045,
+      "step": 2873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.931540143330284e-05,
+      "loss": 0.634,
+      "step": 2874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.929499600691014e-05,
+      "loss": 0.6126,
+      "step": 2875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.927458680387905e-05,
+      "loss": 0.6481,
+      "step": 2876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.925417382820426e-05,
+      "loss": 0.6507,
+      "step": 2877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.923375708388124e-05,
+      "loss": 0.7079,
+      "step": 2878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921333657490621e-05,
+      "loss": 0.6615,
+      "step": 2879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919291230527608e-05,
+      "loss": 0.6418,
+      "step": 2880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.917248427898853e-05,
+      "loss": 0.643,
+      "step": 2881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.915205250004197e-05,
+      "loss": 0.7504,
+      "step": 2882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.913161697243557e-05,
+      "loss": 0.6503,
+      "step": 2883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.911117770016915e-05,
+      "loss": 0.6755,
+      "step": 2884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.909073468724337e-05,
+      "loss": 0.708,
+      "step": 2885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.907028793765954e-05,
+      "loss": 0.6648,
+      "step": 2886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.904983745541973e-05,
+      "loss": 0.6457,
+      "step": 2887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.902938324452678e-05,
+      "loss": 0.6956,
+      "step": 2888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.900892530898419e-05,
+      "loss": 0.6443,
+      "step": 2889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.898846365279625e-05,
+      "loss": 0.6485,
+      "step": 2890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.896799827996792e-05,
+      "loss": 0.6711,
+      "step": 2891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.894752919450494e-05,
+      "loss": 0.6579,
+      "step": 2892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.892705640041373e-05,
+      "loss": 0.617,
+      "step": 2893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.890657990170148e-05,
+      "loss": 0.6846,
+      "step": 2894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.888609970237609e-05,
+      "loss": 0.7261,
+      "step": 2895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.886561580644619e-05,
+      "loss": 0.7084,
+      "step": 2896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.884512821792107e-05,
+      "loss": 0.5809,
+      "step": 2897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.882463694081085e-05,
+      "loss": 0.6399,
+      "step": 2898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.88041419791263e-05,
+      "loss": 0.6012,
+      "step": 2899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.878364333687894e-05,
+      "loss": 0.7187,
+      "step": 2900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.876314101808098e-05,
+      "loss": 0.6705,
+      "step": 2901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.874263502674538e-05,
+      "loss": 0.6197,
+      "step": 2902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.87221253668858e-05,
+      "loss": 0.6966,
+      "step": 2903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.870161204251665e-05,
+      "loss": 0.6842,
+      "step": 2904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.868109505765302e-05,
+      "loss": 0.6713,
+      "step": 2905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.866057441631071e-05,
+      "loss": 0.6737,
+      "step": 2906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864005012250629e-05,
+      "loss": 0.7297,
+      "step": 2907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.861952218025699e-05,
+      "loss": 0.673,
+      "step": 2908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.859899059358076e-05,
+      "loss": 0.66,
+      "step": 2909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.857845536649635e-05,
+      "loss": 0.6713,
+      "step": 2910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.855791650302305e-05,
+      "loss": 0.6495,
+      "step": 2911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.853737400718103e-05,
+      "loss": 0.6502,
+      "step": 2912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.851682788299108e-05,
+      "loss": 0.6302,
+      "step": 2913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.849627813447474e-05,
+      "loss": 0.6475,
+      "step": 2914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.847572476565421e-05,
+      "loss": 0.6766,
+      "step": 2915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.845516778055245e-05,
+      "loss": 0.6663,
+      "step": 2916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843460718319313e-05,
+      "loss": 0.6724,
+      "step": 2917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.841404297760059e-05,
+      "loss": 0.6766,
+      "step": 2918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.839347516779988e-05,
+      "loss": 0.6848,
+      "step": 2919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.837290375781678e-05,
+      "loss": 0.6606,
+      "step": 2920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.835232875167778e-05,
+      "loss": 0.7058,
+      "step": 2921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.833175015341002e-05,
+      "loss": 0.6962,
+      "step": 2922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.83111679670414e-05,
+      "loss": 0.5985,
+      "step": 2923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82905821966005e-05,
+      "loss": 0.6758,
+      "step": 2924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82699928461166e-05,
+      "loss": 0.6309,
+      "step": 2925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82493999196197e-05,
+      "loss": 0.6857,
+      "step": 2926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.822880342114046e-05,
+      "loss": 0.7035,
+      "step": 2927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82082033547103e-05,
+      "loss": 0.7037,
+      "step": 2928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.818759972436125e-05,
+      "loss": 0.7506,
+      "step": 2929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.816699253412613e-05,
+      "loss": 0.6801,
+      "step": 2930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.814638178803842e-05,
+      "loss": 0.6885,
+      "step": 2931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.812576749013224e-05,
+      "loss": 0.6369,
+      "step": 2932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.810514964444252e-05,
+      "loss": 0.6591,
+      "step": 2933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80845282550048e-05,
+      "loss": 0.7019,
+      "step": 2934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.806390332585531e-05,
+      "loss": 0.7117,
+      "step": 2935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.804327486103104e-05,
+      "loss": 0.6258,
+      "step": 2936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.802264286456958e-05,
+      "loss": 0.703,
+      "step": 2937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.800200734050931e-05,
+      "loss": 0.6413,
+      "step": 2938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798136829288924e-05,
+      "loss": 0.611,
+      "step": 2939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.796072572574906e-05,
+      "loss": 0.6934,
+      "step": 2940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.794007964312919e-05,
+      "loss": 0.7109,
+      "step": 2941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.79194300490707e-05,
+      "loss": 0.6568,
+      "step": 2942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78987769476154e-05,
+      "loss": 0.6654,
+      "step": 2943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.787812034280572e-05,
+      "loss": 0.6343,
+      "step": 2944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.785746023868482e-05,
+      "loss": 0.5883,
+      "step": 2945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.783679663929654e-05,
+      "loss": 0.6326,
+      "step": 2946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781612954868538e-05,
+      "loss": 0.7134,
+      "step": 2947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.779545897089655e-05,
+      "loss": 0.6792,
+      "step": 2948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.777478490997594e-05,
+      "loss": 0.7163,
+      "step": 2949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.775410736997011e-05,
+      "loss": 0.621,
+      "step": 2950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77334263549263e-05,
+      "loss": 0.6815,
+      "step": 2951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.771274186889244e-05,
+      "loss": 0.6371,
+      "step": 2952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.769205391591713e-05,
+      "loss": 0.6277,
+      "step": 2953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.767136250004967e-05,
+      "loss": 0.4071,
+      "step": 2954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.765066762534e-05,
+      "loss": 0.3697,
+      "step": 2955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.762996929583878e-05,
+      "loss": 0.7175,
+      "step": 2956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.76092675155973e-05,
+      "loss": 0.6489,
+      "step": 2957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.758856228866759e-05,
+      "loss": 0.5929,
+      "step": 2958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.756785361910227e-05,
+      "loss": 0.6339,
+      "step": 2959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.75471415109547e-05,
+      "loss": 0.6567,
+      "step": 2960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.752642596827891e-05,
+      "loss": 0.6503,
+      "step": 2961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.750570699512955e-05,
+      "loss": 0.7045,
+      "step": 2962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.748498459556199e-05,
+      "loss": 0.6487,
+      "step": 2963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.746425877363226e-05,
+      "loss": 0.3949,
+      "step": 2964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.744352953339706e-05,
+      "loss": 0.622,
+      "step": 2965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742279687891375e-05,
+      "loss": 0.6854,
+      "step": 2966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.740206081424034e-05,
+      "loss": 0.585,
+      "step": 2967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.738132134343557e-05,
+      "loss": 0.687,
+      "step": 2968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.736057847055879e-05,
+      "loss": 0.6598,
+      "step": 2969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.733983219967002e-05,
+      "loss": 0.6947,
+      "step": 2970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.731908253482999e-05,
+      "loss": 0.7107,
+      "step": 2971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.729832948010004e-05,
+      "loss": 0.6904,
+      "step": 2972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.72775730395422e-05,
+      "loss": 0.6448,
+      "step": 2973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.725681321721916e-05,
+      "loss": 0.6606,
+      "step": 2974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.723605001719428e-05,
+      "loss": 0.6951,
+      "step": 2975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.721528344353154e-05,
+      "loss": 0.6233,
+      "step": 2976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.719451350029566e-05,
+      "loss": 0.6415,
+      "step": 2977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.717374019155195e-05,
+      "loss": 0.6358,
+      "step": 2978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.715296352136638e-05,
+      "loss": 0.6264,
+      "step": 2979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.713218349380565e-05,
+      "loss": 0.6395,
+      "step": 2980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7111400112937e-05,
+      "loss": 0.6741,
+      "step": 2981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.709061338282845e-05,
+      "loss": 0.6805,
+      "step": 2982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.706982330754858e-05,
+      "loss": 0.6089,
+      "step": 2983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.704902989116669e-05,
+      "loss": 0.6289,
+      "step": 2984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.702823313775268e-05,
+      "loss": 0.6716,
+      "step": 2985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.700743305137714e-05,
+      "loss": 0.6485,
+      "step": 2986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.698662963611131e-05,
+      "loss": 0.6642,
+      "step": 2987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696582289602708e-05,
+      "loss": 0.6593,
+      "step": 2988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.694501283519695e-05,
+      "loss": 0.4124,
+      "step": 2989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.692419945769415e-05,
+      "loss": 0.6203,
+      "step": 2990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.690338276759248e-05,
+      "loss": 0.6913,
+      "step": 2991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.688256276896643e-05,
+      "loss": 0.7332,
+      "step": 2992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.686173946589114e-05,
+      "loss": 0.6923,
+      "step": 2993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684091286244238e-05,
+      "loss": 0.6632,
+      "step": 2994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.682008296269655e-05,
+      "loss": 0.6195,
+      "step": 2995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.679924977073076e-05,
+      "loss": 0.6648,
+      "step": 2996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.677841329062269e-05,
+      "loss": 0.3866,
+      "step": 2997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.67575735264507e-05,
+      "loss": 0.6744,
+      "step": 2998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.67367304822938e-05,
+      "loss": 0.6431,
+      "step": 2999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.671588416223163e-05,
+      "loss": 0.618,
+      "step": 3000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669503457034446e-05,
+      "loss": 0.6395,
+      "step": 3001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667418171071319e-05,
+      "loss": 0.6811,
+      "step": 3002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.665332558741944e-05,
+      "loss": 0.6773,
+      "step": 3003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.663246620454537e-05,
+      "loss": 0.6785,
+      "step": 3004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.661160356617382e-05,
+      "loss": 0.6149,
+      "step": 3005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.659073767638827e-05,
+      "loss": 0.7866,
+      "step": 3006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.656986853927286e-05,
+      "loss": 0.6091,
+      "step": 3007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.654899615891226e-05,
+      "loss": 0.6187,
+      "step": 3008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.652812053939194e-05,
+      "loss": 0.6414,
+      "step": 3009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.650724168479789e-05,
+      "loss": 0.6904,
+      "step": 3010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.648635959921672e-05,
+      "loss": 0.656,
+      "step": 3011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.646547428673575e-05,
+      "loss": 0.6903,
+      "step": 3012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.644458575144288e-05,
+      "loss": 0.7231,
+      "step": 3013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.642369399742666e-05,
+      "loss": 0.6714,
+      "step": 3014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640279902877627e-05,
+      "loss": 0.7021,
+      "step": 3015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.63819008495815e-05,
+      "loss": 0.694,
+      "step": 3016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.636099946393278e-05,
+      "loss": 0.6983,
+      "step": 3017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.634009487592118e-05,
+      "loss": 0.6488,
+      "step": 3018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.63191870896384e-05,
+      "loss": 0.659,
+      "step": 3019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.629827610917674e-05,
+      "loss": 0.6459,
+      "step": 3020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.62773619386291e-05,
+      "loss": 0.6858,
+      "step": 3021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.62564445820891e-05,
+      "loss": 0.6899,
+      "step": 3022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.62355240436509e-05,
+      "loss": 0.6777,
+      "step": 3023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.62146003274093e-05,
+      "loss": 0.7242,
+      "step": 3024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.619367343745974e-05,
+      "loss": 0.6372,
+      "step": 3025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.617274337789828e-05,
+      "loss": 0.5971,
+      "step": 3026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.615181015282156e-05,
+      "loss": 0.6433,
+      "step": 3027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.61308737663269e-05,
+      "loss": 0.6595,
+      "step": 3028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610993422251223e-05,
+      "loss": 0.6307,
+      "step": 3029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.608899152547602e-05,
+      "loss": 0.6422,
+      "step": 3030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.606804567931746e-05,
+      "loss": 0.6113,
+      "step": 3031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.604709668813631e-05,
+      "loss": 0.7056,
+      "step": 3032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602614455603292e-05,
+      "loss": 0.631,
+      "step": 3033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.60051892871083e-05,
+      "loss": 0.662,
+      "step": 3034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.598423088546406e-05,
+      "loss": 0.6553,
+      "step": 3035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.596326935520241e-05,
+      "loss": 0.6952,
+      "step": 3036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59423047004262e-05,
+      "loss": 0.6783,
+      "step": 3037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.592133692523886e-05,
+      "loss": 0.6612,
+      "step": 3038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.590036603374441e-05,
+      "loss": 0.6556,
+      "step": 3039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.587939203004758e-05,
+      "loss": 0.6201,
+      "step": 3040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.585841491825361e-05,
+      "loss": 0.6504,
+      "step": 3041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583743470246837e-05,
+      "loss": 0.5883,
+      "step": 3042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.581645138679836e-05,
+      "loss": 0.7305,
+      "step": 3043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.579546497535068e-05,
+      "loss": 0.7111,
+      "step": 3044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.577447547223304e-05,
+      "loss": 0.6574,
+      "step": 3045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57534828815537e-05,
+      "loss": 0.5969,
+      "step": 3046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.573248720742164e-05,
+      "loss": 0.711,
+      "step": 3047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.571148845394633e-05,
+      "loss": 0.7166,
+      "step": 3048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.569048662523788e-05,
+      "loss": 0.6806,
+      "step": 3049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.566948172540705e-05,
+      "loss": 0.6994,
+      "step": 3050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.564847375856513e-05,
+      "loss": 0.683,
+      "step": 3051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.562746272882405e-05,
+      "loss": 0.659,
+      "step": 3052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.560644864029632e-05,
+      "loss": 0.6676,
+      "step": 3053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.558543149709507e-05,
+      "loss": 0.6527,
+      "step": 3054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.556441130333403e-05,
+      "loss": 0.6613,
+      "step": 3055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554338806312748e-05,
+      "loss": 0.5796,
+      "step": 3056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.552236178059036e-05,
+      "loss": 0.6856,
+      "step": 3057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.550133245983818e-05,
+      "loss": 0.6901,
+      "step": 3058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5480300104987e-05,
+      "loss": 0.3927,
+      "step": 3059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545926472015354e-05,
+      "loss": 0.66,
+      "step": 3060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.54382263094551e-05,
+      "loss": 0.3918,
+      "step": 3061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.541718487700954e-05,
+      "loss": 0.6697,
+      "step": 3062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.539614042693532e-05,
+      "loss": 0.627,
+      "step": 3063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.537509296335155e-05,
+      "loss": 0.6793,
+      "step": 3064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.535404249037782e-05,
+      "loss": 0.6828,
+      "step": 3065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.533298901213443e-05,
+      "loss": 0.7152,
+      "step": 3066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.531193253274217e-05,
+      "loss": 0.6833,
+      "step": 3067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.529087305632245e-05,
+      "loss": 0.6192,
+      "step": 3068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.52698105869973e-05,
+      "loss": 0.6908,
+      "step": 3069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.52487451288893e-05,
+      "loss": 0.659,
+      "step": 3070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.522767668612162e-05,
+      "loss": 0.6195,
+      "step": 3071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.520660526281801e-05,
+      "loss": 0.636,
+      "step": 3072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.518553086310285e-05,
+      "loss": 0.6469,
+      "step": 3073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5164453491101e-05,
+      "loss": 0.6532,
+      "step": 3074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5143373150938e-05,
+      "loss": 0.6741,
+      "step": 3075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.512228984673994e-05,
+      "loss": 0.7248,
+      "step": 3076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.51012035826335e-05,
+      "loss": 0.6652,
+      "step": 3077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.50801143627459e-05,
+      "loss": 0.6233,
+      "step": 3078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.505902219120497e-05,
+      "loss": 0.5845,
+      "step": 3079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.50379270721391e-05,
+      "loss": 0.609,
+      "step": 3080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.501682900967729e-05,
+      "loss": 0.7445,
+      "step": 3081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.499572800794911e-05,
+      "loss": 0.6355,
+      "step": 3082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.497462407108465e-05,
+      "loss": 0.5798,
+      "step": 3083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.495351720321464e-05,
+      "loss": 0.6598,
+      "step": 3084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.493240740847036e-05,
+      "loss": 0.3952,
+      "step": 3085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.491129469098365e-05,
+      "loss": 0.6559,
+      "step": 3086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.489017905488695e-05,
+      "loss": 0.688,
+      "step": 3087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.486906050431323e-05,
+      "loss": 0.6787,
+      "step": 3088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.484793904339607e-05,
+      "loss": 0.712,
+      "step": 3089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.48268146762696e-05,
+      "loss": 0.7392,
+      "step": 3090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.480568740706856e-05,
+      "loss": 0.6774,
+      "step": 3091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.478455723992817e-05,
+      "loss": 0.7426,
+      "step": 3092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.476342417898429e-05,
+      "loss": 0.6456,
+      "step": 3093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.474228822837332e-05,
+      "loss": 0.5946,
+      "step": 3094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472114939223225e-05,
+      "loss": 0.7177,
+      "step": 3095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.47000076746986e-05,
+      "loss": 0.6889,
+      "step": 3096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.467886307991047e-05,
+      "loss": 0.6133,
+      "step": 3097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.465771561200653e-05,
+      "loss": 0.7176,
+      "step": 3098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.463656527512599e-05,
+      "loss": 0.738,
+      "step": 3099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.461541207340866e-05,
+      "loss": 0.6251,
+      "step": 3100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.459425601099487e-05,
+      "loss": 0.649,
+      "step": 3101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.457309709202555e-05,
+      "loss": 0.6593,
+      "step": 3102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.455193532064212e-05,
+      "loss": 0.6619,
+      "step": 3103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.453077070098667e-05,
+      "loss": 0.7086,
+      "step": 3104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.450960323720173e-05,
+      "loss": 0.3848,
+      "step": 3105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.448843293343045e-05,
+      "loss": 0.6883,
+      "step": 3106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.446725979381655e-05,
+      "loss": 0.7225,
+      "step": 3107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.444608382250426e-05,
+      "loss": 0.7554,
+      "step": 3108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.442490502363838e-05,
+      "loss": 0.6781,
+      "step": 3109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.44037234013643e-05,
+      "loss": 0.6222,
+      "step": 3110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.438253895982789e-05,
+      "loss": 0.6736,
+      "step": 3111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.43613517031756e-05,
+      "loss": 0.7172,
+      "step": 3112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.434016163555452e-05,
+      "loss": 0.6494,
+      "step": 3113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.431896876111216e-05,
+      "loss": 0.7025,
+      "step": 3114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.429777308399661e-05,
+      "loss": 0.685,
+      "step": 3115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.427657460835659e-05,
+      "loss": 0.6855,
+      "step": 3116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.425537333834127e-05,
+      "loss": 0.6905,
+      "step": 3117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.42341692781004e-05,
+      "loss": 0.6551,
+      "step": 3118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.421296243178432e-05,
+      "loss": 0.5949,
+      "step": 3119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.419175280354384e-05,
+      "loss": 0.6623,
+      "step": 3120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.417054039753037e-05,
+      "loss": 0.6328,
+      "step": 3121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.414932521789585e-05,
+      "loss": 0.6918,
+      "step": 3122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.412810726879277e-05,
+      "loss": 0.6871,
+      "step": 3123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.410688655437412e-05,
+      "loss": 0.7081,
+      "step": 3124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.408566307879347e-05,
+      "loss": 0.6849,
+      "step": 3125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.406443684620494e-05,
+      "loss": 0.6539,
+      "step": 3126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.404320786076317e-05,
+      "loss": 0.638,
+      "step": 3127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.40219761266233e-05,
+      "loss": 0.633,
+      "step": 3128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.400074164794111e-05,
+      "loss": 0.619,
+      "step": 3129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.397950442887282e-05,
+      "loss": 0.6879,
+      "step": 3130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.395826447357524e-05,
+      "loss": 0.7071,
+      "step": 3131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.393702178620566e-05,
+      "loss": 0.6365,
+      "step": 3132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.391577637092199e-05,
+      "loss": 0.646,
+      "step": 3133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.389452823188263e-05,
+      "loss": 0.6992,
+      "step": 3134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.387327737324644e-05,
+      "loss": 0.6779,
+      "step": 3135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.385202379917297e-05,
+      "loss": 0.5934,
+      "step": 3136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.383076751382217e-05,
+      "loss": 0.6783,
+      "step": 3137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.380950852135454e-05,
+      "loss": 0.6295,
+      "step": 3138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.37882468259312e-05,
+      "loss": 0.617,
+      "step": 3139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.376698243171368e-05,
+      "loss": 0.7035,
+      "step": 3140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.374571534286411e-05,
+      "loss": 0.6821,
+      "step": 3141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.372444556354513e-05,
+      "loss": 0.7385,
+      "step": 3142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.37031730979199e-05,
+      "loss": 0.7077,
+      "step": 3143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.368189795015212e-05,
+      "loss": 0.7381,
+      "step": 3144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.366062012440599e-05,
+      "loss": 0.6287,
+      "step": 3145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.363933962484628e-05,
+      "loss": 0.6464,
+      "step": 3146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.361805645563822e-05,
+      "loss": 0.6365,
+      "step": 3147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.359677062094761e-05,
+      "loss": 0.5841,
+      "step": 3148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.357548212494077e-05,
+      "loss": 0.6098,
+      "step": 3149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.355419097178452e-05,
+      "loss": 0.6529,
+      "step": 3150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.35328971656462e-05,
+      "loss": 0.7233,
+      "step": 3151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.351160071069371e-05,
+      "loss": 0.588,
+      "step": 3152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.349030161109541e-05,
+      "loss": 0.641,
+      "step": 3153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.346899987102019e-05,
+      "loss": 0.698,
+      "step": 3154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.34476954946375e-05,
+      "loss": 0.6868,
+      "step": 3155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.342638848611726e-05,
+      "loss": 0.6648,
+      "step": 3156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.340507884962994e-05,
+      "loss": 0.7126,
+      "step": 3157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.33837665893465e-05,
+      "loss": 0.7225,
+      "step": 3158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.336245170943842e-05,
+      "loss": 0.6497,
+      "step": 3159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.334113421407768e-05,
+      "loss": 0.6776,
+      "step": 3160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.33198141074368e-05,
+      "loss": 0.7087,
+      "step": 3161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.329849139368879e-05,
+      "loss": 0.7145,
+      "step": 3162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.327716607700719e-05,
+      "loss": 0.3729,
+      "step": 3163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.325583816156602e-05,
+      "loss": 0.6027,
+      "step": 3164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.323450765153985e-05,
+      "loss": 0.6593,
+      "step": 3165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.321317455110368e-05,
+      "loss": 0.6757,
+      "step": 3166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.319183886443312e-05,
+      "loss": 0.6022,
+      "step": 3167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.317050059570422e-05,
+      "loss": 0.6949,
+      "step": 3168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.314915974909355e-05,
+      "loss": 0.6603,
+      "step": 3169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.312781632877818e-05,
+      "loss": 0.6782,
+      "step": 3170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31064703389357e-05,
+      "loss": 0.6431,
+      "step": 3171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.308512178374419e-05,
+      "loss": 0.6468,
+      "step": 3172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.306377066738222e-05,
+      "loss": 0.6265,
+      "step": 3173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.30424169940289e-05,
+      "loss": 0.7161,
+      "step": 3174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.302106076786381e-05,
+      "loss": 0.6333,
+      "step": 3175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.299970199306701e-05,
+      "loss": 0.6583,
+      "step": 3176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.297834067381913e-05,
+      "loss": 0.6815,
+      "step": 3177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.295697681430121e-05,
+      "loss": 0.6253,
+      "step": 3178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.293561041869487e-05,
+      "loss": 0.6993,
+      "step": 3179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.291424149118214e-05,
+      "loss": 0.6604,
+      "step": 3180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.289287003594564e-05,
+      "loss": 0.6618,
+      "step": 3181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.28714960571684e-05,
+      "loss": 0.6634,
+      "step": 3182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2850119559034e-05,
+      "loss": 0.7216,
+      "step": 3183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.28287405457265e-05,
+      "loss": 0.6813,
+      "step": 3184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.280735902143041e-05,
+      "loss": 0.6572,
+      "step": 3185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.278597499033081e-05,
+      "loss": 0.6915,
+      "step": 3186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.27645884566132e-05,
+      "loss": 0.6706,
+      "step": 3187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.274319942446363e-05,
+      "loss": 0.6674,
+      "step": 3188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.272180789806857e-05,
+      "loss": 0.6857,
+      "step": 3189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.270041388161503e-05,
+      "loss": 0.7123,
+      "step": 3190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.267901737929051e-05,
+      "loss": 0.6494,
+      "step": 3191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.265761839528296e-05,
+      "loss": 0.5965,
+      "step": 3192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263621693378085e-05,
+      "loss": 0.7164,
+      "step": 3193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26148129989731e-05,
+      "loss": 0.7028,
+      "step": 3194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.259340659504914e-05,
+      "loss": 0.7121,
+      "step": 3195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.257199772619888e-05,
+      "loss": 0.5683,
+      "step": 3196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.255058639661272e-05,
+      "loss": 0.6763,
+      "step": 3197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.25291726104815e-05,
+      "loss": 0.6551,
+      "step": 3198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.250775637199661e-05,
+      "loss": 0.6671,
+      "step": 3199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.248633768534988e-05,
+      "loss": 0.69,
+      "step": 3200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.246491655473362e-05,
+      "loss": 0.6373,
+      "step": 3201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.244349298434058e-05,
+      "loss": 0.6518,
+      "step": 3202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.242206697836406e-05,
+      "loss": 0.6772,
+      "step": 3203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.240063854099781e-05,
+      "loss": 0.6661,
+      "step": 3204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.237920767643603e-05,
+      "loss": 0.6268,
+      "step": 3205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.235777438887345e-05,
+      "loss": 0.682,
+      "step": 3206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.233633868250518e-05,
+      "loss": 0.7137,
+      "step": 3207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.231490056152692e-05,
+      "loss": 0.6521,
+      "step": 3208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.229346003013476e-05,
+      "loss": 0.5986,
+      "step": 3209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.22720170925253e-05,
+      "loss": 0.5906,
+      "step": 3210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.225057175289558e-05,
+      "loss": 0.6696,
+      "step": 3211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.222912401544314e-05,
+      "loss": 0.7073,
+      "step": 3212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.220767388436598e-05,
+      "loss": 0.6808,
+      "step": 3213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218622136386255e-05,
+      "loss": 0.6953,
+      "step": 3214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.216476645813181e-05,
+      "loss": 0.3685,
+      "step": 3215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.214330917137313e-05,
+      "loss": 0.654,
+      "step": 3216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.21218495077864e-05,
+      "loss": 0.6068,
+      "step": 3217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.210038747157194e-05,
+      "loss": 0.6517,
+      "step": 3218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.207892306693057e-05,
+      "loss": 0.6625,
+      "step": 3219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.205745629806351e-05,
+      "loss": 0.5889,
+      "step": 3220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.203598716917252e-05,
+      "loss": 0.7059,
+      "step": 3221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.201451568445977e-05,
+      "loss": 0.6258,
+      "step": 3222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.199304184812788e-05,
+      "loss": 0.6804,
+      "step": 3223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.197156566438001e-05,
+      "loss": 0.6423,
+      "step": 3224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.195008713741968e-05,
+      "loss": 0.6909,
+      "step": 3225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.192860627145094e-05,
+      "loss": 0.4012,
+      "step": 3226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.190712307067824e-05,
+      "loss": 0.7534,
+      "step": 3227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.188563753930657e-05,
+      "loss": 0.709,
+      "step": 3228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.186414968154128e-05,
+      "loss": 0.6453,
+      "step": 3229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.184265950158824e-05,
+      "loss": 0.7336,
+      "step": 3230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.182116700365376e-05,
+      "loss": 0.6782,
+      "step": 3231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.179967219194459e-05,
+      "loss": 0.6447,
+      "step": 3232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.177817507066793e-05,
+      "loss": 0.6938,
+      "step": 3233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.175667564403147e-05,
+      "loss": 0.6382,
+      "step": 3234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.173517391624331e-05,
+      "loss": 0.6397,
+      "step": 3235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.171366989151201e-05,
+      "loss": 0.6508,
+      "step": 3236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.169216357404659e-05,
+      "loss": 0.6381,
+      "step": 3237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.167065496805653e-05,
+      "loss": 0.719,
+      "step": 3238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.16491440777517e-05,
+      "loss": 0.6032,
+      "step": 3239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.162763090734252e-05,
+      "loss": 0.3781,
+      "step": 3240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.160611546103974e-05,
+      "loss": 0.5958,
+      "step": 3241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.158459774305462e-05,
+      "loss": 0.6458,
+      "step": 3242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.156307775759888e-05,
+      "loss": 0.654,
+      "step": 3243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.154155550888466e-05,
+      "loss": 0.5962,
+      "step": 3244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.152003100112451e-05,
+      "loss": 0.395,
+      "step": 3245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.149850423853148e-05,
+      "loss": 0.5995,
+      "step": 3246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.147697522531902e-05,
+      "loss": 0.655,
+      "step": 3247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.145544396570105e-05,
+      "loss": 0.6362,
+      "step": 3248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.143391046389188e-05,
+      "loss": 0.7295,
+      "step": 3249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.141237472410633e-05,
+      "loss": 0.649,
+      "step": 3250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.139083675055964e-05,
+      "loss": 0.7135,
+      "step": 3251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.136929654746744e-05,
+      "loss": 0.7339,
+      "step": 3252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.134775411904581e-05,
+      "loss": 0.6215,
+      "step": 3253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.132620946951132e-05,
+      "loss": 0.6603,
+      "step": 3254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.130466260308092e-05,
+      "loss": 0.5971,
+      "step": 3255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.128311352397199e-05,
+      "loss": 0.624,
+      "step": 3256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.126156223640241e-05,
+      "loss": 0.6264,
+      "step": 3257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.124000874459041e-05,
+      "loss": 0.6534,
+      "step": 3258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.121845305275468e-05,
+      "loss": 0.6064,
+      "step": 3259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.119689516511437e-05,
+      "loss": 0.6634,
+      "step": 3260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.117533508588905e-05,
+      "loss": 0.6016,
+      "step": 3261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.115377281929867e-05,
+      "loss": 0.6324,
+      "step": 3262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.113220836956367e-05,
+      "loss": 0.6874,
+      "step": 3263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.111064174090489e-05,
+      "loss": 0.676,
+      "step": 3264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.108907293754356e-05,
+      "loss": 0.6801,
+      "step": 3265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.106750196370144e-05,
+      "loss": 0.6353,
+      "step": 3266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.104592882360062e-05,
+      "loss": 0.661,
+      "step": 3267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.102435352146362e-05,
+      "loss": 0.7069,
+      "step": 3268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1002776061513436e-05,
+      "loss": 0.6556,
+      "step": 3269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0981196447973446e-05,
+      "loss": 0.6208,
+      "step": 3270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0959614685067444e-05,
+      "loss": 0.7216,
+      "step": 3271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.09380307770197e-05,
+      "loss": 0.6597,
+      "step": 3272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.091644472805482e-05,
+      "loss": 0.6424,
+      "step": 3273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0894856542397905e-05,
+      "loss": 0.6014,
+      "step": 3274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.087326622427443e-05,
+      "loss": 0.6873,
+      "step": 3275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.08516737779103e-05,
+      "loss": 0.6617,
+      "step": 3276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0830079207531856e-05,
+      "loss": 0.6289,
+      "step": 3277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0808482517365796e-05,
+      "loss": 0.6352,
+      "step": 3278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.078688371163931e-05,
+      "loss": 0.6389,
+      "step": 3279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0765282794579934e-05,
+      "loss": 0.6557,
+      "step": 3280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0743679770415675e-05,
+      "loss": 0.723,
+      "step": 3281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0722074643374915e-05,
+      "loss": 0.6465,
+      "step": 3282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.070046741768646e-05,
+      "loss": 0.6094,
+      "step": 3283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0678858097579515e-05,
+      "loss": 0.6405,
+      "step": 3284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.065724668728371e-05,
+      "loss": 0.6887,
+      "step": 3285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.063563319102909e-05,
+      "loss": 0.4073,
+      "step": 3286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.061401761304607e-05,
+      "loss": 0.657,
+      "step": 3287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0592399957565526e-05,
+      "loss": 0.3585,
+      "step": 3288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0570780228818705e-05,
+      "loss": 0.6215,
+      "step": 3289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.054915843103726e-05,
+      "loss": 0.5927,
+      "step": 3290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.052753456845327e-05,
+      "loss": 0.6906,
+      "step": 3291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.050590864529921e-05,
+      "loss": 0.6327,
+      "step": 3292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0484280665807936e-05,
+      "loss": 0.7021,
+      "step": 3293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.046265063421274e-05,
+      "loss": 0.6128,
+      "step": 3294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0441018554747283e-05,
+      "loss": 0.7019,
+      "step": 3295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0419384431645675e-05,
+      "loss": 0.6702,
+      "step": 3296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.039774826914235e-05,
+      "loss": 0.6639,
+      "step": 3297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0376110071472234e-05,
+      "loss": 0.591,
+      "step": 3298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.035446984287059e-05,
+      "loss": 0.6314,
+      "step": 3299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.033282758757306e-05,
+      "loss": 0.658,
+      "step": 3300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.031118330981576e-05,
+      "loss": 0.6923,
+      "step": 3301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.028953701383513e-05,
+      "loss": 0.6335,
+      "step": 3302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.026788870386803e-05,
+      "loss": 0.6366,
+      "step": 3303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.024623838415173e-05,
+      "loss": 0.6006,
+      "step": 3304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.022458605892388e-05,
+      "loss": 0.5896,
+      "step": 3305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.020293173242251e-05,
+      "loss": 0.6543,
+      "step": 3306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0181275408886064e-05,
+      "loss": 0.7259,
+      "step": 3307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0159617092553365e-05,
+      "loss": 0.6525,
+      "step": 3308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.013795678766361e-05,
+      "loss": 0.6884,
+      "step": 3309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0116294498456436e-05,
+      "loss": 0.695,
+      "step": 3310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.009463022917181e-05,
+      "loss": 0.6855,
+      "step": 3311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.007296398405011e-05,
+      "loss": 0.6376,
+      "step": 3312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0051295767332126e-05,
+      "loss": 0.5892,
+      "step": 3313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0029625583259e-05,
+      "loss": 0.6864,
+      "step": 3314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0007953436072236e-05,
+      "loss": 0.5983,
+      "step": 3315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.998627933001381e-05,
+      "loss": 0.6532,
+      "step": 3316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.996460326932599e-05,
+      "loss": 0.7024,
+      "step": 3317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.994292525825149e-05,
+      "loss": 0.6163,
+      "step": 3318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9921245301033356e-05,
+      "loss": 0.3862,
+      "step": 3319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.989956340191505e-05,
+      "loss": 0.6083,
+      "step": 3320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9877879565140403e-05,
+      "loss": 0.6279,
+      "step": 3321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.985619379495362e-05,
+      "loss": 0.6907,
+      "step": 3322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9834506095599284e-05,
+      "loss": 0.6928,
+      "step": 3323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9812816471322384e-05,
+      "loss": 0.7048,
+      "step": 3324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.979112492636824e-05,
+      "loss": 0.6975,
+      "step": 3325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9769431464982574e-05,
+      "loss": 0.6267,
+      "step": 3326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9747736091411486e-05,
+      "loss": 0.6791,
+      "step": 3327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.972603880990145e-05,
+      "loss": 0.6219,
+      "step": 3328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.970433962469929e-05,
+      "loss": 0.694,
+      "step": 3329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.968263854005223e-05,
+      "loss": 0.6524,
+      "step": 3330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.966093556020785e-05,
+      "loss": 0.6707,
+      "step": 3331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9639230689414114e-05,
+      "loss": 0.6999,
+      "step": 3332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.961752393191935e-05,
+      "loss": 0.6129,
+      "step": 3333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.959581529197225e-05,
+      "loss": 0.5915,
+      "step": 3334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.957410477382186e-05,
+      "loss": 0.6575,
+      "step": 3335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.955239238171765e-05,
+      "loss": 0.621,
+      "step": 3336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.95306781199094e-05,
+      "loss": 0.6253,
+      "step": 3337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.950896199264725e-05,
+      "loss": 0.6613,
+      "step": 3338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.948724400418178e-05,
+      "loss": 0.6721,
+      "step": 3339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.946552415876385e-05,
+      "loss": 0.6121,
+      "step": 3340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944380246064473e-05,
+      "loss": 0.6432,
+      "step": 3341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9422078914076026e-05,
+      "loss": 0.6938,
+      "step": 3342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.940035352330975e-05,
+      "loss": 0.5781,
+      "step": 3343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9378626292598206e-05,
+      "loss": 0.6583,
+      "step": 3344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.935689722619413e-05,
+      "loss": 0.373,
+      "step": 3345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9335166328350575e-05,
+      "loss": 0.6603,
+      "step": 3346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9313433603320945e-05,
+      "loss": 0.642,
+      "step": 3347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.929169905535905e-05,
+      "loss": 0.6281,
+      "step": 3348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9269962688719e-05,
+      "loss": 0.6842,
+      "step": 3349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9248224507655304e-05,
+      "loss": 0.6484,
+      "step": 3350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9226484516422783e-05,
+      "loss": 0.6867,
+      "step": 3351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9204742719276676e-05,
+      "loss": 0.6549,
+      "step": 3352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.91829991204725e-05,
+      "loss": 0.7034,
+      "step": 3353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9161253724266184e-05,
+      "loss": 0.6558,
+      "step": 3354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.913950653491398e-05,
+      "loss": 0.6883,
+      "step": 3355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.911775755667251e-05,
+      "loss": 0.6839,
+      "step": 3356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.909600679379871e-05,
+      "loss": 0.6212,
+      "step": 3357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.907425425054991e-05,
+      "loss": 0.6413,
+      "step": 3358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.905249993118376e-05,
+      "loss": 0.6864,
+      "step": 3359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.903074383995826e-05,
+      "loss": 0.6,
+      "step": 3360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.900898598113177e-05,
+      "loss": 0.6505,
+      "step": 3361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.898722635896299e-05,
+      "loss": 0.6294,
+      "step": 3362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.896546497771095e-05,
+      "loss": 0.6561,
+      "step": 3363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.894370184163507e-05,
+      "loss": 0.6436,
+      "step": 3364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8921936954995036e-05,
+      "loss": 0.665,
+      "step": 3365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.890017032205095e-05,
+      "loss": 0.7254,
+      "step": 3366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.887840194706323e-05,
+      "loss": 0.683,
+      "step": 3367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.885663183429263e-05,
+      "loss": 0.6707,
+      "step": 3368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8834859988000225e-05,
+      "loss": 0.6823,
+      "step": 3369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.881308641244747e-05,
+      "loss": 0.668,
+      "step": 3370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.879131111189615e-05,
+      "loss": 0.6387,
+      "step": 3371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.876953409060836e-05,
+      "loss": 0.6019,
+      "step": 3372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.874775535284653e-05,
+      "loss": 0.581,
+      "step": 3373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8725974902873495e-05,
+      "loss": 0.6603,
+      "step": 3374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.870419274495235e-05,
+      "loss": 0.7074,
+      "step": 3375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.868240888334653e-05,
+      "loss": 0.6639,
+      "step": 3376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.866062332231984e-05,
+      "loss": 0.6766,
+      "step": 3377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.863883606613642e-05,
+      "loss": 0.6709,
+      "step": 3378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.861704711906067e-05,
+      "loss": 0.6438,
+      "step": 3379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.859525648535742e-05,
+      "loss": 0.688,
+      "step": 3380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.857346416929178e-05,
+      "loss": 0.6265,
+      "step": 3381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.855167017512918e-05,
+      "loss": 0.6497,
+      "step": 3382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.852987450713538e-05,
+      "loss": 0.6913,
+      "step": 3383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.850807716957651e-05,
+      "loss": 0.645,
+      "step": 3384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.848627816671896e-05,
+      "loss": 0.7513,
+      "step": 3385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.846447750282951e-05,
+      "loss": 0.6647,
+      "step": 3386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.844267518217522e-05,
+      "loss": 0.6795,
+      "step": 3387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.842087120902351e-05,
+      "loss": 0.6777,
+      "step": 3388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.839906558764206e-05,
+      "loss": 0.6506,
+      "step": 3389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.837725832229898e-05,
+      "loss": 0.6508,
+      "step": 3390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8355449417262587e-05,
+      "loss": 0.6312,
+      "step": 3391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8333638876801585e-05,
+      "loss": 0.3773,
+      "step": 3392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8311826705185e-05,
+      "loss": 0.7309,
+      "step": 3393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8290012906682145e-05,
+      "loss": 0.5983,
+      "step": 3394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.826819748556266e-05,
+      "loss": 0.6373,
+      "step": 3395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.824638044609654e-05,
+      "loss": 0.6532,
+      "step": 3396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8224561792554036e-05,
+      "loss": 0.6495,
+      "step": 3397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.820274152920575e-05,
+      "loss": 0.6076,
+      "step": 3398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.818091966032262e-05,
+      "loss": 0.6504,
+      "step": 3399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.815909619017586e-05,
+      "loss": 0.7439,
+      "step": 3400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8137271123036974e-05,
+      "loss": 0.6449,
+      "step": 3401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811544446317787e-05,
+      "loss": 0.6102,
+      "step": 3402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8093616214870696e-05,
+      "loss": 0.7033,
+      "step": 3403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807178638238792e-05,
+      "loss": 0.6613,
+      "step": 3404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8049954970002306e-05,
+      "loss": 0.6039,
+      "step": 3405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802812198198699e-05,
+      "loss": 0.654,
+      "step": 3406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800628742261536e-05,
+      "loss": 0.5964,
+      "step": 3407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79844512961611e-05,
+      "loss": 0.6996,
+      "step": 3408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7962613606898275e-05,
+      "loss": 0.7564,
+      "step": 3409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794077435910119e-05,
+      "loss": 0.6725,
+      "step": 3410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.791893355704445e-05,
+      "loss": 0.6727,
+      "step": 3411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.789709120500302e-05,
+      "loss": 0.7016,
+      "step": 3412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.787524730725212e-05,
+      "loss": 0.6607,
+      "step": 3413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.785340186806729e-05,
+      "loss": 0.6282,
+      "step": 3414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.783155489172437e-05,
+      "loss": 0.7147,
+      "step": 3415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.780970638249952e-05,
+      "loss": 0.6873,
+      "step": 3416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.778785634466915e-05,
+      "loss": 0.652,
+      "step": 3417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.776600478251001e-05,
+      "loss": 0.6551,
+      "step": 3418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.774415170029917e-05,
+      "loss": 0.6736,
+      "step": 3419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.77222971023139e-05,
+      "loss": 0.5882,
+      "step": 3420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7700440992831905e-05,
+      "loss": 0.6284,
+      "step": 3421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.767858337613108e-05,
+      "loss": 0.6418,
+      "step": 3422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7656724256489646e-05,
+      "loss": 0.6707,
+      "step": 3423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.763486363818613e-05,
+      "loss": 0.6464,
+      "step": 3424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7613001525499344e-05,
+      "loss": 0.6541,
+      "step": 3425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7591137922708395e-05,
+      "loss": 0.6024,
+      "step": 3426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.756927283409266e-05,
+      "loss": 0.623,
+      "step": 3427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7547406263931835e-05,
+      "loss": 0.6284,
+      "step": 3428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.752553821650591e-05,
+      "loss": 0.6854,
+      "step": 3429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7503668696095134e-05,
+      "loss": 0.691,
+      "step": 3430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.748179770698007e-05,
+      "loss": 0.6925,
+      "step": 3431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.745992525344156e-05,
+      "loss": 0.6596,
+      "step": 3432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.743805133976071e-05,
+      "loss": 0.6138,
+      "step": 3433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7416175970218974e-05,
+      "loss": 0.4023,
+      "step": 3434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.739429914909803e-05,
+      "loss": 0.5744,
+      "step": 3435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.737242088067984e-05,
+      "loss": 0.6378,
+      "step": 3436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.735054116924671e-05,
+      "loss": 0.6035,
+      "step": 3437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.732866001908116e-05,
+      "loss": 0.6836,
+      "step": 3438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7306777434466016e-05,
+      "loss": 0.3773,
+      "step": 3439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.728489341968443e-05,
+      "loss": 0.6987,
+      "step": 3440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7263007979019745e-05,
+      "loss": 0.6868,
+      "step": 3441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7241121116755646e-05,
+      "loss": 0.6827,
+      "step": 3442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7219232837176096e-05,
+      "loss": 0.6685,
+      "step": 3443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.719734314456532e-05,
+      "loss": 0.6352,
+      "step": 3444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.71754520432078e-05,
+      "loss": 0.6476,
+      "step": 3445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.715355953738831e-05,
+      "loss": 0.6499,
+      "step": 3446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7131665631391937e-05,
+      "loss": 0.6622,
+      "step": 3447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7109770329504e-05,
+      "loss": 0.6402,
+      "step": 3448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7087873636010056e-05,
+      "loss": 0.6351,
+      "step": 3449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7065975555196036e-05,
+      "loss": 0.5968,
+      "step": 3450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.704407609134805e-05,
+      "loss": 0.7059,
+      "step": 3451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.702217524875252e-05,
+      "loss": 0.3758,
+      "step": 3452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.700027303169614e-05,
+      "loss": 0.5718,
+      "step": 3453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.697836944446584e-05,
+      "loss": 0.6452,
+      "step": 3454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.695646449134887e-05,
+      "loss": 0.684,
+      "step": 3455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.693455817663269e-05,
+      "loss": 0.7022,
+      "step": 3456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.691265050460509e-05,
+      "loss": 0.6793,
+      "step": 3457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6890741479554065e-05,
+      "loss": 0.6217,
+      "step": 3458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6868831105767905e-05,
+      "loss": 0.6803,
+      "step": 3459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.684691938753517e-05,
+      "loss": 0.7066,
+      "step": 3460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.682500632914467e-05,
+      "loss": 0.381,
+      "step": 3461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6803091934885463e-05,
+      "loss": 0.6324,
+      "step": 3462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6781176209046924e-05,
+      "loss": 0.6176,
+      "step": 3463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.675925915591862e-05,
+      "loss": 0.6739,
+      "step": 3464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.67373407797904e-05,
+      "loss": 0.6564,
+      "step": 3465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6715421084952424e-05,
+      "loss": 0.6801,
+      "step": 3466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.669350007569504e-05,
+      "loss": 0.6943,
+      "step": 3467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.667157775630886e-05,
+      "loss": 0.664,
+      "step": 3468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.664965413108481e-05,
+      "loss": 0.6102,
+      "step": 3469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.662772920431401e-05,
+      "loss": 0.6139,
+      "step": 3470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.660580298028785e-05,
+      "loss": 0.6635,
+      "step": 3471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.658387546329802e-05,
+      "loss": 0.6291,
+      "step": 3472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.656194665763639e-05,
+      "loss": 0.3745,
+      "step": 3473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.654001656759513e-05,
+      "loss": 0.6713,
+      "step": 3474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6518085197466666e-05,
+      "loss": 0.6945,
+      "step": 3475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.649615255154364e-05,
+      "loss": 0.5947,
+      "step": 3476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6474218634118956e-05,
+      "loss": 0.712,
+      "step": 3477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6452283449485774e-05,
+      "loss": 0.633,
+      "step": 3478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.643034700193752e-05,
+      "loss": 0.6627,
+      "step": 3479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.640840929576783e-05,
+      "loss": 0.6843,
+      "step": 3480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.63864703352706e-05,
+      "loss": 0.6713,
+      "step": 3481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.636453012474e-05,
+      "loss": 0.7054,
+      "step": 3482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.63425886684704e-05,
+      "loss": 0.6637,
+      "step": 3483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.632064597075643e-05,
+      "loss": 0.6656,
+      "step": 3484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.629870203589298e-05,
+      "loss": 0.6928,
+      "step": 3485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.627675686817517e-05,
+      "loss": 0.7331,
+      "step": 3486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.625481047189835e-05,
+      "loss": 0.6086,
+      "step": 3487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6232862851358125e-05,
+      "loss": 0.6361,
+      "step": 3488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6210914010850335e-05,
+      "loss": 0.6324,
+      "step": 3489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.618896395467106e-05,
+      "loss": 0.6557,
+      "step": 3490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.616701268711663e-05,
+      "loss": 0.3941,
+      "step": 3491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.614506021248358e-05,
+      "loss": 0.6691,
+      "step": 3492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.61231065350687e-05,
+      "loss": 0.6269,
+      "step": 3493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.610115165916904e-05,
+      "loss": 0.6252,
+      "step": 3494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6079195589081846e-05,
+      "loss": 0.6326,
+      "step": 3495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.60572383291046e-05,
+      "loss": 0.5855,
+      "step": 3496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6035279883535044e-05,
+      "loss": 0.7273,
+      "step": 3497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6013320256671145e-05,
+      "loss": 0.3457,
+      "step": 3498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.599135945281106e-05,
+      "loss": 0.666,
+      "step": 3499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.596939747625325e-05,
+      "loss": 0.6348,
+      "step": 3500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5947434331296356e-05,
+      "loss": 0.6731,
+      "step": 3501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.592547002223923e-05,
+      "loss": 0.6225,
+      "step": 3502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5903504553381006e-05,
+      "loss": 0.6642,
+      "step": 3503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.588153792902102e-05,
+      "loss": 0.6779,
+      "step": 3504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.585957015345882e-05,
+      "loss": 0.6078,
+      "step": 3505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.583760123099418e-05,
+      "loss": 0.6688,
+      "step": 3506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.581563116592714e-05,
+      "loss": 0.6311,
+      "step": 3507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.579365996255791e-05,
+      "loss": 0.6201,
+      "step": 3508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.577168762518695e-05,
+      "loss": 0.6739,
+      "step": 3509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.574971415811494e-05,
+      "loss": 0.6873,
+      "step": 3510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.57277395656428e-05,
+      "loss": 0.6944,
+      "step": 3511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5705763852071624e-05,
+      "loss": 0.622,
+      "step": 3512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.568378702170276e-05,
+      "loss": 0.6773,
+      "step": 3513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.566180907883777e-05,
+      "loss": 0.6433,
+      "step": 3514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.563983002777843e-05,
+      "loss": 0.6538,
+      "step": 3515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.561784987282671e-05,
+      "loss": 0.6388,
+      "step": 3516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.559586861828485e-05,
+      "loss": 0.6284,
+      "step": 3517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5573886268455256e-05,
+      "loss": 0.7263,
+      "step": 3518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5551902827640566e-05,
+      "loss": 0.6916,
+      "step": 3519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5529918300143646e-05,
+      "loss": 0.7325,
+      "step": 3520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.550793269026755e-05,
+      "loss": 0.6499,
+      "step": 3521
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.548594600231553e-05,
+      "loss": 0.6518,
+      "step": 3522
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.546395824059113e-05,
+      "loss": 0.6387,
+      "step": 3523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.544196940939801e-05,
+      "loss": 0.6428,
+      "step": 3524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.541997951304007e-05,
+      "loss": 0.7234,
+      "step": 3525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.539798855582144e-05,
+      "loss": 0.5686,
+      "step": 3526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.537599654204645e-05,
+      "loss": 0.6619,
+      "step": 3527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.535400347601961e-05,
+      "loss": 0.6694,
+      "step": 3528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.533200936204567e-05,
+      "loss": 0.6734,
+      "step": 3529
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5310014204429575e-05,
+      "loss": 0.6749,
+      "step": 3530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.528801800747645e-05,
+      "loss": 0.6876,
+      "step": 3531
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5266020775491654e-05,
+      "loss": 0.6635,
+      "step": 3532
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.524402251278076e-05,
+      "loss": 0.5932,
+      "step": 3533
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.522202322364948e-05,
+      "loss": 0.6957,
+      "step": 3534
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.520002291240379e-05,
+      "loss": 0.6313,
+      "step": 3535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5178021583349846e-05,
+      "loss": 0.7134,
+      "step": 3536
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5156019240794e-05,
+      "loss": 0.6716,
+      "step": 3537
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.513401588904279e-05,
+      "loss": 0.6079,
+      "step": 3538
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5112011532402986e-05,
+      "loss": 0.6671,
+      "step": 3539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509000617518152e-05,
+      "loss": 0.7169,
+      "step": 3540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.506799982168553e-05,
+      "loss": 0.6528,
+      "step": 3541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.504599247622237e-05,
+      "loss": 0.6642,
+      "step": 3542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.502398414309956e-05,
+      "loss": 0.6946,
+      "step": 3543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.50019748266248e-05,
+      "loss": 0.642,
+      "step": 3544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.497996453110606e-05,
+      "loss": 0.6694,
+      "step": 3545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.49579532608514e-05,
+      "loss": 0.6143,
+      "step": 3546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.493594102016914e-05,
+      "loss": 0.6308,
+      "step": 3547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.491392781336777e-05,
+      "loss": 0.6707,
+      "step": 3548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.489191364475597e-05,
+      "loss": 0.6466,
+      "step": 3549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.486989851864258e-05,
+      "loss": 0.6648,
+      "step": 3550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.484788243933668e-05,
+      "loss": 0.6616,
+      "step": 3551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.482586541114751e-05,
+      "loss": 0.6832,
+      "step": 3552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.480384743838449e-05,
+      "loss": 0.6206,
+      "step": 3553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.478182852535722e-05,
+      "loss": 0.6862,
+      "step": 3554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.475980867637551e-05,
+      "loss": 0.6558,
+      "step": 3555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4737787895749336e-05,
+      "loss": 0.6841,
+      "step": 3556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.471576618778884e-05,
+      "loss": 0.599,
+      "step": 3557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.46937435568044e-05,
+      "loss": 0.6728,
+      "step": 3558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4671720007106507e-05,
+      "loss": 0.6195,
+      "step": 3559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4649695543005866e-05,
+      "loss": 0.6509,
+      "step": 3560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.462767016881337e-05,
+      "loss": 0.6617,
+      "step": 3561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.460564388884008e-05,
+      "loss": 0.6264,
+      "step": 3562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.458361670739721e-05,
+      "loss": 0.6509,
+      "step": 3563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.456158862879621e-05,
+      "loss": 0.6417,
+      "step": 3564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.453955965734865e-05,
+      "loss": 0.6289,
+      "step": 3565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4517529797366276e-05,
+      "loss": 0.6565,
+      "step": 3566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4495499053161046e-05,
+      "loss": 0.618,
+      "step": 3567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.447346742904508e-05,
+      "loss": 0.4162,
+      "step": 3568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4451434929330626e-05,
+      "loss": 0.6834,
+      "step": 3569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.442940155833017e-05,
+      "loss": 0.5861,
+      "step": 3570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.440736732035633e-05,
+      "loss": 0.6553,
+      "step": 3571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4385332219721875e-05,
+      "loss": 0.6594,
+      "step": 3572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.436329626073982e-05,
+      "loss": 0.6758,
+      "step": 3573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.434125944772325e-05,
+      "loss": 0.6068,
+      "step": 3574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.431922178498548e-05,
+      "loss": 0.6323,
+      "step": 3575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4297183276839945e-05,
+      "loss": 0.6747,
+      "step": 3576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.427514392760034e-05,
+      "loss": 0.6779,
+      "step": 3577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4253103741580414e-05,
+      "loss": 0.6455,
+      "step": 3578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.423106272309413e-05,
+      "loss": 0.7327,
+      "step": 3579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4209020876455616e-05,
+      "loss": 0.6758,
+      "step": 3580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4186978205979156e-05,
+      "loss": 0.6069,
+      "step": 3581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.416493471597919e-05,
+      "loss": 0.6904,
+      "step": 3582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.414289041077033e-05,
+      "loss": 0.6763,
+      "step": 3583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4120845294667335e-05,
+      "loss": 0.6449,
+      "step": 3584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.409879937198513e-05,
+      "loss": 0.703,
+      "step": 3585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.40767526470388e-05,
+      "loss": 0.6112,
+      "step": 3586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.405470512414358e-05,
+      "loss": 0.6398,
+      "step": 3587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4032656807614854e-05,
+      "loss": 0.6675,
+      "step": 3588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.401060770176819e-05,
+      "loss": 0.661,
+      "step": 3589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3988557810919293e-05,
+      "loss": 0.6763,
+      "step": 3590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3966507139384006e-05,
+      "loss": 0.6675,
+      "step": 3591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.394445569147834e-05,
+      "loss": 0.5963,
+      "step": 3592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.392240347151848e-05,
+      "loss": 0.6424,
+      "step": 3593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.390035048382071e-05,
+      "loss": 0.676,
+      "step": 3594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3878296732701515e-05,
+      "loss": 0.6969,
+      "step": 3595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.38562422224775e-05,
+      "loss": 0.6403,
+      "step": 3596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3834186957465446e-05,
+      "loss": 0.6473,
+      "step": 3597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.381213094198223e-05,
+      "loss": 0.6892,
+      "step": 3598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.379007418034493e-05,
+      "loss": 0.6301,
+      "step": 3599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3768016676870745e-05,
+      "loss": 0.6825,
+      "step": 3600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.374595843587702e-05,
+      "loss": 0.6343,
+      "step": 3601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.372389946168126e-05,
+      "loss": 0.6088,
+      "step": 3602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.370183975860109e-05,
+      "loss": 0.6912,
+      "step": 3603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.367977933095428e-05,
+      "loss": 0.6267,
+      "step": 3604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.365771818305877e-05,
+      "loss": 0.6456,
+      "step": 3605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.363565631923262e-05,
+      "loss": 0.5864,
+      "step": 3606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.361359374379401e-05,
+      "loss": 0.7039,
+      "step": 3607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.359153046106129e-05,
+      "loss": 0.6375,
+      "step": 3608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.356946647535296e-05,
+      "loss": 0.7229,
+      "step": 3609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3547401790987614e-05,
+      "loss": 0.6604,
+      "step": 3610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.352533641228401e-05,
+      "loss": 0.5877,
+      "step": 3611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.350327034356104e-05,
+      "loss": 0.6665,
+      "step": 3612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.348120358913773e-05,
+      "loss": 0.6718,
+      "step": 3613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3459136153333234e-05,
+      "loss": 0.6188,
+      "step": 3614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.343706804046685e-05,
+      "loss": 0.6868,
+      "step": 3615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.341499925485801e-05,
+      "loss": 0.6449,
+      "step": 3616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3392929800826244e-05,
+      "loss": 0.7091,
+      "step": 3617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3370859682691265e-05,
+      "loss": 0.3884,
+      "step": 3618
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3348788904772887e-05,
+      "loss": 0.6488,
+      "step": 3619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.332671747139104e-05,
+      "loss": 0.6608,
+      "step": 3620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.330464538686581e-05,
+      "loss": 0.5912,
+      "step": 3621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3282572655517416e-05,
+      "loss": 0.6714,
+      "step": 3622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.326049928166614e-05,
+      "loss": 0.714,
+      "step": 3623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3238425269632485e-05,
+      "loss": 0.663,
+      "step": 3624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.321635062373701e-05,
+      "loss": 0.6379,
+      "step": 3625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3194275348300414e-05,
+      "loss": 0.7235,
+      "step": 3626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.317219944764353e-05,
+      "loss": 0.701,
+      "step": 3627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.315012292608731e-05,
+      "loss": 0.6331,
+      "step": 3628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.312804578795282e-05,
+      "loss": 0.6201,
+      "step": 3629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.310596803756125e-05,
+      "loss": 0.7249,
+      "step": 3630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.308388967923391e-05,
+      "loss": 0.6289,
+      "step": 3631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3061810717292235e-05,
+      "loss": 0.6338,
+      "step": 3632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.303973115605777e-05,
+      "loss": 0.6695,
+      "step": 3633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.301765099985219e-05,
+      "loss": 0.6627,
+      "step": 3634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2995570252997265e-05,
+      "loss": 0.6621,
+      "step": 3635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.297348891981489e-05,
+      "loss": 0.5892,
+      "step": 3636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2951407004627085e-05,
+      "loss": 0.6818,
+      "step": 3637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2929324511755976e-05,
+      "loss": 0.6557,
+      "step": 3638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.290724144552379e-05,
+      "loss": 0.6599,
+      "step": 3639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2885157810252915e-05,
+      "loss": 0.6123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.286307361026577e-05,
+      "loss": 0.6311,
+      "step": 3641
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.284098884988495e-05,
+      "loss": 0.6308,
+      "step": 3642
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.281890353343314e-05,
+      "loss": 0.6636,
+      "step": 3643
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.279681766523312e-05,
+      "loss": 0.6691,
+      "step": 3644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.277473124960781e-05,
+      "loss": 0.6499,
+      "step": 3645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2752644290880195e-05,
+      "loss": 0.6882,
+      "step": 3646
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.273055679337341e-05,
+      "loss": 0.6374,
+      "step": 3647
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2708468761410647e-05,
+      "loss": 0.6552,
+      "step": 3648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2686380199315244e-05,
+      "loss": 0.6617,
+      "step": 3649
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.266429111141065e-05,
+      "loss": 0.3743,
+      "step": 3650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.264220150202036e-05,
+      "loss": 0.6517,
+      "step": 3651
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2620111375468006e-05,
+      "loss": 0.3708,
+      "step": 3652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.259802073607736e-05,
+      "loss": 0.6793,
+      "step": 3653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.257592958817222e-05,
+      "loss": 0.675,
+      "step": 3654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255383793607652e-05,
+      "loss": 0.6333,
+      "step": 3655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.253174578411433e-05,
+      "loss": 0.6523,
+      "step": 3656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.250965313660974e-05,
+      "loss": 0.6129,
+      "step": 3657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.248755999788699e-05,
+      "loss": 0.6811,
+      "step": 3658
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2465466372270385e-05,
+      "loss": 0.6701,
+      "step": 3659
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.244337226408439e-05,
+      "loss": 0.7071,
+      "step": 3660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.242127767765348e-05,
+      "loss": 0.6413,
+      "step": 3661
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.239918261730226e-05,
+      "loss": 0.5761,
+      "step": 3662
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2377087087355444e-05,
+      "loss": 0.7021,
+      "step": 3663
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.235499109213783e-05,
+      "loss": 0.6506,
+      "step": 3664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.233289463597427e-05,
+      "loss": 0.367,
+      "step": 3665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.231079772318978e-05,
+      "loss": 0.5715,
+      "step": 3666
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2288700358109375e-05,
+      "loss": 0.6767,
+      "step": 3667
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.226660254505823e-05,
+      "loss": 0.6549,
+      "step": 3668
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2244504288361595e-05,
+      "loss": 0.6836,
+      "step": 3669
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.222240559234478e-05,
+      "loss": 0.7029,
+      "step": 3670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.220030646133318e-05,
+      "loss": 0.6728,
+      "step": 3671
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.217820689965231e-05,
+      "loss": 0.7182,
+      "step": 3672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2156106911627765e-05,
+      "loss": 0.622,
+      "step": 3673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.213400650158518e-05,
+      "loss": 0.7099,
+      "step": 3674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2111905673850304e-05,
+      "loss": 0.7174,
+      "step": 3675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.208980443274899e-05,
+      "loss": 0.6714,
+      "step": 3676
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.206770278260711e-05,
+      "loss": 0.6151,
+      "step": 3677
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2045600727750686e-05,
+      "loss": 0.6595,
+      "step": 3678
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2023498272505775e-05,
+      "loss": 0.6591,
+      "step": 3679
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.200139542119852e-05,
+      "loss": 0.5929,
+      "step": 3680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.197929217815515e-05,
+      "loss": 0.6381,
+      "step": 3681
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1957188547701965e-05,
+      "loss": 0.6317,
+      "step": 3682
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1935084534165336e-05,
+      "loss": 0.6881,
+      "step": 3683
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1912980141871704e-05,
+      "loss": 0.6739,
+      "step": 3684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.189087537514763e-05,
+      "loss": 0.6925,
+      "step": 3685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1868770238319684e-05,
+      "loss": 0.6376,
+      "step": 3686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1846664735714543e-05,
+      "loss": 0.703,
+      "step": 3687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1824558871658957e-05,
+      "loss": 0.6638,
+      "step": 3688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.180245265047974e-05,
+      "loss": 0.6677,
+      "step": 3689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.178034607650376e-05,
+      "loss": 0.6444,
+      "step": 3690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.175823915405799e-05,
+      "loss": 0.6253,
+      "step": 3691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.173613188746944e-05,
+      "loss": 0.6183,
+      "step": 3692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.171402428106519e-05,
+      "loss": 0.609,
+      "step": 3693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.169191633917242e-05,
+      "loss": 0.3802,
+      "step": 3694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.166980806611833e-05,
+      "loss": 0.6811,
+      "step": 3695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1647699466230205e-05,
+      "loss": 0.6342,
+      "step": 3696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.162559054383541e-05,
+      "loss": 0.633,
+      "step": 3697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1603481303261346e-05,
+      "loss": 0.5939,
+      "step": 3698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1581371748835484e-05,
+      "loss": 0.6233,
+      "step": 3699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1559261884885365e-05,
+      "loss": 0.662,
+      "step": 3700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1537151715738596e-05,
+      "loss": 0.6342,
+      "step": 3701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.151504124572282e-05,
+      "loss": 0.615,
+      "step": 3702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.149293047916576e-05,
+      "loss": 0.6479,
+      "step": 3703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.147081942039518e-05,
+      "loss": 0.6898,
+      "step": 3704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.144870807373892e-05,
+      "loss": 0.6803,
+      "step": 3705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.142659644352483e-05,
+      "loss": 0.7162,
+      "step": 3706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.140448453408092e-05,
+      "loss": 0.6506,
+      "step": 3707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1382372349735155e-05,
+      "loss": 0.6565,
+      "step": 3708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1360259894815565e-05,
+      "loss": 0.6323,
+      "step": 3709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1338147173650297e-05,
+      "loss": 0.684,
+      "step": 3710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1316034190567466e-05,
+      "loss": 0.6848,
+      "step": 3711
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.12939209498953e-05,
+      "loss": 0.6395,
+      "step": 3712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.127180745596205e-05,
+      "loss": 0.6076,
+      "step": 3713
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1249693713096024e-05,
+      "loss": 0.6773,
+      "step": 3714
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.122757972562559e-05,
+      "loss": 0.6485,
+      "step": 3715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.120546549787912e-05,
+      "loss": 0.6359,
+      "step": 3716
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.11833510341851e-05,
+      "loss": 0.6338,
+      "step": 3717
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116123633887202e-05,
+      "loss": 0.6491,
+      "step": 3718
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1139121416268396e-05,
+      "loss": 0.7017,
+      "step": 3719
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.111700627070285e-05,
+      "loss": 0.6292,
+      "step": 3720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1094890906504e-05,
+      "loss": 0.7004,
+      "step": 3721
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.10727753280005e-05,
+      "loss": 0.699,
+      "step": 3722
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1050659539521095e-05,
+      "loss": 0.6263,
+      "step": 3723
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1028543545394525e-05,
+      "loss": 0.6303,
+      "step": 3724
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.100642734994958e-05,
+      "loss": 0.6665,
+      "step": 3725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0984310957515114e-05,
+      "loss": 0.6339,
+      "step": 3726
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.096219437241999e-05,
+      "loss": 0.6651,
+      "step": 3727
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.094007759899312e-05,
+      "loss": 0.6985,
+      "step": 3728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.091796064156347e-05,
+      "loss": 0.6101,
+      "step": 3729
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0895843504460005e-05,
+      "loss": 0.6539,
+      "step": 3730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.087372619201174e-05,
+      "loss": 0.6122,
+      "step": 3731
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085160870854775e-05,
+      "loss": 0.6585,
+      "step": 3732
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0829491058397115e-05,
+      "loss": 0.6686,
+      "step": 3733
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.080737324588896e-05,
+      "loss": 0.6692,
+      "step": 3734
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.078525527535242e-05,
+      "loss": 0.6186,
+      "step": 3735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0763137151116705e-05,
+      "loss": 0.6363,
+      "step": 3736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0741018877511005e-05,
+      "loss": 0.6619,
+      "step": 3737
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.071890045886457e-05,
+      "loss": 0.7014,
+      "step": 3738
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0696781899506686e-05,
+      "loss": 0.6056,
+      "step": 3739
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.067466320376664e-05,
+      "loss": 0.6925,
+      "step": 3740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.065254437597373e-05,
+      "loss": 0.6236,
+      "step": 3741
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.063042542045736e-05,
+      "loss": 0.6968,
+      "step": 3742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.060830634154688e-05,
+      "loss": 0.612,
+      "step": 3743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0586187143571676e-05,
+      "loss": 0.6576,
+      "step": 3744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.056406783086119e-05,
+      "loss": 0.6044,
+      "step": 3745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0541948407744865e-05,
+      "loss": 0.6262,
+      "step": 3746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.051982887855217e-05,
+      "loss": 0.6735,
+      "step": 3747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.049770924761259e-05,
+      "loss": 0.7,
+      "step": 3748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.047558951925564e-05,
+      "loss": 0.6441,
+      "step": 3749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.045346969781083e-05,
+      "loss": 0.6434,
+      "step": 3750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.043134978760773e-05,
+      "loss": 0.7245,
+      "step": 3751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.04092297929759e-05,
+      "loss": 0.6673,
+      "step": 3752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.038710971824488e-05,
+      "loss": 0.6959,
+      "step": 3753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0364989567744324e-05,
+      "loss": 0.6323,
+      "step": 3754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0342869345803814e-05,
+      "loss": 0.7387,
+      "step": 3755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.032074905675297e-05,
+      "loss": 0.5892,
+      "step": 3756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.029862870492142e-05,
+      "loss": 0.6974,
+      "step": 3757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.027650829463883e-05,
+      "loss": 0.7349,
+      "step": 3758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0254387830234884e-05,
+      "loss": 0.6485,
+      "step": 3759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0232267316039194e-05,
+      "loss": 0.6384,
+      "step": 3760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.021014675638149e-05,
+      "loss": 0.6846,
+      "step": 3761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.018802615559146e-05,
+      "loss": 0.6469,
+      "step": 3762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.016590551799877e-05,
+      "loss": 0.388,
+      "step": 3763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.014378484793315e-05,
+      "loss": 0.6474,
+      "step": 3764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0121664149724315e-05,
+      "loss": 0.7304,
+      "step": 3765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0099543427701956e-05,
+      "loss": 0.4065,
+      "step": 3766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.007742268619583e-05,
+      "loss": 0.6283,
+      "step": 3767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0055301929535635e-05,
+      "loss": 0.6623,
+      "step": 3768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.003318116205111e-05,
+      "loss": 0.6479,
+      "step": 3769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0011060388072004e-05,
+      "loss": 0.7029,
+      "step": 3770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.998893961192801e-05,
+      "loss": 0.6211,
+      "step": 3771
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.99668188379489e-05,
+      "loss": 0.6998,
+      "step": 3772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.994469807046437e-05,
+      "loss": 0.3708,
+      "step": 3773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.992257731380419e-05,
+      "loss": 0.6398,
+      "step": 3774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9900456572298055e-05,
+      "loss": 0.6597,
+      "step": 3775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.987833585027571e-05,
+      "loss": 0.6254,
+      "step": 3776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.985621515206686e-05,
+      "loss": 0.6175,
+      "step": 3777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.983409448200126e-05,
+      "loss": 0.6913,
+      "step": 3778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.981197384440856e-05,
+      "loss": 0.6769,
+      "step": 3779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.978985324361853e-05,
+      "loss": 0.6031,
+      "step": 3780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.976773268396082e-05,
+      "loss": 0.6649,
+      "step": 3781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974561216976514e-05,
+      "loss": 0.6742,
+      "step": 3782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.972349170536117e-05,
+      "loss": 0.6734,
+      "step": 3783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9701371295078603e-05,
+      "loss": 0.6623,
+      "step": 3784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.967925094324704e-05,
+      "loss": 0.5818,
+      "step": 3785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.965713065419621e-05,
+      "loss": 0.6646,
+      "step": 3786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.963501043225569e-05,
+      "loss": 0.6328,
+      "step": 3787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.961289028175513e-05,
+      "loss": 0.6634,
+      "step": 3788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.959077020702411e-05,
+      "loss": 0.6076,
+      "step": 3789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.956865021239229e-05,
+      "loss": 0.6143,
+      "step": 3790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9546530302189175e-05,
+      "loss": 0.685,
+      "step": 3791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.952441048074438e-05,
+      "loss": 0.671,
+      "step": 3792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.950229075238742e-05,
+      "loss": 0.668,
+      "step": 3793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.948017112144784e-05,
+      "loss": 0.6542,
+      "step": 3794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.945805159225514e-05,
+      "loss": 0.6976,
+      "step": 3795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.943593216913883e-05,
+      "loss": 0.5978,
+      "step": 3796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.941381285642834e-05,
+      "loss": 0.6693,
+      "step": 3797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.939169365845315e-05,
+      "loss": 0.6447,
+      "step": 3798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.936957457954266e-05,
+      "loss": 0.6557,
+      "step": 3799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.934745562402628e-05,
+      "loss": 0.6622,
+      "step": 3800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.932533679623338e-05,
+      "loss": 0.614,
+      "step": 3801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930321810049334e-05,
+      "loss": 0.7104,
+      "step": 3802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928109954113544e-05,
+      "loss": 0.6655,
+      "step": 3803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9258981122489e-05,
+      "loss": 0.6797,
+      "step": 3804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9236862848883307e-05,
+      "loss": 0.6321,
+      "step": 3805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9214744724647585e-05,
+      "loss": 0.5975,
+      "step": 3806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9192626754111046e-05,
+      "loss": 0.6207,
+      "step": 3807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917050894160288e-05,
+      "loss": 0.6995,
+      "step": 3808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.914839129145226e-05,
+      "loss": 0.6297,
+      "step": 3809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.912627380798826e-05,
+      "loss": 0.5911,
+      "step": 3810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.910415649554001e-05,
+      "loss": 0.6477,
+      "step": 3811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9082039358436536e-05,
+      "loss": 0.6827,
+      "step": 3812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.905992240100689e-05,
+      "loss": 0.6854,
+      "step": 3813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9037805627580006e-05,
+      "loss": 0.6476,
+      "step": 3814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.90156890424849e-05,
+      "loss": 0.689,
+      "step": 3815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.899357265005042e-05,
+      "loss": 0.4018,
+      "step": 3816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8971456454605486e-05,
+      "loss": 0.6638,
+      "step": 3817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.894934046047891e-05,
+      "loss": 0.6434,
+      "step": 3818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.892722467199951e-05,
+      "loss": 0.6437,
+      "step": 3819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.890510909349602e-05,
+      "loss": 0.6149,
+      "step": 3820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8882993729297164e-05,
+      "loss": 0.6641,
+      "step": 3821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.88608785837316e-05,
+      "loss": 0.6353,
+      "step": 3822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.883876366112799e-05,
+      "loss": 0.6386,
+      "step": 3823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.881664896581489e-05,
+      "loss": 0.6711,
+      "step": 3824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.879453450212089e-05,
+      "loss": 0.6464,
+      "step": 3825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8772420274374425e-05,
+      "loss": 0.6684,
+      "step": 3826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.875030628690399e-05,
+      "loss": 0.6897,
+      "step": 3827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8728192544037956e-05,
+      "loss": 0.6612,
+      "step": 3828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.870607905010471e-05,
+      "loss": 0.5903,
+      "step": 3829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.868396580943253e-05,
+      "loss": 0.6583,
+      "step": 3830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.866185282634972e-05,
+      "loss": 0.6343,
+      "step": 3831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.863974010518443e-05,
+      "loss": 0.6384,
+      "step": 3832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.861762765026485e-05,
+      "loss": 0.6165,
+      "step": 3833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.859551546591907e-05,
+      "loss": 0.7072,
+      "step": 3834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8573403556475174e-05,
+      "loss": 0.6445,
+      "step": 3835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8551291926261095e-05,
+      "loss": 0.6878,
+      "step": 3836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.852918057960485e-05,
+      "loss": 0.6623,
+      "step": 3837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.850706952083426e-05,
+      "loss": 0.5888,
+      "step": 3838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.84849587542772e-05,
+      "loss": 0.3921,
+      "step": 3839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.846284828426141e-05,
+      "loss": 0.6805,
+      "step": 3840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.844073811511465e-05,
+      "loss": 0.6597,
+      "step": 3841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.841862825116453e-05,
+      "loss": 0.6243,
+      "step": 3842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.839651869673867e-05,
+      "loss": 0.5899,
+      "step": 3843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.83744094561646e-05,
+      "loss": 0.6213,
+      "step": 3844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.835230053376981e-05,
+      "loss": 0.725,
+      "step": 3845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.833019193388168e-05,
+      "loss": 0.7206,
+      "step": 3846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.83080836608276e-05,
+      "loss": 0.6171,
+      "step": 3847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.828597571893482e-05,
+      "loss": 0.6584,
+      "step": 3848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8263868112530587e-05,
+      "loss": 0.6463,
+      "step": 3849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.824176084594202e-05,
+      "loss": 0.6407,
+      "step": 3850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.821965392349626e-05,
+      "loss": 0.6249,
+      "step": 3851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8197547349520274e-05,
+      "loss": 0.7284,
+      "step": 3852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.817544112834106e-05,
+      "loss": 0.6298,
+      "step": 3853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.815333526428547e-05,
+      "loss": 0.6512,
+      "step": 3854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8131229761680334e-05,
+      "loss": 0.6636,
+      "step": 3855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8109124624852386e-05,
+      "loss": 0.6546,
+      "step": 3856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.808701985812832e-05,
+      "loss": 0.6545,
+      "step": 3857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.806491546583468e-05,
+      "loss": 0.6063,
+      "step": 3858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.804281145229807e-05,
+      "loss": 0.6433,
+      "step": 3859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.802070782184487e-05,
+      "loss": 0.5962,
+      "step": 3860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7998604578801495e-05,
+      "loss": 0.6583,
+      "step": 3861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.797650172749423e-05,
+      "loss": 0.6627,
+      "step": 3862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.795439927224933e-05,
+      "loss": 0.6097,
+      "step": 3863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7932297217392904e-05,
+      "loss": 0.6541,
+      "step": 3864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.791019556725104e-05,
+      "loss": 0.6398,
+      "step": 3865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.788809432614971e-05,
+      "loss": 0.678,
+      "step": 3866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.786599349841484e-05,
+      "loss": 0.691,
+      "step": 3867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.784389308837225e-05,
+      "loss": 0.6884,
+      "step": 3868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7821793100347686e-05,
+      "loss": 0.6654,
+      "step": 3869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.779969353866683e-05,
+      "loss": 0.5579,
+      "step": 3870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.777759440765524e-05,
+      "loss": 0.6688,
+      "step": 3871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.775549571163842e-05,
+      "loss": 0.724,
+      "step": 3872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7733397454941766e-05,
+      "loss": 0.5863,
+      "step": 3873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.771129964189063e-05,
+      "loss": 0.6241,
+      "step": 3874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.768920227681022e-05,
+      "loss": 0.6602,
+      "step": 3875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.766710536402574e-05,
+      "loss": 0.5828,
+      "step": 3876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.764500890786218e-05,
+      "loss": 0.6397,
+      "step": 3877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.762291291264456e-05,
+      "loss": 0.6796,
+      "step": 3878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7600817382697746e-05,
+      "loss": 0.7043,
+      "step": 3879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.757872232234654e-05,
+      "loss": 0.6944,
+      "step": 3880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.755662773591561e-05,
+      "loss": 0.6806,
+      "step": 3881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.753453362772962e-05,
+      "loss": 0.638,
+      "step": 3882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.751244000211302e-05,
+      "loss": 0.7323,
+      "step": 3883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.749034686339028e-05,
+      "loss": 0.7049,
+      "step": 3884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7468254215885674e-05,
+      "loss": 0.6216,
+      "step": 3885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7446162063923485e-05,
+      "loss": 0.6849,
+      "step": 3886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.742407041182778e-05,
+      "loss": 0.6377,
+      "step": 3887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740197926392266e-05,
+      "loss": 0.6432,
+      "step": 3888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737988862453199e-05,
+      "loss": 0.6266,
+      "step": 3889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.735779849797965e-05,
+      "loss": 0.5972,
+      "step": 3890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7335708888589356e-05,
+      "loss": 0.6337,
+      "step": 3891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.731361980068476e-05,
+      "loss": 0.6744,
+      "step": 3892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.729153123858936e-05,
+      "loss": 0.3713,
+      "step": 3893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.726944320662661e-05,
+      "loss": 0.6523,
+      "step": 3894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.72473557091198e-05,
+      "loss": 0.5979,
+      "step": 3895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7225268750392196e-05,
+      "loss": 0.704,
+      "step": 3896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7203182334766875e-05,
+      "loss": 0.6531,
+      "step": 3897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.718109646656688e-05,
+      "loss": 0.6783,
+      "step": 3898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.715901115011506e-05,
+      "loss": 0.6343,
+      "step": 3899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.713692638973424e-05,
+      "loss": 0.387,
+      "step": 3900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7114842189747096e-05,
+      "loss": 0.6886,
+      "step": 3901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.709275855447621e-05,
+      "loss": 0.6624,
+      "step": 3902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.707067548824403e-05,
+      "loss": 0.7348,
+      "step": 3903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.704859299537294e-05,
+      "loss": 0.6811,
+      "step": 3904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.702651108018513e-05,
+      "loss": 0.6677,
+      "step": 3905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.700442974700276e-05,
+      "loss": 0.6376,
+      "step": 3906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6982349000147824e-05,
+      "loss": 0.6185,
+      "step": 3907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.696026884394225e-05,
+      "loss": 0.6828,
+      "step": 3908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.693818928270777e-05,
+      "loss": 0.6184,
+      "step": 3909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.691611032076611e-05,
+      "loss": 0.658,
+      "step": 3910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.689403196243877e-05,
+      "loss": 0.6834,
+      "step": 3911
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.68719542120472e-05,
+      "loss": 0.6709,
+      "step": 3912
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.68498770739127e-05,
+      "loss": 0.5492,
+      "step": 3913
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.682780055235649e-05,
+      "loss": 0.6302,
+      "step": 3914
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.68057246516996e-05,
+      "loss": 0.6514,
+      "step": 3915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.678364937626301e-05,
+      "loss": 0.6711,
+      "step": 3916
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.676157473036753e-05,
+      "loss": 0.5984,
+      "step": 3917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.673950071833387e-05,
+      "loss": 0.5772,
+      "step": 3918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.67174273444826e-05,
+      "loss": 0.3622,
+      "step": 3919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6695354613134205e-05,
+      "loss": 0.5364,
+      "step": 3920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.667328252860897e-05,
+      "loss": 0.6567,
+      "step": 3921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665121109522713e-05,
+      "loss": 0.6285,
+      "step": 3922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.662914031730874e-05,
+      "loss": 0.6407,
+      "step": 3923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660707019917376e-05,
+      "loss": 0.7049,
+      "step": 3924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6585000745142e-05,
+      "loss": 0.6412,
+      "step": 3925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6562931959533165e-05,
+      "loss": 0.6714,
+      "step": 3926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.654086384666678e-05,
+      "loss": 0.605,
+      "step": 3927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6518796410862284e-05,
+      "loss": 0.6089,
+      "step": 3928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.649672965643897e-05,
+      "loss": 0.6262,
+      "step": 3929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6474663587716016e-05,
+      "loss": 0.612,
+      "step": 3930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.645259820901239e-05,
+      "loss": 0.6406,
+      "step": 3931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.643053352464706e-05,
+      "loss": 0.6321,
+      "step": 3932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6408469538938716e-05,
+      "loss": 0.6487,
+      "step": 3933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6386406256206e-05,
+      "loss": 0.6947,
+      "step": 3934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.63643436807674e-05,
+      "loss": 0.6347,
+      "step": 3935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.634228181694123e-05,
+      "loss": 0.6879,
+      "step": 3936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.632022066904573e-05,
+      "loss": 0.6217,
+      "step": 3937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.629816024139891e-05,
+      "loss": 0.7174,
+      "step": 3938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.627610053831875e-05,
+      "loss": 0.677,
+      "step": 3939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6254041564122974e-05,
+      "loss": 0.6962,
+      "step": 3940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.623198332312926e-05,
+      "loss": 0.3661,
+      "step": 3941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.620992581965507e-05,
+      "loss": 0.6456,
+      "step": 3942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6187869058017787e-05,
+      "loss": 0.6612,
+      "step": 3943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.616581304253456e-05,
+      "loss": 0.6538,
+      "step": 3944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614375777752251e-05,
+      "loss": 0.7118,
+      "step": 3945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.612170326729849e-05,
+      "loss": 0.6142,
+      "step": 3946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609964951617929e-05,
+      "loss": 0.6474,
+      "step": 3947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6077596528481525e-05,
+      "loss": 0.684,
+      "step": 3948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.605554430852167e-05,
+      "loss": 0.6673,
+      "step": 3949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6033492860616005e-05,
+      "loss": 0.691,
+      "step": 3950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6011442189080725e-05,
+      "loss": 0.6308,
+      "step": 3951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.598939229823181e-05,
+      "loss": 0.6077,
+      "step": 3952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.596734319238515e-05,
+      "loss": 0.657,
+      "step": 3953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.594529487585643e-05,
+      "loss": 0.6281,
+      "step": 3954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.592324735296122e-05,
+      "loss": 0.6435,
+      "step": 3955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.590120062801487e-05,
+      "loss": 0.7042,
+      "step": 3956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.587915470533267e-05,
+      "loss": 0.5998,
+      "step": 3957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.585710958922968e-05,
+      "loss": 0.6526,
+      "step": 3958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.583506528402082e-05,
+      "loss": 0.6732,
+      "step": 3959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.581302179402084e-05,
+      "loss": 0.6382,
+      "step": 3960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5790979123544395e-05,
+      "loss": 0.6088,
+      "step": 3961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5768937276905875e-05,
+      "loss": 0.5853,
+      "step": 3962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5746896258419605e-05,
+      "loss": 0.6762,
+      "step": 3963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5724856072399666e-05,
+      "loss": 0.6336,
+      "step": 3964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.570281672316007e-05,
+      "loss": 0.6751,
+      "step": 3965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5680778215014534e-05,
+      "loss": 0.5213,
+      "step": 3966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.565874055227678e-05,
+      "loss": 0.6402,
+      "step": 3967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.56367037392602e-05,
+      "loss": 0.6481,
+      "step": 3968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.561466778027813e-05,
+      "loss": 0.594,
+      "step": 3969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.559263267964369e-05,
+      "loss": 0.6569,
+      "step": 3970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5570598441669855e-05,
+      "loss": 0.6397,
+      "step": 3971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.554856507066939e-05,
+      "loss": 0.6311,
+      "step": 3972
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.552653257095495e-05,
+      "loss": 0.7076,
+      "step": 3973
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5504500946838966e-05,
+      "loss": 0.6246,
+      "step": 3974
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.548247020263374e-05,
+      "loss": 0.6112,
+      "step": 3975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.546044034265137e-05,
+      "loss": 0.6625,
+      "step": 3976
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.543841137120381e-05,
+      "loss": 0.6245,
+      "step": 3977
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5416383292602796e-05,
+      "loss": 0.3743,
+      "step": 3978
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.539435611115994e-05,
+      "loss": 0.3636,
+      "step": 3979
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.537232983118664e-05,
+      "loss": 0.6723,
+      "step": 3980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.535030445699416e-05,
+      "loss": 0.6197,
+      "step": 3981
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5328279992893505e-05,
+      "loss": 0.6909,
+      "step": 3982
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.530625644319563e-05,
+      "loss": 0.6417,
+      "step": 3983
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.528423381221117e-05,
+      "loss": 0.6569,
+      "step": 3984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.526221210425068e-05,
+      "loss": 0.6312,
+      "step": 3985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.52401913236245e-05,
+      "loss": 0.6094,
+      "step": 3986
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.52181714746428e-05,
+      "loss": 0.6095,
+      "step": 3987
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.519615256161552e-05,
+      "loss": 0.6454,
+      "step": 3988
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5174134588852505e-05,
+      "loss": 0.6549,
+      "step": 3989
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.515211756066333e-05,
+      "loss": 0.6262,
+      "step": 3990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.513010148135743e-05,
+      "loss": 0.5934,
+      "step": 3991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5108086355244044e-05,
+      "loss": 0.6637,
+      "step": 3992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.508607218663225e-05,
+      "loss": 0.691,
+      "step": 3993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.506405897983087e-05,
+      "loss": 0.6425,
+      "step": 3994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5042046739148616e-05,
+      "loss": 0.6344,
+      "step": 3995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.502003546889395e-05,
+      "loss": 0.3864,
+      "step": 3996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.49980251733752e-05,
+      "loss": 0.3385,
+      "step": 3997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.497601585690045e-05,
+      "loss": 0.6489,
+      "step": 3998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.495400752377763e-05,
+      "loss": 0.6699,
+      "step": 3999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.493200017831448e-05,
+      "loss": 0.6059,
+      "step": 4000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490999382481849e-05,
+      "loss": 0.6399,
+      "step": 4001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4887988467597025e-05,
+      "loss": 0.6815,
+      "step": 4002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.486598411095721e-05,
+      "loss": 0.3441,
+      "step": 4003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.484398075920601e-05,
+      "loss": 0.6951,
+      "step": 4004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.482197841665015e-05,
+      "loss": 0.6753,
+      "step": 4005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4799977087596226e-05,
+      "loss": 0.4228,
+      "step": 4006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4777976776350526e-05,
+      "loss": 0.5758,
+      "step": 4007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.475597748721926e-05,
+      "loss": 0.6491,
+      "step": 4008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4733979224508344e-05,
+      "loss": 0.6906,
+      "step": 4009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.471198199252356e-05,
+      "loss": 0.5745,
+      "step": 4010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.468998579557043e-05,
+      "loss": 0.6405,
+      "step": 4011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4667990637954336e-05,
+      "loss": 0.609,
+      "step": 4012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.46459965239804e-05,
+      "loss": 0.6486,
+      "step": 4013
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462400345795356e-05,
+      "loss": 0.6536,
+      "step": 4014
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.460201144417856e-05,
+      "loss": 0.6459,
+      "step": 4015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4580020486959944e-05,
+      "loss": 0.6298,
+      "step": 4016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4558030590601994e-05,
+      "loss": 0.6215,
+      "step": 4017
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.453604175940888e-05,
+      "loss": 0.6553,
+      "step": 4018
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4514053997684465e-05,
+      "loss": 0.7258,
+      "step": 4019
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.449206730973247e-05,
+      "loss": 0.5507,
+      "step": 4020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.447008169985636e-05,
+      "loss": 0.6707,
+      "step": 4021
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4448097172359445e-05,
+      "loss": 0.6683,
+      "step": 4022
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4426113731544755e-05,
+      "loss": 0.6618,
+      "step": 4023
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.440413138171516e-05,
+      "loss": 0.6299,
+      "step": 4024
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4382150127173296e-05,
+      "loss": 0.6417,
+      "step": 4025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.436016997222159e-05,
+      "loss": 0.6359,
+      "step": 4026
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.433819092116223e-05,
+      "loss": 0.6453,
+      "step": 4027
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.431621297829725e-05,
+      "loss": 0.6546,
+      "step": 4028
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.429423614792838e-05,
+      "loss": 0.7233,
+      "step": 4029
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.427226043435721e-05,
+      "loss": 0.6733,
+      "step": 4030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.425028584188505e-05,
+      "loss": 0.6477,
+      "step": 4031
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.422831237481306e-05,
+      "loss": 0.675,
+      "step": 4032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.42063400374421e-05,
+      "loss": 0.6617,
+      "step": 4033
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4184368834072885e-05,
+      "loss": 0.6448,
+      "step": 4034
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.416239876900583e-05,
+      "loss": 0.6656,
+      "step": 4035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.41404298465412e-05,
+      "loss": 0.6433,
+      "step": 4036
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.411846207097899e-05,
+      "loss": 0.5825,
+      "step": 4037
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.409649544661901e-05,
+      "loss": 0.6879,
+      "step": 4038
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.407452997776078e-05,
+      "loss": 0.3453,
+      "step": 4039
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.405256566870367e-05,
+      "loss": 0.6321,
+      "step": 4040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4030602523746753e-05,
+      "loss": 0.6703,
+      "step": 4041
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.400864054718895e-05,
+      "loss": 0.5774,
+      "step": 4042
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3986679743328867e-05,
+      "loss": 0.6208,
+      "step": 4043
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3964720116464974e-05,
+      "loss": 0.6924,
+      "step": 4044
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3942761670895416e-05,
+      "loss": 0.6188,
+      "step": 4045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.392080441091818e-05,
+      "loss": 0.6425,
+      "step": 4046
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3898848340830966e-05,
+      "loss": 0.6727,
+      "step": 4047
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.387689346493131e-05,
+      "loss": 0.6428,
+      "step": 4048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.385493978751643e-05,
+      "loss": 0.6318,
+      "step": 4049
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3832987312883396e-05,
+      "loss": 0.6745,
+      "step": 4050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3811036045328944e-05,
+      "loss": 0.6431,
+      "step": 4051
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.378908598914968e-05,
+      "loss": 0.6301,
+      "step": 4052
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.376713714864189e-05,
+      "loss": 0.6788,
+      "step": 4053
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.374518952810167e-05,
+      "loss": 0.6271,
+      "step": 4054
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.372324313182484e-05,
+      "loss": 0.6657,
+      "step": 4055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.370129796410704e-05,
+      "loss": 0.6882,
+      "step": 4056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3679354029243584e-05,
+      "loss": 0.7494,
+      "step": 4057
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.365741133152963e-05,
+      "loss": 0.6376,
+      "step": 4058
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3635469875260016e-05,
+      "loss": 0.6664,
+      "step": 4059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.361352966472942e-05,
+      "loss": 0.6297,
+      "step": 4060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.359159070423218e-05,
+      "loss": 0.6584,
+      "step": 4061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3569652998062506e-05,
+      "loss": 0.693,
+      "step": 4062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.354771655051424e-05,
+      "loss": 0.6314,
+      "step": 4063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3525781365881056e-05,
+      "loss": 0.6844,
+      "step": 4064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.350384744845638e-05,
+      "loss": 0.3683,
+      "step": 4065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.348191480253333e-05,
+      "loss": 0.6375,
+      "step": 4066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.345998343240488e-05,
+      "loss": 0.6465,
+      "step": 4067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.343805334236361e-05,
+      "loss": 0.6399,
+      "step": 4068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3416124536701995e-05,
+      "loss": 0.6103,
+      "step": 4069
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.339419701971215e-05,
+      "loss": 0.6532,
+      "step": 4070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.337227079568601e-05,
+      "loss": 0.6346,
+      "step": 4071
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.33503458689152e-05,
+      "loss": 0.7006,
+      "step": 4072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.332842224369115e-05,
+      "loss": 0.6162,
+      "step": 4073
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.330649992430498e-05,
+      "loss": 0.6997,
+      "step": 4074
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.328457891504759e-05,
+      "loss": 0.6012,
+      "step": 4075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.326265922020959e-05,
+      "loss": 0.6567,
+      "step": 4076
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.324074084408139e-05,
+      "loss": 0.6772,
+      "step": 4077
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.321882379095308e-05,
+      "loss": 0.6695,
+      "step": 4078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.319690806511454e-05,
+      "loss": 0.6852,
+      "step": 4079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.317499367085534e-05,
+      "loss": 0.6689,
+      "step": 4080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3153080612464835e-05,
+      "loss": 0.6469,
+      "step": 4081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3131168894232093e-05,
+      "loss": 0.6375,
+      "step": 4082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3109258520445946e-05,
+      "loss": 0.6815,
+      "step": 4083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.308734949539491e-05,
+      "loss": 0.6836,
+      "step": 4084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3065441823367316e-05,
+      "loss": 0.6718,
+      "step": 4085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3043535508651136e-05,
+      "loss": 0.6746,
+      "step": 4086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3021630555534165e-05,
+      "loss": 0.6877,
+      "step": 4087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2999726968303866e-05,
+      "loss": 0.5981,
+      "step": 4088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297782475124749e-05,
+      "loss": 0.7474,
+      "step": 4089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.295592390865194e-05,
+      "loss": 0.3693,
+      "step": 4090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2934024444803975e-05,
+      "loss": 0.5652,
+      "step": 4091
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2912126363989935e-05,
+      "loss": 0.5968,
+      "step": 4092
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.289022967049602e-05,
+      "loss": 0.6327,
+      "step": 4093
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.286833436860806e-05,
+      "loss": 0.671,
+      "step": 4094
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.28464404626117e-05,
+      "loss": 0.6333,
+      "step": 4095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2824547956792213e-05,
+      "loss": 0.6292,
+      "step": 4096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.28026568554347e-05,
+      "loss": 0.6951,
+      "step": 4097
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.278076716282391e-05,
+      "loss": 0.6618,
+      "step": 4098
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2758878883244366e-05,
+      "loss": 0.6459,
+      "step": 4099
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2736992020980267e-05,
+      "loss": 0.6728,
+      "step": 4100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.27151065803156e-05,
+      "loss": 0.6576,
+      "step": 4101
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.269322256553399e-05,
+      "loss": 0.6428,
+      "step": 4102
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2671339980918864e-05,
+      "loss": 0.6722,
+      "step": 4103
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.264945883075331e-05,
+      "loss": 0.6946,
+      "step": 4104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.262757911932017e-05,
+      "loss": 0.6551,
+      "step": 4105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.260570085090199e-05,
+      "loss": 0.6604,
+      "step": 4106
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2583824029781044e-05,
+      "loss": 0.6675,
+      "step": 4107
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.256194866023929e-05,
+      "loss": 0.6569,
+      "step": 4108
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.254007474655846e-05,
+      "loss": 0.6128,
+      "step": 4109
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2518202293019936e-05,
+      "loss": 0.6709,
+      "step": 4110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.249633130390489e-05,
+      "loss": 0.6128,
+      "step": 4111
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2474461783494093e-05,
+      "loss": 0.3915,
+      "step": 4112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.245259373606818e-05,
+      "loss": 0.6451,
+      "step": 4113
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2430727165907355e-05,
+      "loss": 0.6376,
+      "step": 4114
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.240886207729163e-05,
+      "loss": 0.659,
+      "step": 4115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.238699847450066e-05,
+      "loss": 0.6635,
+      "step": 4116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.236513636181389e-05,
+      "loss": 0.6231,
+      "step": 4117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2343275743510366e-05,
+      "loss": 0.6348,
+      "step": 4118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.232141662386894e-05,
+      "loss": 0.6491,
+      "step": 4119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.22995590071681e-05,
+      "loss": 0.6288,
+      "step": 4120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.22777028976861e-05,
+      "loss": 0.6391,
+      "step": 4121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.225584829970085e-05,
+      "loss": 0.6657,
+      "step": 4122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.223399521749001e-05,
+      "loss": 0.6623,
+      "step": 4123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.221214365533087e-05,
+      "loss": 0.6049,
+      "step": 4124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.219029361750051e-05,
+      "loss": 0.69,
+      "step": 4125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2168445108275636e-05,
+      "loss": 0.6147,
+      "step": 4126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.214659813193272e-05,
+      "loss": 0.6848,
+      "step": 4127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.212475269274789e-05,
+      "loss": 0.5927,
+      "step": 4128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2102908794996984e-05,
+      "loss": 0.636,
+      "step": 4129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.208106644295556e-05,
+      "loss": 0.5884,
+      "step": 4130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.205922564089883e-05,
+      "loss": 0.609,
+      "step": 4131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.203738639310173e-05,
+      "loss": 0.6183,
+      "step": 4132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.201554870383889e-05,
+      "loss": 0.6338,
+      "step": 4133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.199371257738466e-05,
+      "loss": 0.5918,
+      "step": 4134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.197187801801301e-05,
+      "loss": 0.6411,
+      "step": 4135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.195004502999771e-05,
+      "loss": 0.36,
+      "step": 4136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.19282136176121e-05,
+      "loss": 0.6445,
+      "step": 4137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1906383785129316e-05,
+      "loss": 0.6268,
+      "step": 4138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1884555536822125e-05,
+      "loss": 0.6408,
+      "step": 4139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.186272887696303e-05,
+      "loss": 0.5758,
+      "step": 4140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.184090380982415e-05,
+      "loss": 0.6418,
+      "step": 4141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.181908033967739e-05,
+      "loss": 0.6875,
+      "step": 4142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.179725847079424e-05,
+      "loss": 0.6558,
+      "step": 4143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.177543820744597e-05,
+      "loss": 0.5908,
+      "step": 4144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1753619553903465e-05,
+      "loss": 0.6252,
+      "step": 4145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1731802514437354e-05,
+      "loss": 0.4088,
+      "step": 4146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1709987093317853e-05,
+      "loss": 0.674,
+      "step": 4147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.168817329481501e-05,
+      "loss": 0.6628,
+      "step": 4148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.166636112319842e-05,
+      "loss": 0.6258,
+      "step": 4149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1644550582737425e-05,
+      "loss": 0.5942,
+      "step": 4150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.162274167770103e-05,
+      "loss": 0.6058,
+      "step": 4151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.160093441235794e-05,
+      "loss": 0.6532,
+      "step": 4152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.157912879097651e-05,
+      "loss": 0.653,
+      "step": 4153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1557324817824786e-05,
+      "loss": 0.6751,
+      "step": 4154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.153552249717049e-05,
+      "loss": 0.6132,
+      "step": 4155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.151372183328104e-05,
+      "loss": 0.5918,
+      "step": 4156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1491922830423494e-05,
+      "loss": 0.6245,
+      "step": 4157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.147012549286462e-05,
+      "loss": 0.3483,
+      "step": 4158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1448329824870825e-05,
+      "loss": 0.6211,
+      "step": 4159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1426535830708227e-05,
+      "loss": 0.6299,
+      "step": 4160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.140474351464257e-05,
+      "loss": 0.6085,
+      "step": 4161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1382952880939346e-05,
+      "loss": 0.6199,
+      "step": 4162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1361163933863594e-05,
+      "loss": 0.6676,
+      "step": 4163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1339376677680175e-05,
+      "loss": 0.6785,
+      "step": 4164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.131759111665349e-05,
+      "loss": 0.3854,
+      "step": 4165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1295807255047676e-05,
+      "loss": 0.6437,
+      "step": 4166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.127402509712651e-05,
+      "loss": 0.7108,
+      "step": 4167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1252244647153485e-05,
+      "loss": 0.652,
+      "step": 4168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.123046590939165e-05,
+      "loss": 0.6813,
+      "step": 4169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.120868888810387e-05,
+      "loss": 0.6604,
+      "step": 4170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.118691358755254e-05,
+      "loss": 0.6364,
+      "step": 4171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.116514001199979e-05,
+      "loss": 0.6704,
+      "step": 4172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.114336816570739e-05,
+      "loss": 0.6332,
+      "step": 4173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1121598052936785e-05,
+      "loss": 0.6264,
+      "step": 4174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.109982967794906e-05,
+      "loss": 0.5981,
+      "step": 4175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1078063045004976e-05,
+      "loss": 0.6794,
+      "step": 4176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.105629815836495e-05,
+      "loss": 0.652,
+      "step": 4177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.103453502228905e-05,
+      "loss": 0.6115,
+      "step": 4178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1012773641037015e-05,
+      "loss": 0.6341,
+      "step": 4179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0991014018868255e-05,
+      "loss": 0.6544,
+      "step": 4180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.096925616004176e-05,
+      "loss": 0.6756,
+      "step": 4181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0947500068816264e-05,
+      "loss": 0.6658,
+      "step": 4182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0925745749450107e-05,
+      "loss": 0.6486,
+      "step": 4183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.090399320620131e-05,
+      "loss": 0.652,
+      "step": 4184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.08822424433275e-05,
+      "loss": 0.648,
+      "step": 4185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.086049346508604e-05,
+      "loss": 0.6035,
+      "step": 4186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.083874627573383e-05,
+      "loss": 0.6122,
+      "step": 4187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0817000879527515e-05,
+      "loss": 0.6746,
+      "step": 4188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.079525728072334e-05,
+      "loss": 0.6566,
+      "step": 4189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0773515483577235e-05,
+      "loss": 0.6504,
+      "step": 4190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.075177549234471e-05,
+      "loss": 0.5967,
+      "step": 4191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.073003731128102e-05,
+      "loss": 0.6815,
+      "step": 4192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070830094464096e-05,
+      "loss": 0.6395,
+      "step": 4193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.068656639667905e-05,
+      "loss": 0.5825,
+      "step": 4194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066483367164944e-05,
+      "loss": 0.6585,
+      "step": 4195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.064310277380587e-05,
+      "loss": 0.6468,
+      "step": 4196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0621373707401805e-05,
+      "loss": 0.643,
+      "step": 4197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059964647669025e-05,
+      "loss": 0.5977,
+      "step": 4198
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057792108592398e-05,
+      "loss": 0.6639,
+      "step": 4199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055619753935528e-05,
+      "loss": 0.6697,
+      "step": 4200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053447584123616e-05,
+      "loss": 0.6925,
+      "step": 4201
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051275599581822e-05,
+      "loss": 0.6738,
+      "step": 4202
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0491038007352754e-05,
+      "loss": 0.6335,
+      "step": 4203
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046932188009062e-05,
+      "loss": 0.5305,
+      "step": 4204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0447607618282363e-05,
+      "loss": 0.6057,
+      "step": 4205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.042589522617814e-05,
+      "loss": 0.7,
+      "step": 4206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0404184708027764e-05,
+      "loss": 0.68,
+      "step": 4207
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0382476068080655e-05,
+      "loss": 0.6777,
+      "step": 4208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.03607693105859e-05,
+      "loss": 0.6339,
+      "step": 4209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0339064439792154e-05,
+      "loss": 0.6329,
+      "step": 4210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031736145994778e-05,
+      "loss": 0.7093,
+      "step": 4211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0295660375300714e-05,
+      "loss": 0.609,
+      "step": 4212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0273961190098555e-05,
+      "loss": 0.6039,
+      "step": 4213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0252263908588505e-05,
+      "loss": 0.5615,
+      "step": 4214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.023056853501743e-05,
+      "loss": 0.6023,
+      "step": 4215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0208875073631767e-05,
+      "loss": 0.6406,
+      "step": 4216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.018718352867763e-05,
+      "loss": 0.6582,
+      "step": 4217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.016549390440071e-05,
+      "loss": 0.6146,
+      "step": 4218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.01438062050464e-05,
+      "loss": 0.6133,
+      "step": 4219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0122120434859595e-05,
+      "loss": 0.65,
+      "step": 4220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.010043659808496e-05,
+      "loss": 0.6265,
+      "step": 4221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.007875469896665e-05,
+      "loss": 0.6385,
+      "step": 4222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.005707474174852e-05,
+      "loss": 0.6027,
+      "step": 4223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.003539673067401e-05,
+      "loss": 0.6559,
+      "step": 4224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.001372066998621e-05,
+      "loss": 0.6568,
+      "step": 4225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.999204656392777e-05,
+      "loss": 0.6577,
+      "step": 4226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.997037441674103e-05,
+      "loss": 0.6198,
+      "step": 4227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9948704232667886e-05,
+      "loss": 0.6893,
+      "step": 4228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.99270360159499e-05,
+      "loss": 0.6715,
+      "step": 4229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.99053697708282e-05,
+      "loss": 0.6606,
+      "step": 4230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.988370550154359e-05,
+      "loss": 0.6044,
+      "step": 4231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.98620432123364e-05,
+      "loss": 0.6712,
+      "step": 4232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9840382907446654e-05,
+      "loss": 0.6538,
+      "step": 4233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.981872459111394e-05,
+      "loss": 0.5862,
+      "step": 4234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.97970682675775e-05,
+      "loss": 0.7187,
+      "step": 4235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.977541394107613e-05,
+      "loss": 0.6338,
+      "step": 4236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9753761615848286e-05,
+      "loss": 0.6799,
+      "step": 4237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.973211129613198e-05,
+      "loss": 0.6014,
+      "step": 4238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.971046298616489e-05,
+      "loss": 0.6084,
+      "step": 4239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9688816690184254e-05,
+      "loss": 0.7508,
+      "step": 4240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9667172412426964e-05,
+      "loss": 0.6277,
+      "step": 4241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9645530157129423e-05,
+      "loss": 0.6389,
+      "step": 4242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.962388992852778e-05,
+      "loss": 0.601,
+      "step": 4243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.960225173085765e-05,
+      "loss": 0.6178,
+      "step": 4244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.958061556835435e-05,
+      "loss": 0.7073,
+      "step": 4245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.955898144525272e-05,
+      "loss": 0.6823,
+      "step": 4246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.953734936578729e-05,
+      "loss": 0.6391,
+      "step": 4247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.951571933419208e-05,
+      "loss": 0.6794,
+      "step": 4248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9494091354700814e-05,
+      "loss": 0.6253,
+      "step": 4249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9472465431546735e-05,
+      "loss": 0.6121,
+      "step": 4250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.945084156896275e-05,
+      "loss": 0.6263,
+      "step": 4251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.94292197711813e-05,
+      "loss": 0.6776,
+      "step": 4252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.940760004243449e-05,
+      "loss": 0.6444,
+      "step": 4253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.938598238695394e-05,
+      "loss": 0.6508,
+      "step": 4254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.936436680897093e-05,
+      "loss": 0.6621,
+      "step": 4255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9342753312716294e-05,
+      "loss": 0.6462,
+      "step": 4256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.93211419024205e-05,
+      "loss": 0.7466,
+      "step": 4257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.929953258231355e-05,
+      "loss": 0.5873,
+      "step": 4258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.92779253566251e-05,
+      "loss": 0.568,
+      "step": 4259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.925632022958434e-05,
+      "loss": 0.6756,
+      "step": 4260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.923471720542007e-05,
+      "loss": 0.6271,
+      "step": 4261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.921311628836071e-05,
+      "loss": 0.6369,
+      "step": 4262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.919151748263421e-05,
+      "loss": 0.6424,
+      "step": 4263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.916992079246816e-05,
+      "loss": 0.6201,
+      "step": 4264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.91483262220897e-05,
+      "loss": 0.6266,
+      "step": 4265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.912673377572558e-05,
+      "loss": 0.6545,
+      "step": 4266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.91051434576021e-05,
+      "loss": 0.6212,
+      "step": 4267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.908355527194519e-05,
+      "loss": 0.6088,
+      "step": 4268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.906196922298031e-05,
+      "loss": 0.6103,
+      "step": 4269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.904038531493257e-05,
+      "loss": 0.6564,
+      "step": 4270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.901880355202656e-05,
+      "loss": 0.5656,
+      "step": 4271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8997223938486576e-05,
+      "loss": 0.6166,
+      "step": 4272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.897564647853638e-05,
+      "loss": 0.617,
+      "step": 4273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.895407117639939e-05,
+      "loss": 0.634,
+      "step": 4274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8932498036298555e-05,
+      "loss": 0.6644,
+      "step": 4275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8910927062456436e-05,
+      "loss": 0.645,
+      "step": 4276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8889358259095126e-05,
+      "loss": 0.6476,
+      "step": 4277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8867791630436346e-05,
+      "loss": 0.6442,
+      "step": 4278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8846227180701335e-05,
+      "loss": 0.5708,
+      "step": 4279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8824664914110966e-05,
+      "loss": 0.6117,
+      "step": 4280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8803104834885626e-05,
+      "loss": 0.6918,
+      "step": 4281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.878154694724533e-05,
+      "loss": 0.6071,
+      "step": 4282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.87599912554096e-05,
+      "loss": 0.6408,
+      "step": 4283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.873843776359761e-05,
+      "loss": 0.6628,
+      "step": 4284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.871688647602801e-05,
+      "loss": 0.6727,
+      "step": 4285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8695337396919093e-05,
+      "loss": 0.6249,
+      "step": 4286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8673790530488686e-05,
+      "loss": 0.558,
+      "step": 4287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.86522458809542e-05,
+      "loss": 0.6931,
+      "step": 4288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.863070345253258e-05,
+      "loss": 0.6132,
+      "step": 4289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.860916324944037e-05,
+      "loss": 0.6077,
+      "step": 4290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8587625275893665e-05,
+      "loss": 0.6176,
+      "step": 4291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.856608953610814e-05,
+      "loss": 0.6613,
+      "step": 4292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8544556034298976e-05,
+      "loss": 0.6357,
+      "step": 4293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.852302477468101e-05,
+      "loss": 0.6324,
+      "step": 4294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.850149576146854e-05,
+      "loss": 0.6325,
+      "step": 4295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.847996899887551e-05,
+      "loss": 0.4372,
+      "step": 4296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.845844449111535e-05,
+      "loss": 0.613,
+      "step": 4297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.843692224240113e-05,
+      "loss": 0.627,
+      "step": 4298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.841540225694539e-05,
+      "loss": 0.6755,
+      "step": 4299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8393884538960276e-05,
+      "loss": 0.6843,
+      "step": 4300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8372369092657504e-05,
+      "loss": 0.6149,
+      "step": 4301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.83508559222483e-05,
+      "loss": 0.6425,
+      "step": 4302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.832934503194348e-05,
+      "loss": 0.6416,
+      "step": 4303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8307836425953426e-05,
+      "loss": 0.6529,
+      "step": 4304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.828633010848801e-05,
+      "loss": 0.6449,
+      "step": 4305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.826482608375671e-05,
+      "loss": 0.6389,
+      "step": 4306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.824332435596854e-05,
+      "loss": 0.7075,
+      "step": 4307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.822182492933208e-05,
+      "loss": 0.6582,
+      "step": 4308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.820032780805542e-05,
+      "loss": 0.6331,
+      "step": 4309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.817883299634626e-05,
+      "loss": 0.5934,
+      "step": 4310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.815734049841177e-05,
+      "loss": 0.7217,
+      "step": 4311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.813585031845873e-05,
+      "loss": 0.6651,
+      "step": 4312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.811436246069344e-05,
+      "loss": 0.6199,
+      "step": 4313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8092876929321775e-05,
+      "loss": 0.6028,
+      "step": 4314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8071393728549074e-05,
+      "loss": 0.6788,
+      "step": 4315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.804991286258035e-05,
+      "loss": 0.6843,
+      "step": 4316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.802843433562001e-05,
+      "loss": 0.7044,
+      "step": 4317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.800695815187213e-05,
+      "loss": 0.6273,
+      "step": 4318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.798548431554025e-05,
+      "loss": 0.6598,
+      "step": 4319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.79640128308275e-05,
+      "loss": 0.6535,
+      "step": 4320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.79425437019365e-05,
+      "loss": 0.6554,
+      "step": 4321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.792107693306946e-05,
+      "loss": 0.687,
+      "step": 4322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.789961252842806e-05,
+      "loss": 0.6295,
+      "step": 4323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.787815049221361e-05,
+      "loss": 0.6781,
+      "step": 4324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.785669082862688e-05,
+      "loss": 0.6202,
+      "step": 4325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7835233541868196e-05,
+      "loss": 0.5885,
+      "step": 4326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7813778636137465e-05,
+      "loss": 0.6988,
+      "step": 4327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7792326115634016e-05,
+      "loss": 0.5908,
+      "step": 4328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.777087598455687e-05,
+      "loss": 0.6286,
+      "step": 4329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7749428247104425e-05,
+      "loss": 0.6309,
+      "step": 4330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7727982907474704e-05,
+      "loss": 0.656,
+      "step": 4331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.770653996986523e-05,
+      "loss": 0.6678,
+      "step": 4332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.768509943847309e-05,
+      "loss": 0.6174,
+      "step": 4333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.766366131749481e-05,
+      "loss": 0.6394,
+      "step": 4334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.764222561112657e-05,
+      "loss": 0.6854,
+      "step": 4335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7620792323563966e-05,
+      "loss": 0.5918,
+      "step": 4336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.75993614590022e-05,
+      "loss": 0.683,
+      "step": 4337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7577933021635934e-05,
+      "loss": 0.6819,
+      "step": 4338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7556507015659435e-05,
+      "loss": 0.6924,
+      "step": 4339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7535083445266395e-05,
+      "loss": 0.6451,
+      "step": 4340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.751366231465012e-05,
+      "loss": 0.6768,
+      "step": 4341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.749224362800338e-05,
+      "loss": 0.6352,
+      "step": 4342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.74708273895185e-05,
+      "loss": 0.6421,
+      "step": 4343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7449413603387284e-05,
+      "loss": 0.5915,
+      "step": 4344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.742800227380113e-05,
+      "loss": 0.6194,
+      "step": 4345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7406593404950876e-05,
+      "loss": 0.6037,
+      "step": 4346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7385187001026916e-05,
+      "loss": 0.6311,
+      "step": 4347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.736378306621916e-05,
+      "loss": 0.6293,
+      "step": 4348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7342381604717046e-05,
+      "loss": 0.5772,
+      "step": 4349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.732098262070948e-05,
+      "loss": 0.7023,
+      "step": 4350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.729958611838496e-05,
+      "loss": 0.686,
+      "step": 4351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.727819210193143e-05,
+      "loss": 0.6757,
+      "step": 4352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7256800575536376e-05,
+      "loss": 0.7331,
+      "step": 4353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.723541154338679e-05,
+      "loss": 0.3547,
+      "step": 4354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.721402500966921e-05,
+      "loss": 0.6451,
+      "step": 4355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.71926409785696e-05,
+      "loss": 0.6322,
+      "step": 4356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.717125945427352e-05,
+      "loss": 0.665,
+      "step": 4357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7149880440966e-05,
+      "loss": 0.6519,
+      "step": 4358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.712850394283161e-05,
+      "loss": 0.6706,
+      "step": 4359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7107129964054366e-05,
+      "loss": 0.7147,
+      "step": 4360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.708575850881787e-05,
+      "loss": 0.6517,
+      "step": 4361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7064389581305146e-05,
+      "loss": 0.6742,
+      "step": 4362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.70430231856988e-05,
+      "loss": 0.6315,
+      "step": 4363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.702165932618088e-05,
+      "loss": 0.672,
+      "step": 4364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.700029800693299e-05,
+      "loss": 0.6333,
+      "step": 4365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.69789392321362e-05,
+      "loss": 0.365,
+      "step": 4366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6957583005971115e-05,
+      "loss": 0.6719,
+      "step": 4367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.693622933261779e-05,
+      "loss": 0.7064,
+      "step": 4368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.691487821625583e-05,
+      "loss": 0.6782,
+      "step": 4369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.689352966106431e-05,
+      "loss": 0.55,
+      "step": 4370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6872183671221836e-05,
+      "loss": 0.7069,
+      "step": 4371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.685084025090646e-05,
+      "loss": 0.6238,
+      "step": 4372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.68294994042958e-05,
+      "loss": 0.3795,
+      "step": 4373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.680816113556689e-05,
+      "loss": 0.6544,
+      "step": 4374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.678682544889633e-05,
+      "loss": 0.3939,
+      "step": 4375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6765492348460176e-05,
+      "loss": 0.6182,
+      "step": 4376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6744161838434e-05,
+      "loss": 0.6259,
+      "step": 4377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.672283392299282e-05,
+      "loss": 0.6431,
+      "step": 4378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.670150860631122e-05,
+      "loss": 0.6226,
+      "step": 4379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.668018589256321e-05,
+      "loss": 0.6826,
+      "step": 4380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.665886578592234e-05,
+      "loss": 0.6559,
+      "step": 4381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6637548290561595e-05,
+      "loss": 0.689,
+      "step": 4382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.661623341065352e-05,
+      "loss": 0.5936,
+      "step": 4383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6594921150370075e-05,
+      "loss": 0.6533,
+      "step": 4384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6573611513882755e-05,
+      "loss": 0.7053,
+      "step": 4385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.655230450536251e-05,
+      "loss": 0.6088,
+      "step": 4386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.653100012897983e-05,
+      "loss": 0.581,
+      "step": 4387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.650969838890461e-05,
+      "loss": 0.6345,
+      "step": 4388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.648839928930632e-05,
+      "loss": 0.6856,
+      "step": 4389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.646710283435381e-05,
+      "loss": 0.674,
+      "step": 4390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.644580902821548e-05,
+      "loss": 0.6255,
+      "step": 4391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.642451787505924e-05,
+      "loss": 0.5913,
+      "step": 4392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6403229379052384e-05,
+      "loss": 0.6796,
+      "step": 4393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6381943544361786e-05,
+      "loss": 0.6322,
+      "step": 4394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.636066037515373e-05,
+      "loss": 0.5822,
+      "step": 4395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.633937987559402e-05,
+      "loss": 0.7148,
+      "step": 4396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.631810204984789e-05,
+      "loss": 0.6188,
+      "step": 4397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.629682690208011e-05,
+      "loss": 0.6153,
+      "step": 4398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.627555443645487e-05,
+      "loss": 0.6671,
+      "step": 4399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6254284657135905e-05,
+      "loss": 0.3898,
+      "step": 4400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.623301756828632e-05,
+      "loss": 0.6752,
+      "step": 4401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.621175317406881e-05,
+      "loss": 0.6644,
+      "step": 4402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.619049147864545e-05,
+      "loss": 0.6401,
+      "step": 4403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6169232486177843e-05,
+      "loss": 0.634,
+      "step": 4404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.614797620082703e-05,
+      "loss": 0.6272,
+      "step": 4405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.612672262675356e-05,
+      "loss": 0.723,
+      "step": 4406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.610547176811739e-05,
+      "loss": 0.6497,
+      "step": 4407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.608422362907801e-05,
+      "loss": 0.6087,
+      "step": 4408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.606297821379434e-05,
+      "loss": 0.7093,
+      "step": 4409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.604173552642478e-05,
+      "loss": 0.7288,
+      "step": 4410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6020495571127175e-05,
+      "loss": 0.6312,
+      "step": 4411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.599925835205891e-05,
+      "loss": 0.6257,
+      "step": 4412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.59780238733767e-05,
+      "loss": 0.6712,
+      "step": 4413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.595679213923685e-05,
+      "loss": 0.6986,
+      "step": 4414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.593556315379506e-05,
+      "loss": 0.6792,
+      "step": 4415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.591433692120653e-05,
+      "loss": 0.6346,
+      "step": 4416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.589311344562588e-05,
+      "loss": 0.3896,
+      "step": 4417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5871892731207246e-05,
+      "loss": 0.59,
+      "step": 4418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5850674782104146e-05,
+      "loss": 0.607,
+      "step": 4419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.582945960246963e-05,
+      "loss": 0.6584,
+      "step": 4420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.580824719645616e-05,
+      "loss": 0.3633,
+      "step": 4421
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.578703756821571e-05,
+      "loss": 0.6621,
+      "step": 4422
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.57658307218996e-05,
+      "loss": 0.6079,
+      "step": 4423
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.574462666165876e-05,
+      "loss": 0.6821,
+      "step": 4424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.572342539164343e-05,
+      "loss": 0.5772,
+      "step": 4425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.57022269160034e-05,
+      "loss": 0.5645,
+      "step": 4426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5681031238887866e-05,
+      "loss": 0.654,
+      "step": 4427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5659838364445505e-05,
+      "loss": 0.7291,
+      "step": 4428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.563864829682439e-05,
+      "loss": 0.6708,
+      "step": 4429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.561746104017214e-05,
+      "loss": 0.6068,
+      "step": 4430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.559627659863572e-05,
+      "loss": 0.6623,
+      "step": 4431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5575094976361625e-05,
+      "loss": 0.6531,
+      "step": 4432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.555391617749575e-05,
+      "loss": 0.6001,
+      "step": 4433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.553274020618347e-05,
+      "loss": 0.6322,
+      "step": 4434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5511567066569556e-05,
+      "loss": 0.6274,
+      "step": 4435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5490396762798296e-05,
+      "loss": 0.619,
+      "step": 4436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.546922929901335e-05,
+      "loss": 0.6284,
+      "step": 4437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.544806467935788e-05,
+      "loss": 0.6739,
+      "step": 4438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.542690290797447e-05,
+      "loss": 0.6493,
+      "step": 4439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5405743989005146e-05,
+      "loss": 0.6572,
+      "step": 4440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5384587926591355e-05,
+      "loss": 0.5705,
+      "step": 4441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5363434724874026e-05,
+      "loss": 0.6102,
+      "step": 4442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.534228438799348e-05,
+      "loss": 0.6613,
+      "step": 4443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.532113692008956e-05,
+      "loss": 0.6834,
+      "step": 4444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.529999232530141e-05,
+      "loss": 0.5908,
+      "step": 4445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5278850607767765e-05,
+      "loss": 0.6481,
+      "step": 4446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.525771177162669e-05,
+      "loss": 0.6397,
+      "step": 4447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5236575821015726e-05,
+      "loss": 0.63,
+      "step": 4448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.521544276007184e-05,
+      "loss": 0.6616,
+      "step": 4449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.519431259293147e-05,
+      "loss": 0.6148,
+      "step": 4450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.51731853237304e-05,
+      "loss": 0.6348,
+      "step": 4451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.515206095660394e-05,
+      "loss": 0.5913,
+      "step": 4452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.513093949568678e-05,
+      "loss": 0.6212,
+      "step": 4453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.510982094511307e-05,
+      "loss": 0.6101,
+      "step": 4454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.508870530901636e-05,
+      "loss": 0.6425,
+      "step": 4455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.506759259152964e-05,
+      "loss": 0.6807,
+      "step": 4456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.504648279678537e-05,
+      "loss": 0.6259,
+      "step": 4457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.502537592891535e-05,
+      "loss": 0.6531,
+      "step": 4458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.500427199205091e-05,
+      "loss": 0.6223,
+      "step": 4459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.49831709903227e-05,
+      "loss": 0.6545,
+      "step": 4460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.49620729278609e-05,
+      "loss": 0.6461,
+      "step": 4461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.494097780879504e-05,
+      "loss": 0.6499,
+      "step": 4462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.491988563725412e-05,
+      "loss": 0.664,
+      "step": 4463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.489879641736651e-05,
+      "loss": 0.3804,
+      "step": 4464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4877710153260055e-05,
+      "loss": 0.6356,
+      "step": 4465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4856626849062e-05,
+      "loss": 0.5577,
+      "step": 4466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.483554650889901e-05,
+      "loss": 0.6555,
+      "step": 4467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4814469136897165e-05,
+      "loss": 0.6239,
+      "step": 4468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.479339473718199e-05,
+      "loss": 0.6737,
+      "step": 4469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.477232331387839e-05,
+      "loss": 0.6037,
+      "step": 4470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.475125487111071e-05,
+      "loss": 0.6531,
+      "step": 4471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.47301894130027e-05,
+      "loss": 0.5717,
+      "step": 4472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4709126943677564e-05,
+      "loss": 0.6249,
+      "step": 4473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.468806746725783e-05,
+      "loss": 0.6434,
+      "step": 4474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4667010987865584e-05,
+      "loss": 0.355,
+      "step": 4475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.464595750962217e-05,
+      "loss": 0.6473,
+      "step": 4476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.462490703664846e-05,
+      "loss": 0.6883,
+      "step": 4477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.460385957306467e-05,
+      "loss": 0.6507,
+      "step": 4478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.458281512299048e-05,
+      "loss": 0.645,
+      "step": 4479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4561773690544916e-05,
+      "loss": 0.6528,
+      "step": 4480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.454073527984647e-05,
+      "loss": 0.6151,
+      "step": 4481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.451969989501301e-05,
+      "loss": 0.6313,
+      "step": 4482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.449866754016185e-05,
+      "loss": 0.6186,
+      "step": 4483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4477638219409635e-05,
+      "loss": 0.628,
+      "step": 4484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.445661193687253e-05,
+      "loss": 0.615,
+      "step": 4485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.443558869666598e-05,
+      "loss": 0.6269,
+      "step": 4486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4414568502904935e-05,
+      "loss": 0.6951,
+      "step": 4487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.439355135970368e-05,
+      "loss": 0.6563,
+      "step": 4488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.437253727117596e-05,
+      "loss": 0.692,
+      "step": 4489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.435152624143488e-05,
+      "loss": 0.5667,
+      "step": 4490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.433051827459297e-05,
+      "loss": 0.6664,
+      "step": 4491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.430951337476213e-05,
+      "loss": 0.6133,
+      "step": 4492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.42885115460537e-05,
+      "loss": 0.6405,
+      "step": 4493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4267512792578374e-05,
+      "loss": 0.6691,
+      "step": 4494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.424651711844632e-05,
+      "loss": 0.6696,
+      "step": 4495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.422552452776698e-05,
+      "loss": 0.676,
+      "step": 4496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.420453502464934e-05,
+      "loss": 0.5951,
+      "step": 4497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4183548613201655e-05,
+      "loss": 0.63,
+      "step": 4498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.416256529753165e-05,
+      "loss": 0.6773,
+      "step": 4499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.41415850817464e-05,
+      "loss": 0.6044,
+      "step": 4500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4120607969952436e-05,
+      "loss": 0.6235,
+      "step": 4501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.409963396625559e-05,
+      "loss": 0.6918,
+      "step": 4502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.407866307476116e-05,
+      "loss": 0.6558,
+      "step": 4503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.405769529957381e-05,
+      "loss": 0.5835,
+      "step": 4504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4036730644797584e-05,
+      "loss": 0.6464,
+      "step": 4505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.401576911453594e-05,
+      "loss": 0.5816,
+      "step": 4506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.399481071289171e-05,
+      "loss": 0.6795,
+      "step": 4507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3973855443967095e-05,
+      "loss": 0.6618,
+      "step": 4508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3952903311863714e-05,
+      "loss": 0.6379,
+      "step": 4509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.393195432068255e-05,
+      "loss": 0.6259,
+      "step": 4510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.391100847452399e-05,
+      "loss": 0.6685,
+      "step": 4511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.389006577748779e-05,
+      "loss": 0.6183,
+      "step": 4512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.386912623367311e-05,
+      "loss": 0.6647,
+      "step": 4513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.384818984717845e-05,
+      "loss": 0.6429,
+      "step": 4514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.382725662210175e-05,
+      "loss": 0.6189,
+      "step": 4515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.380632656254027e-05,
+      "loss": 0.6609,
+      "step": 4516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.378539967259073e-05,
+      "loss": 0.6886,
+      "step": 4517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.376447595634911e-05,
+      "loss": 0.6458,
+      "step": 4518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.374355541791092e-05,
+      "loss": 0.647,
+      "step": 4519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3722638061370903e-05,
+      "loss": 0.6191,
+      "step": 4520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3701723890823276e-05,
+      "loss": 0.6117,
+      "step": 4521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.36808129103616e-05,
+      "loss": 0.601,
+      "step": 4522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.365990512407881e-05,
+      "loss": 0.6809,
+      "step": 4523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3639000536067224e-05,
+      "loss": 0.7217,
+      "step": 4524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3618099150418496e-05,
+      "loss": 0.6722,
+      "step": 4525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.359720097122374e-05,
+      "loss": 0.657,
+      "step": 4526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.357630600257334e-05,
+      "loss": 0.7002,
+      "step": 4527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.355541424855713e-05,
+      "loss": 0.6164,
+      "step": 4528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.353452571326425e-05,
+      "loss": 0.6828,
+      "step": 4529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3513640400783295e-05,
+      "loss": 0.6997,
+      "step": 4530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.349275831520212e-05,
+      "loss": 0.6113,
+      "step": 4531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.347187946060807e-05,
+      "loss": 0.6449,
+      "step": 4532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3451003841087735e-05,
+      "loss": 0.6277,
+      "step": 4533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.343013146072717e-05,
+      "loss": 0.6444,
+      "step": 4534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3409262323611724e-05,
+      "loss": 0.613,
+      "step": 4535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3388396433826196e-05,
+      "loss": 0.5704,
+      "step": 4536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3367533795454643e-05,
+      "loss": 0.6808,
+      "step": 4537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.334667441258057e-05,
+      "loss": 0.6924,
+      "step": 4538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3325818289286805e-05,
+      "loss": 0.6322,
+      "step": 4539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.330496542965556e-05,
+      "loss": 0.7051,
+      "step": 4540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.328411583776838e-05,
+      "loss": 0.5794,
+      "step": 4541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.326326951770622e-05,
+      "loss": 0.7505,
+      "step": 4542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3242426473549304e-05,
+      "loss": 0.6894,
+      "step": 4543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.322158670937733e-05,
+      "loss": 0.7104,
+      "step": 4544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.320075022926925e-05,
+      "loss": 0.6563,
+      "step": 4545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3179917037303464e-05,
+      "loss": 0.6525,
+      "step": 4546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.315908713755763e-05,
+      "loss": 0.6657,
+      "step": 4547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.313826053410888e-05,
+      "loss": 0.5346,
+      "step": 4548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.311743723103357e-05,
+      "loss": 0.6601,
+      "step": 4549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.309661723240753e-05,
+      "loss": 0.6136,
+      "step": 4550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.307580054230585e-05,
+      "loss": 0.6228,
+      "step": 4551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.305498716480305e-05,
+      "loss": 0.6455,
+      "step": 4552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.303417710397292e-05,
+      "loss": 0.5948,
+      "step": 4553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.30133703638887e-05,
+      "loss": 0.6706,
+      "step": 4554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.299256694862286e-05,
+      "loss": 0.6784,
+      "step": 4555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2971766862247335e-05,
+      "loss": 0.7039,
+      "step": 4556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.295097010883332e-05,
+      "loss": 0.6618,
+      "step": 4557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.293017669245144e-05,
+      "loss": 0.6499,
+      "step": 4558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2909386617171565e-05,
+      "loss": 0.676,
+      "step": 4559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.288859988706301e-05,
+      "loss": 0.6097,
+      "step": 4560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.286781650619437e-05,
+      "loss": 0.712,
+      "step": 4561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2847036478633617e-05,
+      "loss": 0.6429,
+      "step": 4562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.282625980844806e-05,
+      "loss": 0.6542,
+      "step": 4563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2805486499704354e-05,
+      "loss": 0.6472,
+      "step": 4564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.278471655646847e-05,
+      "loss": 0.3928,
+      "step": 4565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.276394998280574e-05,
+      "loss": 0.6324,
+      "step": 4566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.274318678278085e-05,
+      "loss": 0.7163,
+      "step": 4567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2722426960457816e-05,
+      "loss": 0.5937,
+      "step": 4568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.270167051989997e-05,
+      "loss": 0.617,
+      "step": 4569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.268091746517003e-05,
+      "loss": 0.6492,
+      "step": 4570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.266016780032999e-05,
+      "loss": 0.6535,
+      "step": 4571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2639421529441225e-05,
+      "loss": 0.6377,
+      "step": 4572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2618678656564436e-05,
+      "loss": 0.6271,
+      "step": 4573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.259793918575969e-05,
+      "loss": 0.6187,
+      "step": 4574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2577203121086265e-05,
+      "loss": 0.64,
+      "step": 4575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.255647046660296e-05,
+      "loss": 0.3761,
+      "step": 4576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.253574122636775e-05,
+      "loss": 0.7206,
+      "step": 4577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.251501540443802e-05,
+      "loss": 0.6542,
+      "step": 4578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.249429300487046e-05,
+      "loss": 0.6366,
+      "step": 4579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2473574031721115e-05,
+      "loss": 0.6548,
+      "step": 4580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.24528584890453e-05,
+      "loss": 0.6296,
+      "step": 4581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.243214638089774e-05,
+      "loss": 0.6065,
+      "step": 4582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.241143771133243e-05,
+      "loss": 0.6463,
+      "step": 4583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.239073248440271e-05,
+      "loss": 0.6638,
+      "step": 4584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.237003070416122e-05,
+      "loss": 0.6589,
+      "step": 4585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.234933237466e-05,
+      "loss": 0.6028,
+      "step": 4586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.232863749995035e-05,
+      "loss": 0.5917,
+      "step": 4587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2307946084082876e-05,
+      "loss": 0.5996,
+      "step": 4588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.228725813110757e-05,
+      "loss": 0.6294,
+      "step": 4589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.22665736450737e-05,
+      "loss": 0.3829,
+      "step": 4590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.22458926300299e-05,
+      "loss": 0.66,
+      "step": 4591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2225215090024056e-05,
+      "loss": 0.6313,
+      "step": 4592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.220454102910346e-05,
+      "loss": 0.6161,
+      "step": 4593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2183870451314624e-05,
+      "loss": 0.6022,
+      "step": 4594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.216320336070348e-05,
+      "loss": 0.6731,
+      "step": 4595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2142539761315184e-05,
+      "loss": 0.6589,
+      "step": 4596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.212187965719429e-05,
+      "loss": 0.6927,
+      "step": 4597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2101223052384597e-05,
+      "loss": 0.7106,
+      "step": 4598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2080569950929305e-05,
+      "loss": 0.588,
+      "step": 4599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.205992035687082e-05,
+      "loss": 0.6538,
+      "step": 4600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.203927427425095e-05,
+      "loss": 0.6121,
+      "step": 4601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.201863170711077e-05,
+      "loss": 0.6073,
+      "step": 4602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.19979926594907e-05,
+      "loss": 0.6782,
+      "step": 4603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.197735713543041e-05,
+      "loss": 0.6448,
+      "step": 4604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.195672513896898e-05,
+      "loss": 0.589,
+      "step": 4605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.193609667414469e-05,
+      "loss": 0.6551,
+      "step": 4606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1915471744995215e-05,
+      "loss": 0.6872,
+      "step": 4607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.189485035555748e-05,
+      "loss": 0.6557,
+      "step": 4608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1874232509867764e-05,
+      "loss": 0.6231,
+      "step": 4609
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.18536182119616e-05,
+      "loss": 0.6113,
+      "step": 4610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.183300746587388e-05,
+      "loss": 0.5841,
+      "step": 4611
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.181240027563875e-05,
+      "loss": 0.6649,
+      "step": 4612
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1791796645289715e-05,
+      "loss": 0.6287,
+      "step": 4613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.177119657885954e-05,
+      "loss": 0.6921,
+      "step": 4614
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1750600080380324e-05,
+      "loss": 0.6246,
+      "step": 4615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.173000715388341e-05,
+      "loss": 0.6181,
+      "step": 4616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.170941780339952e-05,
+      "loss": 0.6276,
+      "step": 4617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1688832032958616e-05,
+      "loss": 0.6778,
+      "step": 4618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.166824984659001e-05,
+      "loss": 0.6709,
+      "step": 4619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.164767124832224e-05,
+      "loss": 0.5687,
+      "step": 4620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1627096242183243e-05,
+      "loss": 0.6679,
+      "step": 4621
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1606524832200135e-05,
+      "loss": 0.6311,
+      "step": 4622
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.158595702239943e-05,
+      "loss": 0.649,
+      "step": 4623
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1565392816806874e-05,
+      "loss": 0.6374,
+      "step": 4624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.154483221944756e-05,
+      "loss": 0.6467,
+      "step": 4625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1524275234345804e-05,
+      "loss": 0.6223,
+      "step": 4626
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1503721865525295e-05,
+      "loss": 0.6441,
+      "step": 4627
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.148317211700893e-05,
+      "loss": 0.614,
+      "step": 4628
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.146262599281898e-05,
+      "loss": 0.6259,
+      "step": 4629
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.144208349697696e-05,
+      "loss": 0.5698,
+      "step": 4630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1421544633503684e-05,
+      "loss": 0.6468,
+      "step": 4631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.140100940641923e-05,
+      "loss": 0.7317,
+      "step": 4632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.138047781974303e-05,
+      "loss": 0.6529,
+      "step": 4633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.135994987749372e-05,
+      "loss": 0.6511,
+      "step": 4634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1339425583689295e-05,
+      "loss": 0.3959,
+      "step": 4635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1318904942347e-05,
+      "loss": 0.6244,
+      "step": 4636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.129838795748336e-05,
+      "loss": 0.6983,
+      "step": 4637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.127787463311421e-05,
+      "loss": 0.7295,
+      "step": 4638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.125736497325464e-05,
+      "loss": 0.6405,
+      "step": 4639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.123685898191904e-05,
+      "loss": 0.6584,
+      "step": 4640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.121635666312108e-05,
+      "loss": 0.626,
+      "step": 4641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.11958580208737e-05,
+      "loss": 0.6708,
+      "step": 4642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.117536305918916e-05,
+      "loss": 0.6375,
+      "step": 4643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.115487178207894e-05,
+      "loss": 0.5907,
+      "step": 4644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1134384193553846e-05,
+      "loss": 0.6442,
+      "step": 4645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.111390029762392e-05,
+      "loss": 0.5965,
+      "step": 4646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1093420098298544e-05,
+      "loss": 0.5582,
+      "step": 4647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.107294359958628e-05,
+      "loss": 0.6953,
+      "step": 4648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1052470805495094e-05,
+      "loss": 0.6769,
+      "step": 4649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.10320017200321e-05,
+      "loss": 0.6678,
+      "step": 4650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.101153634720376e-05,
+      "loss": 0.5853,
+      "step": 4651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.099107469101581e-05,
+      "loss": 0.6096,
+      "step": 4652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0970616755473214e-05,
+      "loss": 0.6439,
+      "step": 4653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0950162544580274e-05,
+      "loss": 0.6684,
+      "step": 4654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.092971206234047e-05,
+      "loss": 0.645,
+      "step": 4655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.090926531275665e-05,
+      "loss": 0.6787,
+      "step": 4656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0888822299830854e-05,
+      "loss": 0.6238,
+      "step": 4657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.086838302756445e-05,
+      "loss": 0.5918,
+      "step": 4658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.084794749995802e-05,
+      "loss": 0.6295,
+      "step": 4659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.082751572101148e-05,
+      "loss": 0.6326,
+      "step": 4660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0807087694723924e-05,
+      "loss": 0.6204,
+      "step": 4661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0786663425093805e-05,
+      "loss": 0.5956,
+      "step": 4662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.076624291611876e-05,
+      "loss": 0.6065,
+      "step": 4663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.074582617179574e-05,
+      "loss": 0.646,
+      "step": 4664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0725413196120956e-05,
+      "loss": 0.6282,
+      "step": 4665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.070500399308987e-05,
+      "loss": 0.7059,
+      "step": 4666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.068459856669718e-05,
+      "loss": 0.6523,
+      "step": 4667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0664196920936896e-05,
+      "loss": 0.6418,
+      "step": 4668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.064379905980224e-05,
+      "loss": 0.7183,
+      "step": 4669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.062340498728573e-05,
+      "loss": 0.6886,
+      "step": 4670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.060301470737913e-05,
+      "loss": 0.6208,
+      "step": 4671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0582628224073483e-05,
+      "loss": 0.6302,
+      "step": 4672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.056224554135901e-05,
+      "loss": 0.6982,
+      "step": 4673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0541866663225293e-05,
+      "loss": 0.7017,
+      "step": 4674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.052149159366109e-05,
+      "loss": 0.5503,
+      "step": 4675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0501120336654464e-05,
+      "loss": 0.6755,
+      "step": 4676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.048075289619269e-05,
+      "loss": 0.6843,
+      "step": 4677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0460389276262348e-05,
+      "loss": 0.6877,
+      "step": 4678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.044002948084921e-05,
+      "loss": 0.7037,
+      "step": 4679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.041967351393834e-05,
+      "loss": 0.6415,
+      "step": 4680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0399321379514047e-05,
+      "loss": 0.6698,
+      "step": 4681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0378973081559892e-05,
+      "loss": 0.6159,
+      "step": 4682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.035862862405865e-05,
+      "loss": 0.6548,
+      "step": 4683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0338288010992398e-05,
+      "loss": 0.6113,
+      "step": 4684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0317951246342425e-05,
+      "loss": 0.7454,
+      "step": 4685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.029761833408927e-05,
+      "loss": 0.7165,
+      "step": 4686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.027728927821274e-05,
+      "loss": 0.6241,
+      "step": 4687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.025696408269187e-05,
+      "loss": 0.5411,
+      "step": 4688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0236642751504916e-05,
+      "loss": 0.7052,
+      "step": 4689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0216325288629434e-05,
+      "loss": 0.6276,
+      "step": 4690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.019601169804216e-05,
+      "loss": 0.6401,
+      "step": 4691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.017570198371914e-05,
+      "loss": 0.6179,
+      "step": 4692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0155396149635585e-05,
+      "loss": 0.5454,
+      "step": 4693
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0135094199766035e-05,
+      "loss": 0.6541,
+      "step": 4694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0114796138084172e-05,
+      "loss": 0.6034,
+      "step": 4695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0094501968562992e-05,
+      "loss": 0.6051,
+      "step": 4696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0074211695174696e-05,
+      "loss": 0.5791,
+      "step": 4697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0053925321890754e-05,
+      "loss": 0.634,
+      "step": 4698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0033642852681788e-05,
+      "loss": 0.6624,
+      "step": 4699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0013364291517788e-05,
+      "loss": 0.6517,
+      "step": 4700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9993089642367845e-05,
+      "loss": 0.5694,
+      "step": 4701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9972818909200396e-05,
+      "loss": 0.5949,
+      "step": 4702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9952552095983023e-05,
+      "loss": 0.63,
+      "step": 4703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9932289206682617e-05,
+      "loss": 0.5826,
+      "step": 4704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.991203024526522e-05,
+      "loss": 0.6081,
+      "step": 4705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9891775215696194e-05,
+      "loss": 0.6281,
+      "step": 4706
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.987152412194004e-05,
+      "loss": 0.3761,
+      "step": 4707
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9851276967960583e-05,
+      "loss": 0.6682,
+      "step": 4708
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9831033757720783e-05,
+      "loss": 0.6871,
+      "step": 4709
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9810794495182925e-05,
+      "loss": 0.6703,
+      "step": 4710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.979055918430842e-05,
+      "loss": 0.6449,
+      "step": 4711
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9770327829057986e-05,
+      "loss": 0.6259,
+      "step": 4712
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9750100433391514e-05,
+      "loss": 0.7241,
+      "step": 4713
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.972987700126817e-05,
+      "loss": 0.5938,
+      "step": 4714
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9709657536646286e-05,
+      "loss": 0.6028,
+      "step": 4715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.968944204348347e-05,
+      "loss": 0.6791,
+      "step": 4716
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9669230525736534e-05,
+      "loss": 0.6218,
+      "step": 4717
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9649022987361492e-05,
+      "loss": 0.6601,
+      "step": 4718
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.962881943231361e-05,
+      "loss": 0.6369,
+      "step": 4719
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9608619864547337e-05,
+      "loss": 0.6447,
+      "step": 4720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9588424288016398e-05,
+      "loss": 0.73,
+      "step": 4721
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9568232706673672e-05,
+      "loss": 0.3729,
+      "step": 4722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9548045124471324e-05,
+      "loss": 0.6777,
+      "step": 4723
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9527861545360658e-05,
+      "loss": 0.6748,
+      "step": 4724
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.950768197329226e-05,
+      "loss": 0.6905,
+      "step": 4725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9487506412215894e-05,
+      "loss": 0.6902,
+      "step": 4726
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.946733486608059e-05,
+      "loss": 0.6285,
+      "step": 4727
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9447167338834485e-05,
+      "loss": 0.3813,
+      "step": 4728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9427003834425075e-05,
+      "loss": 0.6823,
+      "step": 4729
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9406844356798946e-05,
+      "loss": 0.6474,
+      "step": 4730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9386688909901953e-05,
+      "loss": 0.6619,
+      "step": 4731
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9366537497679153e-05,
+      "loss": 0.6498,
+      "step": 4732
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.934639012407483e-05,
+      "loss": 0.6642,
+      "step": 4733
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9326246793032418e-05,
+      "loss": 0.6383,
+      "step": 4734
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.930610750849465e-05,
+      "loss": 0.6559,
+      "step": 4735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.928597227440337e-05,
+      "loss": 0.6251,
+      "step": 4736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.926584109469972e-05,
+      "loss": 0.5567,
+      "step": 4737
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.924571397332398e-05,
+      "loss": 0.6833,
+      "step": 4738
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9225590914215683e-05,
+      "loss": 0.6515,
+      "step": 4739
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9205471921313516e-05,
+      "loss": 0.5995,
+      "step": 4740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9185356998555435e-05,
+      "loss": 0.6287,
+      "step": 4741
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.91652461498785e-05,
+      "loss": 0.6068,
+      "step": 4742
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.914513937921912e-05,
+      "loss": 0.6418,
+      "step": 4743
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.912503669051277e-05,
+      "loss": 0.637,
+      "step": 4744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9104938087694213e-05,
+      "loss": 0.6166,
+      "step": 4745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.908484357469734e-05,
+      "loss": 0.356,
+      "step": 4746
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.906475315545532e-05,
+      "loss": 0.3243,
+      "step": 4747
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9044666833900437e-05,
+      "loss": 0.6134,
+      "step": 4748
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9024584613964245e-05,
+      "loss": 0.6848,
+      "step": 4749
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9004506499577467e-05,
+      "loss": 0.6084,
+      "step": 4750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8984432494670027e-05,
+      "loss": 0.6877,
+      "step": 4751
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8964362603171013e-05,
+      "loss": 0.661,
+      "step": 4752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8944296829008776e-05,
+      "loss": 0.6057,
+      "step": 4753
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8924235176110776e-05,
+      "loss": 0.6855,
+      "step": 4754
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.890417764840373e-05,
+      "loss": 0.6339,
+      "step": 4755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8884124249813526e-05,
+      "loss": 0.7365,
+      "step": 4756
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8864074984265277e-05,
+      "loss": 0.6955,
+      "step": 4757
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8844029855683196e-05,
+      "loss": 0.6576,
+      "step": 4758
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8823988867990803e-05,
+      "loss": 0.5779,
+      "step": 4759
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8803952025110704e-05,
+      "loss": 0.632,
+      "step": 4760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8783919330964764e-05,
+      "loss": 0.6559,
+      "step": 4761
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8763890789474003e-05,
+      "loss": 0.6669,
+      "step": 4762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8743866404558668e-05,
+      "loss": 0.64,
+      "step": 4763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8723846180138114e-05,
+      "loss": 0.5878,
+      "step": 4764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8703830120130976e-05,
+      "loss": 0.6806,
+      "step": 4765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8683818228454985e-05,
+      "loss": 0.6213,
+      "step": 4766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.866381050902711e-05,
+      "loss": 0.5939,
+      "step": 4767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8643806965763502e-05,
+      "loss": 0.592,
+      "step": 4768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.86238076025795e-05,
+      "loss": 0.6781,
+      "step": 4769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8603812423389576e-05,
+      "loss": 0.6542,
+      "step": 4770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.858382143210745e-05,
+      "loss": 0.6761,
+      "step": 4771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8563834632645946e-05,
+      "loss": 0.633,
+      "step": 4772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.854385202891714e-05,
+      "loss": 0.4111,
+      "step": 4773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8523873624832247e-05,
+      "loss": 0.6071,
+      "step": 4774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8503899424301694e-05,
+      "loss": 0.6826,
+      "step": 4775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8483929431235024e-05,
+      "loss": 0.6433,
+      "step": 4776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8463963649541025e-05,
+      "loss": 0.6598,
+      "step": 4777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8444002083127596e-05,
+      "loss": 0.6379,
+      "step": 4778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8424044735901867e-05,
+      "loss": 0.6673,
+      "step": 4779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8404091611770112e-05,
+      "loss": 0.595,
+      "step": 4780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.838414271463779e-05,
+      "loss": 0.6366,
+      "step": 4781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8364198048409552e-05,
+      "loss": 0.6975,
+      "step": 4782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8344257616989144e-05,
+      "loss": 0.5919,
+      "step": 4783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8324321424279593e-05,
+      "loss": 0.6398,
+      "step": 4784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8304389474182992e-05,
+      "loss": 0.6046,
+      "step": 4785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.828446177060067e-05,
+      "loss": 0.5889,
+      "step": 4786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8264538317433103e-05,
+      "loss": 0.5793,
+      "step": 4787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8244619118579967e-05,
+      "loss": 0.686,
+      "step": 4788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.822470417794003e-05,
+      "loss": 0.6282,
+      "step": 4789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.820479349941132e-05,
+      "loss": 0.6704,
+      "step": 4790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8184887086890943e-05,
+      "loss": 0.6102,
+      "step": 4791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8164984944275242e-05,
+      "loss": 0.7253,
+      "step": 4792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.814508707545964e-05,
+      "loss": 0.3792,
+      "step": 4793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.812519348433885e-05,
+      "loss": 0.5977,
+      "step": 4794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8105304174806622e-05,
+      "loss": 0.6271,
+      "step": 4795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8085419150755967e-05,
+      "loss": 0.6483,
+      "step": 4796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.806553841607895e-05,
+      "loss": 0.6622,
+      "step": 4797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8045661974666916e-05,
+      "loss": 0.6233,
+      "step": 4798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8025789830410266e-05,
+      "loss": 0.5776,
+      "step": 4799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8005921987198614e-05,
+      "loss": 0.6787,
+      "step": 4800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7986058448920728e-05,
+      "loss": 0.6858,
+      "step": 4801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.796619921946455e-05,
+      "loss": 0.5516,
+      "step": 4802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7946344302717125e-05,
+      "loss": 0.6863,
+      "step": 4803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7926493702564714e-05,
+      "loss": 0.6588,
+      "step": 4804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.790664742289266e-05,
+      "loss": 0.6276,
+      "step": 4805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.788680546758554e-05,
+      "loss": 0.6345,
+      "step": 4806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.786696784052705e-05,
+      "loss": 0.6582,
+      "step": 4807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.784713454560004e-05,
+      "loss": 0.6492,
+      "step": 4808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7827305586686486e-05,
+      "loss": 0.6705,
+      "step": 4809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7807480967667576e-05,
+      "loss": 0.6964,
+      "step": 4810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7787660692423568e-05,
+      "loss": 0.3623,
+      "step": 4811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.776784476483394e-05,
+      "loss": 0.5778,
+      "step": 4812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.774803318877729e-05,
+      "loss": 0.7483,
+      "step": 4813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7728225968131383e-05,
+      "loss": 0.6578,
+      "step": 4814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7708423106773084e-05,
+      "loss": 0.6186,
+      "step": 4815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.768862460857847e-05,
+      "loss": 0.717,
+      "step": 4816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.76688304774227e-05,
+      "loss": 0.67,
+      "step": 4817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7649040717180132e-05,
+      "loss": 0.6052,
+      "step": 4818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7629255331724225e-05,
+      "loss": 0.6022,
+      "step": 4819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.760947432492763e-05,
+      "loss": 0.6064,
+      "step": 4820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7589697700662086e-05,
+      "loss": 0.6969,
+      "step": 4821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7569925462798528e-05,
+      "loss": 0.5918,
+      "step": 4822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7550157615206974e-05,
+      "loss": 0.575,
+      "step": 4823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.753039416175662e-05,
+      "loss": 0.646,
+      "step": 4824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7510635106315818e-05,
+      "loss": 0.3964,
+      "step": 4825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7490880452752033e-05,
+      "loss": 0.618,
+      "step": 4826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7471130204931857e-05,
+      "loss": 0.6929,
+      "step": 4827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7451384366721057e-05,
+      "loss": 0.5864,
+      "step": 4828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7431642941984482e-05,
+      "loss": 0.6377,
+      "step": 4829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.741190593458618e-05,
+      "loss": 0.6371,
+      "step": 4830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7392173348389284e-05,
+      "loss": 0.5988,
+      "step": 4831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7372445187256124e-05,
+      "loss": 0.6075,
+      "step": 4832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.735272145504807e-05,
+      "loss": 0.6148,
+      "step": 4833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.733300215562572e-05,
+      "loss": 0.6678,
+      "step": 4834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7313287292848722e-05,
+      "loss": 0.643,
+      "step": 4835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7293576870575933e-05,
+      "loss": 0.6617,
+      "step": 4836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7273870892665253e-05,
+      "loss": 0.6087,
+      "step": 4837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7254169362973838e-05,
+      "loss": 0.5806,
+      "step": 4838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7234472285357826e-05,
+      "loss": 0.6154,
+      "step": 4839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7214779663672613e-05,
+      "loss": 0.6232,
+      "step": 4840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.719509150177262e-05,
+      "loss": 0.6251,
+      "step": 4841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7175407803511466e-05,
+      "loss": 0.6604,
+      "step": 4842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7155728572741834e-05,
+      "loss": 0.6822,
+      "step": 4843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7136053813315633e-05,
+      "loss": 0.6877,
+      "step": 4844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.711638352908378e-05,
+      "loss": 0.6075,
+      "step": 4845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.709671772389639e-05,
+      "loss": 0.5878,
+      "step": 4846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7077056401602696e-05,
+      "loss": 0.6573,
+      "step": 4847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7057399566050994e-05,
+      "loss": 0.6939,
+      "step": 4848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7037747221088795e-05,
+      "loss": 0.6686,
+      "step": 4849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7018099370562622e-05,
+      "loss": 0.5998,
+      "step": 4850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6998456018318252e-05,
+      "loss": 0.6572,
+      "step": 4851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6978817168200448e-05,
+      "loss": 0.6719,
+      "step": 4852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6959182824053196e-05,
+      "loss": 0.6356,
+      "step": 4853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6939552989719514e-05,
+      "loss": 0.6016,
+      "step": 4854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.691992766904161e-05,
+      "loss": 0.6355,
+      "step": 4855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.690030686586076e-05,
+      "loss": 0.6251,
+      "step": 4856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6880690584017366e-05,
+      "loss": 0.6662,
+      "step": 4857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.686107882735097e-05,
+      "loss": 0.6461,
+      "step": 4858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6841471599700226e-05,
+      "loss": 0.683,
+      "step": 4859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.682186890490285e-05,
+      "loss": 0.6335,
+      "step": 4860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6802270746795733e-05,
+      "loss": 0.608,
+      "step": 4861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6782677129214816e-05,
+      "loss": 0.6239,
+      "step": 4862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6763088055995222e-05,
+      "loss": 0.6613,
+      "step": 4863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6743503530971138e-05,
+      "loss": 0.6979,
+      "step": 4864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.672392355797589e-05,
+      "loss": 0.6711,
+      "step": 4865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6704348140841863e-05,
+      "loss": 0.5962,
+      "step": 4866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6684777283400615e-05,
+      "loss": 0.5781,
+      "step": 4867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6665210989482748e-05,
+      "loss": 0.6146,
+      "step": 4868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.664564926291801e-05,
+      "loss": 0.6368,
+      "step": 4869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6626092107535255e-05,
+      "loss": 0.699,
+      "step": 4870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6606539527162454e-05,
+      "loss": 0.6273,
+      "step": 4871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6586991525626624e-05,
+      "loss": 0.6878,
+      "step": 4872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.656744810675397e-05,
+      "loss": 0.6844,
+      "step": 4873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.654790927436969e-05,
+      "loss": 0.6333,
+      "step": 4874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6528375032298196e-05,
+      "loss": 0.5861,
+      "step": 4875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6508845384362947e-05,
+      "loss": 0.6234,
+      "step": 4876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.648932033438652e-05,
+      "loss": 0.6508,
+      "step": 4877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6469799886190548e-05,
+      "loss": 0.5887,
+      "step": 4878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6450284043595853e-05,
+      "loss": 0.6814,
+      "step": 4879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6430772810422233e-05,
+      "loss": 0.6049,
+      "step": 4880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6411266190488693e-05,
+      "loss": 0.6359,
+      "step": 4881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6391764187613277e-05,
+      "loss": 0.6124,
+      "step": 4882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6372266805613167e-05,
+      "loss": 0.611,
+      "step": 4883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6352774048304586e-05,
+      "loss": 0.6078,
+      "step": 4884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6333285919502904e-05,
+      "loss": 0.6293,
+      "step": 4885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6313802423022526e-05,
+      "loss": 0.668,
+      "step": 4886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.629432356267701e-05,
+      "loss": 0.6792,
+      "step": 4887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.627484934227898e-05,
+      "loss": 0.6238,
+      "step": 4888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6255379765640174e-05,
+      "loss": 0.6295,
+      "step": 4889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.623591483657136e-05,
+      "loss": 0.7022,
+      "step": 4890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6216454558882486e-05,
+      "loss": 0.665,
+      "step": 4891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.619699893638249e-05,
+      "loss": 0.6607,
+      "step": 4892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6177547972879495e-05,
+      "loss": 0.6194,
+      "step": 4893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6158101672180612e-05,
+      "loss": 0.631,
+      "step": 4894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6138660038092168e-05,
+      "loss": 0.664,
+      "step": 4895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6119223074419442e-05,
+      "loss": 0.6239,
+      "step": 4896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.60997907849669e-05,
+      "loss": 0.7039,
+      "step": 4897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.608036317353802e-05,
+      "loss": 0.6511,
+      "step": 4898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.606094024393543e-05,
+      "loss": 0.6375,
+      "step": 4899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.604152199996077e-05,
+      "loss": 0.6069,
+      "step": 4900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6022108445414835e-05,
+      "loss": 0.6419,
+      "step": 4901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6002699584097452e-05,
+      "loss": 0.6194,
+      "step": 4902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.598329541980757e-05,
+      "loss": 0.6321,
+      "step": 4903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.596389595634316e-05,
+      "loss": 0.6018,
+      "step": 4904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.594450119750135e-05,
+      "loss": 0.6195,
+      "step": 4905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5925111147078252e-05,
+      "loss": 0.7202,
+      "step": 4906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5905725808869143e-05,
+      "loss": 0.6833,
+      "step": 4907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5886345186668338e-05,
+      "loss": 0.6298,
+      "step": 4908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.586696928426926e-05,
+      "loss": 0.5868,
+      "step": 4909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5847598105464337e-05,
+      "loss": 0.6428,
+      "step": 4910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5828231654045143e-05,
+      "loss": 0.6658,
+      "step": 4911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.580886993380232e-05,
+      "loss": 0.619,
+      "step": 4912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.578951294852553e-05,
+      "loss": 0.6588,
+      "step": 4913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5770160702003566e-05,
+      "loss": 0.6837,
+      "step": 4914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5750813198024258e-05,
+      "loss": 0.6436,
+      "step": 4915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.573147044037456e-05,
+      "loss": 0.6254,
+      "step": 4916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5712132432840407e-05,
+      "loss": 0.667,
+      "step": 4917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5692799179206906e-05,
+      "loss": 0.6314,
+      "step": 4918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.567347068325814e-05,
+      "loss": 0.5919,
+      "step": 4919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5654146948777313e-05,
+      "loss": 0.6428,
+      "step": 4920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5634827979546706e-05,
+      "loss": 0.6513,
+      "step": 4921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.561551377934766e-05,
+      "loss": 0.6197,
+      "step": 4922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5596204351960535e-05,
+      "loss": 0.6446,
+      "step": 4923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5576899701164824e-05,
+      "loss": 0.631,
+      "step": 4924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5557599830739037e-05,
+      "loss": 0.6151,
+      "step": 4925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5538304744460767e-05,
+      "loss": 0.36,
+      "step": 4926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5519014446106682e-05,
+      "loss": 0.6828,
+      "step": 4927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5499728939452505e-05,
+      "loss": 0.5631,
+      "step": 4928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5480448228272996e-05,
+      "loss": 0.6334,
+      "step": 4929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.546117231634203e-05,
+      "loss": 0.6235,
+      "step": 4930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.544190120743246e-05,
+      "loss": 0.5869,
+      "step": 4931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.542263490531629e-05,
+      "loss": 0.6496,
+      "step": 4932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5403373413764527e-05,
+      "loss": 0.6075,
+      "step": 4933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.538411673654727e-05,
+      "loss": 0.3827,
+      "step": 4934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5364864877433624e-05,
+      "loss": 0.6645,
+      "step": 4935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5345617840191828e-05,
+      "loss": 0.6157,
+      "step": 4936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5326375628589083e-05,
+      "loss": 0.5805,
+      "step": 4937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5307138246391714e-05,
+      "loss": 0.6381,
+      "step": 4938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.52879056973651e-05,
+      "loss": 0.6268,
+      "step": 4939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.526867798527366e-05,
+      "loss": 0.7166,
+      "step": 4940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5249455113880828e-05,
+      "loss": 0.6359,
+      "step": 4941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.523023708694917e-05,
+      "loss": 0.6372,
+      "step": 4942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.521102390824021e-05,
+      "loss": 0.5954,
+      "step": 4943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5191815581514623e-05,
+      "loss": 0.6715,
+      "step": 4944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5172612110532012e-05,
+      "loss": 0.6635,
+      "step": 4945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5153413499051187e-05,
+      "loss": 0.6146,
+      "step": 4946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.513421975082986e-05,
+      "loss": 0.5644,
+      "step": 4947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5115030869624896e-05,
+      "loss": 0.6131,
+      "step": 4948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.509584685919212e-05,
+      "loss": 0.6664,
+      "step": 4949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5076667723286485e-05,
+      "loss": 0.6773,
+      "step": 4950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.505749346566192e-05,
+      "loss": 0.6383,
+      "step": 4951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5038324090071437e-05,
+      "loss": 0.6403,
+      "step": 4952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5019159600267093e-05,
+      "loss": 0.5989,
+      "step": 4953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.500000000000001e-05,
+      "loss": 0.6396,
+      "step": 4954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4980845293020277e-05,
+      "loss": 0.6605,
+      "step": 4955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4961695483077118e-05,
+      "loss": 0.6277,
+      "step": 4956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.494255057391871e-05,
+      "loss": 0.6263,
+      "step": 4957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4923410569292338e-05,
+      "loss": 0.5784,
+      "step": 4958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4904275472944298e-05,
+      "loss": 0.6468,
+      "step": 4959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4885145288619954e-05,
+      "loss": 0.6214,
+      "step": 4960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4866020020063653e-05,
+      "loss": 0.6632,
+      "step": 4961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.484689967101884e-05,
+      "loss": 0.6489,
+      "step": 4962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4827784245227937e-05,
+      "loss": 0.5888,
+      "step": 4963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4808673746432455e-05,
+      "loss": 0.6021,
+      "step": 4964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4789568178372917e-05,
+      "loss": 0.6805,
+      "step": 4965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4770467544788907e-05,
+      "loss": 0.6312,
+      "step": 4966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.475137184941898e-05,
+      "loss": 0.6742,
+      "step": 4967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.47322810960008e-05,
+      "loss": 0.6891,
+      "step": 4968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4713195288271003e-05,
+      "loss": 0.6154,
+      "step": 4969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4694114429965287e-05,
+      "loss": 0.6385,
+      "step": 4970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4675038524818383e-05,
+      "loss": 0.6538,
+      "step": 4971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4655967576564064e-05,
+      "loss": 0.6584,
+      "step": 4972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4636901588935084e-05,
+      "loss": 0.6301,
+      "step": 4973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4617840565663286e-05,
+      "loss": 0.3841,
+      "step": 4974
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4598784510479478e-05,
+      "loss": 0.5337,
+      "step": 4975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4579733427113556e-05,
+      "loss": 0.6624,
+      "step": 4976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.45606873192944e-05,
+      "loss": 0.6839,
+      "step": 4977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.454164619074995e-05,
+      "loss": 0.5994,
+      "step": 4978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4522610045207167e-05,
+      "loss": 0.6993,
+      "step": 4979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4503578886391987e-05,
+      "loss": 0.6193,
+      "step": 4980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4484552718029448e-05,
+      "loss": 0.6882,
+      "step": 4981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4465531543843524e-05,
+      "loss": 0.6916,
+      "step": 4982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.444651536755728e-05,
+      "loss": 0.6738,
+      "step": 4983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.442750419289279e-05,
+      "loss": 0.6396,
+      "step": 4984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4408498023571148e-05,
+      "loss": 0.631,
+      "step": 4985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4389496863312422e-05,
+      "loss": 0.6578,
+      "step": 4986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.437050071583578e-05,
+      "loss": 0.5809,
+      "step": 4987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4351509584859333e-05,
+      "loss": 0.6889,
+      "step": 4988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.433252347410025e-05,
+      "loss": 0.7004,
+      "step": 4989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4313542387274717e-05,
+      "loss": 0.6162,
+      "step": 4990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4294566328097958e-05,
+      "loss": 0.6003,
+      "step": 4991
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4275595300284132e-05,
+      "loss": 0.6706,
+      "step": 4992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4256629307546513e-05,
+      "loss": 0.6506,
+      "step": 4993
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4237668353597305e-05,
+      "loss": 0.6143,
+      "step": 4994
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4218712442147795e-05,
+      "loss": 0.5713,
+      "step": 4995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.419976157690821e-05,
+      "loss": 0.6184,
+      "step": 4996
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4180815761587894e-05,
+      "loss": 0.6282,
+      "step": 4997
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4161874999895085e-05,
+      "loss": 0.6646,
+      "step": 4998
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4142939295537126e-05,
+      "loss": 0.6537,
+      "step": 4999
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.41240086522203e-05,
+      "loss": 0.6078,
+      "step": 5000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4105083073649953e-05,
+      "loss": 0.6617,
+      "step": 5001
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.408616256353039e-05,
+      "loss": 0.6709,
+      "step": 5002
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4067247125564968e-05,
+      "loss": 0.6965,
+      "step": 5003
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.404833676345603e-05,
+      "loss": 0.67,
+      "step": 5004
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.402943148090495e-05,
+      "loss": 0.6929,
+      "step": 5005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4010531281612053e-05,
+      "loss": 0.6532,
+      "step": 5006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3991636169276737e-05,
+      "loss": 0.6416,
+      "step": 5007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3972746147597335e-05,
+      "loss": 0.6074,
+      "step": 5008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.395386122027124e-05,
+      "loss": 0.637,
+      "step": 5009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.393498139099482e-05,
+      "loss": 0.6686,
+      "step": 5010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.391610666346348e-05,
+      "loss": 0.7051,
+      "step": 5011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3897237041371552e-05,
+      "loss": 0.6282,
+      "step": 5012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3878372528412456e-05,
+      "loss": 0.6207,
+      "step": 5013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.385951312827854e-05,
+      "loss": 0.5908,
+      "step": 5014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.384065884466119e-05,
+      "loss": 0.6559,
+      "step": 5015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3821809681250796e-05,
+      "loss": 0.6291,
+      "step": 5016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.380296564173674e-05,
+      "loss": 0.6456,
+      "step": 5017
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3784126729807364e-05,
+      "loss": 0.5937,
+      "step": 5018
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3765292949150076e-05,
+      "loss": 0.6712,
+      "step": 5019
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3746464303451195e-05,
+      "loss": 0.6778,
+      "step": 5020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.37276407963961e-05,
+      "loss": 0.6557,
+      "step": 5021
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3708822431669148e-05,
+      "loss": 0.4001,
+      "step": 5022
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.36900092129537e-05,
+      "loss": 0.3634,
+      "step": 5023
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3671201143932064e-05,
+      "loss": 0.6026,
+      "step": 5024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3652398228285603e-05,
+      "loss": 0.6225,
+      "step": 5025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3633600469694606e-05,
+      "loss": 0.6558,
+      "step": 5026
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.36148078718384e-05,
+      "loss": 0.6196,
+      "step": 5027
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.35960204383953e-05,
+      "loss": 0.6344,
+      "step": 5028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3577238173042603e-05,
+      "loss": 0.634,
+      "step": 5029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.355846107945656e-05,
+      "loss": 0.5592,
+      "step": 5030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.353968916131248e-05,
+      "loss": 0.6474,
+      "step": 5031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.352092242228457e-05,
+      "loss": 0.6247,
+      "step": 5032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3502160866046103e-05,
+      "loss": 0.5907,
+      "step": 5033
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3483404496269296e-05,
+      "loss": 0.6673,
+      "step": 5034
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3464653316625385e-05,
+      "loss": 0.5826,
+      "step": 5035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3445907330784534e-05,
+      "loss": 0.6278,
+      "step": 5036
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.342716654241595e-05,
+      "loss": 0.5992,
+      "step": 5037
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3408430955187765e-05,
+      "loss": 0.3875,
+      "step": 5038
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.338970057276716e-05,
+      "loss": 0.6405,
+      "step": 5039
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3370975398820207e-05,
+      "loss": 0.6362,
+      "step": 5040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3352255437012053e-05,
+      "loss": 0.6095,
+      "step": 5041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.333354069100681e-05,
+      "loss": 0.6614,
+      "step": 5042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3314831164467487e-05,
+      "loss": 0.5783,
+      "step": 5043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3296126861056174e-05,
+      "loss": 0.6047,
+      "step": 5044
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.327742778443384e-05,
+      "loss": 0.6104,
+      "step": 5045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3258733938260525e-05,
+      "loss": 0.6781,
+      "step": 5046
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3240045326195187e-05,
+      "loss": 0.6502,
+      "step": 5047
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3221361951895796e-05,
+      "loss": 0.5757,
+      "step": 5048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3202683819019244e-05,
+      "loss": 0.6417,
+      "step": 5049
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3184010931221467e-05,
+      "loss": 0.6074,
+      "step": 5050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3165343292157297e-05,
+      "loss": 0.5655,
+      "step": 5051
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.314668090548062e-05,
+      "loss": 0.6233,
+      "step": 5052
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3128023774844194e-05,
+      "loss": 0.6599,
+      "step": 5053
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3109371903899883e-05,
+      "loss": 0.6347,
+      "step": 5054
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3090725296298387e-05,
+      "loss": 0.6249,
+      "step": 5055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.307208395568948e-05,
+      "loss": 0.614,
+      "step": 5056
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3053447885721808e-05,
+      "loss": 0.6218,
+      "step": 5057
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.303481709004308e-05,
+      "loss": 0.6209,
+      "step": 5058
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.30161915722999e-05,
+      "loss": 0.67,
+      "step": 5059
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.299757133613788e-05,
+      "loss": 0.6911,
+      "step": 5060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.297895638520159e-05,
+      "loss": 0.6655,
+      "step": 5061
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2960346723134575e-05,
+      "loss": 0.6528,
+      "step": 5062
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.29417423535793e-05,
+      "loss": 0.5934,
+      "step": 5063
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.292314328017726e-05,
+      "loss": 0.6171,
+      "step": 5064
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2904549506568845e-05,
+      "loss": 0.5696,
+      "step": 5065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.288596103639346e-05,
+      "loss": 0.5493,
+      "step": 5066
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.286737787328946e-05,
+      "loss": 0.6528,
+      "step": 5067
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2848800020894155e-05,
+      "loss": 0.6361,
+      "step": 5068
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2830227482843804e-05,
+      "loss": 0.6222,
+      "step": 5069
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2811660262773653e-05,
+      "loss": 0.6498,
+      "step": 5070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2793098364317868e-05,
+      "loss": 0.6117,
+      "step": 5071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2774541791109612e-05,
+      "loss": 0.6624,
+      "step": 5072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2755990546780987e-05,
+      "loss": 0.6153,
+      "step": 5073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2737444634963077e-05,
+      "loss": 0.6862,
+      "step": 5074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2718904059285867e-05,
+      "loss": 0.6647,
+      "step": 5075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2700368823378366e-05,
+      "loss": 0.5855,
+      "step": 5076
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2681838930868464e-05,
+      "loss": 0.6157,
+      "step": 5077
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2663314385383066e-05,
+      "loss": 0.633,
+      "step": 5078
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2644795190548014e-05,
+      "loss": 0.6408,
+      "step": 5079
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2626281349988103e-05,
+      "loss": 0.622,
+      "step": 5080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2607772867327048e-05,
+      "loss": 0.6441,
+      "step": 5081
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.258926974618758e-05,
+      "loss": 0.5827,
+      "step": 5082
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.25707719901913e-05,
+      "loss": 0.3879,
+      "step": 5083
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.255227960295882e-05,
+      "loss": 0.564,
+      "step": 5084
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2533792588109682e-05,
+      "loss": 0.5539,
+      "step": 5085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2515310949262402e-05,
+      "loss": 0.3454,
+      "step": 5086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2496834690034378e-05,
+      "loss": 0.6145,
+      "step": 5087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.247836381404203e-05,
+      "loss": 0.6705,
+      "step": 5088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2459898324900662e-05,
+      "loss": 0.6266,
+      "step": 5089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.244143822622456e-05,
+      "loss": 0.6279,
+      "step": 5090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.242298352162695e-05,
+      "loss": 0.6074,
+      "step": 5091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2404534214720018e-05,
+      "loss": 0.6701,
+      "step": 5092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2386090309114843e-05,
+      "loss": 0.7099,
+      "step": 5093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.23676518084215e-05,
+      "loss": 0.6764,
+      "step": 5094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2349218716248965e-05,
+      "loss": 0.6402,
+      "step": 5095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2330791036205205e-05,
+      "loss": 0.6362,
+      "step": 5096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2312368771897035e-05,
+      "loss": 0.604,
+      "step": 5097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2293951926930356e-05,
+      "loss": 0.5895,
+      "step": 5098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2275540504909865e-05,
+      "loss": 0.673,
+      "step": 5099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2257134509439293e-05,
+      "loss": 0.5979,
+      "step": 5100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2238733944121237e-05,
+      "loss": 0.7078,
+      "step": 5101
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2220338812557306e-05,
+      "loss": 0.6599,
+      "step": 5102
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.220194911834797e-05,
+      "loss": 0.6152,
+      "step": 5103
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2183564865092683e-05,
+      "loss": 0.6596,
+      "step": 5104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.216518605638983e-05,
+      "loss": 0.6615,
+      "step": 5105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.214681269583674e-05,
+      "loss": 0.703,
+      "step": 5106
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2128444787029618e-05,
+      "loss": 0.5831,
+      "step": 5107
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.211008233356366e-05,
+      "loss": 0.6309,
+      "step": 5108
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2091725339033003e-05,
+      "loss": 0.624,
+      "step": 5109
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2073373807030645e-05,
+      "loss": 0.6479,
+      "step": 5110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2055027741148576e-05,
+      "loss": 0.6744,
+      "step": 5111
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.20366871449777e-05,
+      "loss": 0.6003,
+      "step": 5112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2018352022107873e-05,
+      "loss": 0.6168,
+      "step": 5113
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2000022376127816e-05,
+      "loss": 0.6396,
+      "step": 5114
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1981698210625257e-05,
+      "loss": 0.6206,
+      "step": 5115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1963379529186768e-05,
+      "loss": 0.6472,
+      "step": 5116
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.194506633539792e-05,
+      "loss": 0.5916,
+      "step": 5117
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1926758632843174e-05,
+      "loss": 0.6665,
+      "step": 5118
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1908456425105943e-05,
+      "loss": 0.6885,
+      "step": 5119
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1890159715768516e-05,
+      "loss": 0.6693,
+      "step": 5120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1871868508412162e-05,
+      "loss": 0.3516,
+      "step": 5121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.185358280661702e-05,
+      "loss": 0.629,
+      "step": 5122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.183530261396219e-05,
+      "loss": 0.6729,
+      "step": 5123
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.181702793402568e-05,
+      "loss": 0.6474,
+      "step": 5124
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1798758770384442e-05,
+      "loss": 0.6728,
+      "step": 5125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.178049512661429e-05,
+      "loss": 0.6624,
+      "step": 5126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.176223700629003e-05,
+      "loss": 0.6414,
+      "step": 5127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1743984412985313e-05,
+      "loss": 0.6324,
+      "step": 5128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1725737350272768e-05,
+      "loss": 0.637,
+      "step": 5129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1707495821723916e-05,
+      "loss": 0.5656,
+      "step": 5130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1689259830909225e-05,
+      "loss": 0.5976,
+      "step": 5131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1671029381398007e-05,
+      "loss": 0.6693,
+      "step": 5132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.165280447675857e-05,
+      "loss": 0.583,
+      "step": 5133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1634585120558078e-05,
+      "loss": 0.707,
+      "step": 5134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1616371316362638e-05,
+      "loss": 0.651,
+      "step": 5135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.159816306773727e-05,
+      "loss": 0.5926,
+      "step": 5136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.157996037824592e-05,
+      "loss": 0.6701,
+      "step": 5137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.15617632514514e-05,
+      "loss": 0.6576,
+      "step": 5138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.154357169091548e-05,
+      "loss": 0.6046,
+      "step": 5139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.15253857001988e-05,
+      "loss": 0.6609,
+      "step": 5140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1507205282860944e-05,
+      "loss": 0.5926,
+      "step": 5141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1489030442460385e-05,
+      "loss": 0.6703,
+      "step": 5142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1470861182554534e-05,
+      "loss": 0.6773,
+      "step": 5143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1452697506699663e-05,
+      "loss": 0.6018,
+      "step": 5144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1434539418451e-05,
+      "loss": 0.628,
+      "step": 5145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1416386921362615e-05,
+      "loss": 0.6561,
+      "step": 5146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.139824001898757e-05,
+      "loss": 0.6614,
+      "step": 5147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.138009871487772e-05,
+      "loss": 0.5934,
+      "step": 5148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1361963012583974e-05,
+      "loss": 0.6069,
+      "step": 5149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1343832915656004e-05,
+      "loss": 0.5682,
+      "step": 5150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1325708427642466e-05,
+      "loss": 0.6936,
+      "step": 5151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1307589552090866e-05,
+      "loss": 0.6249,
+      "step": 5152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.128947629254767e-05,
+      "loss": 0.5975,
+      "step": 5153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.127136865255818e-05,
+      "loss": 0.6546,
+      "step": 5154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.125326663566666e-05,
+      "loss": 0.6009,
+      "step": 5155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1235170245416223e-05,
+      "loss": 0.6308,
+      "step": 5156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.121707948534894e-05,
+      "loss": 0.5877,
+      "step": 5157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.119899435900569e-05,
+      "loss": 0.6622,
+      "step": 5158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1180914869926355e-05,
+      "loss": 0.5953,
+      "step": 5159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.116284102164961e-05,
+      "loss": 0.655,
+      "step": 5160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1144772817713103e-05,
+      "loss": 0.6813,
+      "step": 5161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1126710261653342e-05,
+      "loss": 0.6479,
+      "step": 5162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.110865335700577e-05,
+      "loss": 0.6924,
+      "step": 5163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.109060210730464e-05,
+      "loss": 0.6545,
+      "step": 5164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1072556516083204e-05,
+      "loss": 0.5632,
+      "step": 5165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1054516586873496e-05,
+      "loss": 0.3842,
+      "step": 5166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.103648232320653e-05,
+      "loss": 0.657,
+      "step": 5167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1018453728612175e-05,
+      "loss": 0.652,
+      "step": 5168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.100043080661922e-05,
+      "loss": 0.6176,
+      "step": 5169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0982413560755272e-05,
+      "loss": 0.601,
+      "step": 5170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.096440199454691e-05,
+      "loss": 0.6322,
+      "step": 5171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0946396111519533e-05,
+      "loss": 0.6009,
+      "step": 5172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0928395915197484e-05,
+      "loss": 0.6277,
+      "step": 5173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.091040140910396e-05,
+      "loss": 0.6244,
+      "step": 5174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0892412596761047e-05,
+      "loss": 0.6586,
+      "step": 5175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0874429481689752e-05,
+      "loss": 0.6191,
+      "step": 5176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0856452067409892e-05,
+      "loss": 0.6373,
+      "step": 5177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0838480357440256e-05,
+      "loss": 0.5953,
+      "step": 5178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0820514355298432e-05,
+      "loss": 0.5821,
+      "step": 5179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0802554064500956e-05,
+      "loss": 0.6976,
+      "step": 5180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0784599488563222e-05,
+      "loss": 0.6679,
+      "step": 5181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0766650630999517e-05,
+      "loss": 0.6015,
+      "step": 5182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0748707495322966e-05,
+      "loss": 0.6648,
+      "step": 5183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0730770085045648e-05,
+      "loss": 0.6416,
+      "step": 5184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0712838403678437e-05,
+      "loss": 0.6269,
+      "step": 5185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0694912454731142e-05,
+      "loss": 0.5804,
+      "step": 5186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.067699224171244e-05,
+      "loss": 0.656,
+      "step": 5187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0659077768129898e-05,
+      "loss": 0.5984,
+      "step": 5188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.064116903748991e-05,
+      "loss": 0.5943,
+      "step": 5189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0623266053297813e-05,
+      "loss": 0.6386,
+      "step": 5190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0605368819057746e-05,
+      "loss": 0.6076,
+      "step": 5191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0587477338272788e-05,
+      "loss": 0.5857,
+      "step": 5192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.056959161444485e-05,
+      "loss": 0.6467,
+      "step": 5193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0551711651074767e-05,
+      "loss": 0.6421,
+      "step": 5194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0533837451662156e-05,
+      "loss": 0.558,
+      "step": 5195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0515969019705615e-05,
+      "loss": 0.6479,
+      "step": 5196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.049810635870251e-05,
+      "loss": 0.6337,
+      "step": 5197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0480249472149177e-05,
+      "loss": 0.6017,
+      "step": 5198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0462398363540697e-05,
+      "loss": 0.6949,
+      "step": 5199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0444553036371184e-05,
+      "loss": 0.6945,
+      "step": 5200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0426713494133466e-05,
+      "loss": 0.6212,
+      "step": 5201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0408879740319342e-05,
+      "loss": 0.6751,
+      "step": 5202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.03910517784194e-05,
+      "loss": 0.6441,
+      "step": 5203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0373229611923177e-05,
+      "loss": 0.666,
+      "step": 5204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0355413244318977e-05,
+      "loss": 0.6316,
+      "step": 5205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0337602679094092e-05,
+      "loss": 0.6442,
+      "step": 5206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0319797919734558e-05,
+      "loss": 0.6354,
+      "step": 5207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0301998969725366e-05,
+      "loss": 0.5968,
+      "step": 5208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.028420583255029e-05,
+      "loss": 0.6427,
+      "step": 5209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0266418511692047e-05,
+      "loss": 0.665,
+      "step": 5210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0248637010632136e-05,
+      "loss": 0.6742,
+      "step": 5211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.023086133285098e-05,
+      "loss": 0.6259,
+      "step": 5212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0213091481827838e-05,
+      "loss": 0.6391,
+      "step": 5213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0195327461040846e-05,
+      "loss": 0.6592,
+      "step": 5214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0177569273966945e-05,
+      "loss": 0.6557,
+      "step": 5215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0159816924082015e-05,
+      "loss": 0.5858,
+      "step": 5216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0142070414860704e-05,
+      "loss": 0.6746,
+      "step": 5217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0124329749776583e-05,
+      "loss": 0.6784,
+      "step": 5218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.010659493230206e-05,
+      "loss": 0.6144,
+      "step": 5219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0088865965908422e-05,
+      "loss": 0.6558,
+      "step": 5220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.007114285406574e-05,
+      "loss": 0.5371,
+      "step": 5221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0053425600243025e-05,
+      "loss": 0.6139,
+      "step": 5222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0035714207908062e-05,
+      "loss": 0.6486,
+      "step": 5223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0018008680527556e-05,
+      "loss": 0.6013,
+      "step": 5224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.000030902156703e-05,
+      "loss": 0.6154,
+      "step": 5225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9982615234490882e-05,
+      "loss": 0.6587,
+      "step": 5226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.996492732276231e-05,
+      "loss": 0.6127,
+      "step": 5227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.994724528984343e-05,
+      "loss": 0.3551,
+      "step": 5228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9929569139195138e-05,
+      "loss": 0.6359,
+      "step": 5229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.991189887427723e-05,
+      "loss": 0.6592,
+      "step": 5230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9894234498548332e-05,
+      "loss": 0.6628,
+      "step": 5231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9876576015465937e-05,
+      "loss": 0.6452,
+      "step": 5232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.985892342848633e-05,
+      "loss": 0.6791,
+      "step": 5233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9841276741064718e-05,
+      "loss": 0.6387,
+      "step": 5234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9823635956655077e-05,
+      "loss": 0.6668,
+      "step": 5235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9806001078710278e-05,
+      "loss": 0.6862,
+      "step": 5236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.978837211068203e-05,
+      "loss": 0.6515,
+      "step": 5237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9770749056020866e-05,
+      "loss": 0.6406,
+      "step": 5238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9753131918176197e-05,
+      "loss": 0.6624,
+      "step": 5239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9735520700596217e-05,
+      "loss": 0.6451,
+      "step": 5240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9717915406728034e-05,
+      "loss": 0.6011,
+      "step": 5241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9700316040017515e-05,
+      "loss": 0.6284,
+      "step": 5242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9682722603909427e-05,
+      "loss": 0.6471,
+      "step": 5243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9665135101847372e-05,
+      "loss": 0.6175,
+      "step": 5244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9647553537273784e-05,
+      "loss": 0.7041,
+      "step": 5245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.96299779136299e-05,
+      "loss": 0.6361,
+      "step": 5246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.961240823435585e-05,
+      "loss": 0.63,
+      "step": 5247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9594844502890553e-05,
+      "loss": 0.6866,
+      "step": 5248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9577286722671788e-05,
+      "loss": 0.6292,
+      "step": 5249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9559734897136177e-05,
+      "loss": 0.5928,
+      "step": 5250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.954218902971917e-05,
+      "loss": 0.6668,
+      "step": 5251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9524649123855028e-05,
+      "loss": 0.5989,
+      "step": 5252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9507115182976877e-05,
+      "loss": 0.5694,
+      "step": 5253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.948958721051664e-05,
+      "loss": 0.6274,
+      "step": 5254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9472065209905126e-05,
+      "loss": 0.59,
+      "step": 5255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.945454918457189e-05,
+      "loss": 0.4112,
+      "step": 5256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9437039137945438e-05,
+      "loss": 0.6161,
+      "step": 5257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9419535073452988e-05,
+      "loss": 0.6556,
+      "step": 5258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.940203699452067e-05,
+      "loss": 0.609,
+      "step": 5259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.938454490457337e-05,
+      "loss": 0.619,
+      "step": 5260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9367058807034884e-05,
+      "loss": 0.6165,
+      "step": 5261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9349578705327752e-05,
+      "loss": 0.6568,
+      "step": 5262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9332104602873398e-05,
+      "loss": 0.6424,
+      "step": 5263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9314636503092053e-05,
+      "loss": 0.6037,
+      "step": 5264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9297174409402802e-05,
+      "loss": 0.6682,
+      "step": 5265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9279718325223474e-05,
+      "loss": 0.6417,
+      "step": 5266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.926226825397082e-05,
+      "loss": 0.6032,
+      "step": 5267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.924482419906034e-05,
+      "loss": 0.6535,
+      "step": 5268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.922738616390639e-05,
+      "loss": 0.6531,
+      "step": 5269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.920995415192215e-05,
+      "loss": 0.6278,
+      "step": 5270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9192528166519623e-05,
+      "loss": 0.6398,
+      "step": 5271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.91751082111096e-05,
+      "loss": 0.6302,
+      "step": 5272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.915769428910174e-05,
+      "loss": 0.6602,
+      "step": 5273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9140286403904468e-05,
+      "loss": 0.6068,
+      "step": 5274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.912288455892507e-05,
+      "loss": 0.6527,
+      "step": 5275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9105488757569635e-05,
+      "loss": 0.5898,
+      "step": 5276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.908809900324308e-05,
+      "loss": 0.6475,
+      "step": 5277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9070715299349105e-05,
+      "loss": 0.6423,
+      "step": 5278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.905333764929028e-05,
+      "loss": 0.6194,
+      "step": 5279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9035966056467918e-05,
+      "loss": 0.6151,
+      "step": 5280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.901860052428221e-05,
+      "loss": 0.6839,
+      "step": 5281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.900124105613213e-05,
+      "loss": 0.6316,
+      "step": 5282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.89838876554155e-05,
+      "loss": 0.6644,
+      "step": 5283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8966540325528885e-05,
+      "loss": 0.5934,
+      "step": 5284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8949199069867742e-05,
+      "loss": 0.6213,
+      "step": 5285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.893186389182626e-05,
+      "loss": 0.6436,
+      "step": 5286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8914534794797513e-05,
+      "loss": 0.6327,
+      "step": 5287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.889721178217333e-05,
+      "loss": 0.6158,
+      "step": 5288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.88798948573444e-05,
+      "loss": 0.649,
+      "step": 5289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8862584023700152e-05,
+      "loss": 0.6713,
+      "step": 5290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.88452792846289e-05,
+      "loss": 0.6628,
+      "step": 5291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8827980643517688e-05,
+      "loss": 0.41,
+      "step": 5292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8810688103752422e-05,
+      "loss": 0.5985,
+      "step": 5293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8793401668717802e-05,
+      "loss": 0.6114,
+      "step": 5294
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8776121341797336e-05,
+      "loss": 0.5951,
+      "step": 5295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8758847126373303e-05,
+      "loss": 0.6518,
+      "step": 5296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8741579025826837e-05,
+      "loss": 0.6661,
+      "step": 5297
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.872431704353782e-05,
+      "loss": 0.6001,
+      "step": 5298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8707061182885e-05,
+      "loss": 0.6394,
+      "step": 5299
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8689811447245836e-05,
+      "loss": 0.65,
+      "step": 5300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8672567839996717e-05,
+      "loss": 0.6072,
+      "step": 5301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8655330364512714e-05,
+      "loss": 0.666,
+      "step": 5302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8638099024167755e-05,
+      "loss": 0.6368,
+      "step": 5303
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8620873822334574e-05,
+      "loss": 0.6159,
+      "step": 5304
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8603654762384642e-05,
+      "loss": 0.6859,
+      "step": 5305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8586441847688325e-05,
+      "loss": 0.6188,
+      "step": 5306
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.856923508161466e-05,
+      "loss": 0.6639,
+      "step": 5307
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8552034467531644e-05,
+      "loss": 0.6048,
+      "step": 5308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8534840008805904e-05,
+      "loss": 0.6335,
+      "step": 5309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.851765170880299e-05,
+      "loss": 0.5891,
+      "step": 5310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.850046957088715e-05,
+      "loss": 0.6354,
+      "step": 5311
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8483293598421507e-05,
+      "loss": 0.5744,
+      "step": 5312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.846612379476791e-05,
+      "loss": 0.5837,
+      "step": 5313
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8448960163287034e-05,
+      "loss": 0.6602,
+      "step": 5314
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8431802707338353e-05,
+      "loss": 0.6757,
+      "step": 5315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8414651430280138e-05,
+      "loss": 0.6636,
+      "step": 5316
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8397506335469395e-05,
+      "loss": 0.6176,
+      "step": 5317
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8380367426261996e-05,
+      "loss": 0.3796,
+      "step": 5318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8363234706012533e-05,
+      "loss": 0.6588,
+      "step": 5319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8346108178074432e-05,
+      "loss": 0.6542,
+      "step": 5320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8328987845799906e-05,
+      "loss": 0.6559,
+      "step": 5321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.831187371253994e-05,
+      "loss": 0.6818,
+      "step": 5322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8294765781644285e-05,
+      "loss": 0.6584,
+      "step": 5323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.827766405646154e-05,
+      "loss": 0.6854,
+      "step": 5324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8260568540339012e-05,
+      "loss": 0.662,
+      "step": 5325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8243479236622857e-05,
+      "loss": 0.6247,
+      "step": 5326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8226396148657977e-05,
+      "loss": 0.6113,
+      "step": 5327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8209319279788095e-05,
+      "loss": 0.6613,
+      "step": 5328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8192248633355652e-05,
+      "loss": 0.6492,
+      "step": 5329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8175184212701958e-05,
+      "loss": 0.6593,
+      "step": 5330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.815812602116701e-05,
+      "loss": 0.67,
+      "step": 5331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8141074062089657e-05,
+      "loss": 0.6906,
+      "step": 5332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.81240283388075e-05,
+      "loss": 0.6703,
+      "step": 5333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8106988854656954e-05,
+      "loss": 0.3886,
+      "step": 5334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.808995561297313e-05,
+      "loss": 0.6692,
+      "step": 5335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8072928617090012e-05,
+      "loss": 0.6375,
+      "step": 5336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8055907870340296e-05,
+      "loss": 0.6234,
+      "step": 5337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8038893376055478e-05,
+      "loss": 0.6932,
+      "step": 5338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8021885137565835e-05,
+      "loss": 0.5934,
+      "step": 5339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8004883158200436e-05,
+      "loss": 0.6355,
+      "step": 5340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.798788744128707e-05,
+      "loss": 0.3694,
+      "step": 5341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.797089799015237e-05,
+      "loss": 0.614,
+      "step": 5342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.795391480812167e-05,
+      "loss": 0.6173,
+      "step": 5343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7936937898519126e-05,
+      "loss": 0.6574,
+      "step": 5344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7919967264667653e-05,
+      "loss": 0.578,
+      "step": 5345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7903002909888965e-05,
+      "loss": 0.6467,
+      "step": 5346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7886044837503486e-05,
+      "loss": 0.5631,
+      "step": 5347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7869093050830466e-05,
+      "loss": 0.6552,
+      "step": 5348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.785214755318788e-05,
+      "loss": 0.5683,
+      "step": 5349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7835208347892535e-05,
+      "loss": 0.6701,
+      "step": 5350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7818275438259902e-05,
+      "loss": 0.6395,
+      "step": 5351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7801348827604363e-05,
+      "loss": 0.575,
+      "step": 5352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7784428519238934e-05,
+      "loss": 0.6142,
+      "step": 5353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7767514516475487e-05,
+      "loss": 0.6064,
+      "step": 5354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7750606822624587e-05,
+      "loss": 0.5753,
+      "step": 5355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7733705440995642e-05,
+      "loss": 0.6278,
+      "step": 5356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771681037489675e-05,
+      "loss": 0.5871,
+      "step": 5357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769992162763482e-05,
+      "loss": 0.6332,
+      "step": 5358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768303920251551e-05,
+      "loss": 0.6202,
+      "step": 5359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.766616310284327e-05,
+      "loss": 0.5522,
+      "step": 5360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7649293331921246e-05,
+      "loss": 0.6627,
+      "step": 5361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.763242989305141e-05,
+      "loss": 0.6414,
+      "step": 5362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7615572789534445e-05,
+      "loss": 0.361,
+      "step": 5363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.759872202466983e-05,
+      "loss": 0.3541,
+      "step": 5364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7581877601755786e-05,
+      "loss": 0.6239,
+      "step": 5365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7565039524089316e-05,
+      "loss": 0.5851,
+      "step": 5366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7548207794966132e-05,
+      "loss": 0.6224,
+      "step": 5367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.753138241768075e-05,
+      "loss": 0.6307,
+      "step": 5368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.751456339552644e-05,
+      "loss": 0.6612,
+      "step": 5369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7497750731795176e-05,
+      "loss": 0.5957,
+      "step": 5370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7480944429777756e-05,
+      "loss": 0.5719,
+      "step": 5371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.746414449276369e-05,
+      "loss": 0.7493,
+      "step": 5372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7447350924041278e-05,
+      "loss": 0.6644,
+      "step": 5373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7430563726897513e-05,
+      "loss": 0.5865,
+      "step": 5374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.741378290461822e-05,
+      "loss": 0.5759,
+      "step": 5375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.739700846048789e-05,
+      "loss": 0.6072,
+      "step": 5376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7380240397789836e-05,
+      "loss": 0.6368,
+      "step": 5377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7363478719806085e-05,
+      "loss": 0.6451,
+      "step": 5378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7346723429817458e-05,
+      "loss": 0.6731,
+      "step": 5379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7329974531103445e-05,
+      "loss": 0.618,
+      "step": 5380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7313232026942373e-05,
+      "loss": 0.6556,
+      "step": 5381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.729649592061124e-05,
+      "loss": 0.6074,
+      "step": 5382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7279766215385852e-05,
+      "loss": 0.6696,
+      "step": 5383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7263042914540727e-05,
+      "loss": 0.6193,
+      "step": 5384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.724632602134918e-05,
+      "loss": 0.6116,
+      "step": 5385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.722961553908318e-05,
+      "loss": 0.6593,
+      "step": 5386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.721291147101353e-05,
+      "loss": 0.5884,
+      "step": 5387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7196213820409725e-05,
+      "loss": 0.6361,
+      "step": 5388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.717952259054002e-05,
+      "loss": 0.7052,
+      "step": 5389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.716283778467142e-05,
+      "loss": 0.6694,
+      "step": 5390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.714615940606969e-05,
+      "loss": 0.6102,
+      "step": 5391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.712948745799927e-05,
+      "loss": 0.6189,
+      "step": 5392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7112821943723433e-05,
+      "loss": 0.6425,
+      "step": 5393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7096162866504094e-05,
+      "loss": 0.6536,
+      "step": 5394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.707951022960199e-05,
+      "loss": 0.5966,
+      "step": 5395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.706286403627656e-05,
+      "loss": 0.6618,
+      "step": 5396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7046224289786e-05,
+      "loss": 0.6537,
+      "step": 5397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7029590993387213e-05,
+      "loss": 0.6196,
+      "step": 5398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.701296415033588e-05,
+      "loss": 0.6232,
+      "step": 5399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6996343763886373e-05,
+      "loss": 0.5819,
+      "step": 5400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6979729837291854e-05,
+      "loss": 0.6195,
+      "step": 5401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6963122373804134e-05,
+      "loss": 0.6592,
+      "step": 5402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.69465213766739e-05,
+      "loss": 0.6638,
+      "step": 5403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6929926849150428e-05,
+      "loss": 0.6803,
+      "step": 5404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.691333879448183e-05,
+      "loss": 0.6326,
+      "step": 5405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.689675721591487e-05,
+      "loss": 0.6882,
+      "step": 5406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6880182116695125e-05,
+      "loss": 0.6693,
+      "step": 5407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.686361350006681e-05,
+      "loss": 0.6621,
+      "step": 5408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.684705136927299e-05,
+      "loss": 0.648,
+      "step": 5409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6830495727555356e-05,
+      "loss": 0.6138,
+      "step": 5410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6813946578154387e-05,
+      "loss": 0.6304,
+      "step": 5411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.679740392430925e-05,
+      "loss": 0.69,
+      "step": 5412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6780867769257898e-05,
+      "loss": 0.6458,
+      "step": 5413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6764338116236934e-05,
+      "loss": 0.644,
+      "step": 5414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6747814968481755e-05,
+      "loss": 0.6132,
+      "step": 5415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6731298329226458e-05,
+      "loss": 0.5941,
+      "step": 5416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6714788201703897e-05,
+      "loss": 0.7092,
+      "step": 5417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6698284589145573e-05,
+      "loss": 0.628,
+      "step": 5418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.668178749478181e-05,
+      "loss": 0.7146,
+      "step": 5419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6665296921841566e-05,
+      "loss": 0.6187,
+      "step": 5420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6648812873552595e-05,
+      "loss": 0.652,
+      "step": 5421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6632335353141333e-05,
+      "loss": 0.6295,
+      "step": 5422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6615864363832966e-05,
+      "loss": 0.6477,
+      "step": 5423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.659939990885136e-05,
+      "loss": 0.6466,
+      "step": 5424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6582941991419155e-05,
+      "loss": 0.5996,
+      "step": 5425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6566490614757652e-05,
+      "loss": 0.6502,
+      "step": 5426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6550045782086922e-05,
+      "loss": 0.6639,
+      "step": 5427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.653360749662573e-05,
+      "loss": 0.5567,
+      "step": 5428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6517175761591592e-05,
+      "loss": 0.6173,
+      "step": 5429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.650075058020068e-05,
+      "loss": 0.6443,
+      "step": 5430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6484331955667947e-05,
+      "loss": 0.6373,
+      "step": 5431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6467919891207007e-05,
+      "loss": 0.5973,
+      "step": 5432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6451514390030237e-05,
+      "loss": 0.6204,
+      "step": 5433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.64351154553487e-05,
+      "loss": 0.596,
+      "step": 5434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6418723090372197e-05,
+      "loss": 0.5407,
+      "step": 5435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6402337298309234e-05,
+      "loss": 0.65,
+      "step": 5436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6385958082367002e-05,
+      "loss": 0.6536,
+      "step": 5437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6369585445751463e-05,
+      "loss": 0.6663,
+      "step": 5438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6353219391667218e-05,
+      "loss": 0.706,
+      "step": 5439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6336859923317643e-05,
+      "loss": 0.6325,
+      "step": 5440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6320507043904797e-05,
+      "loss": 0.3653,
+      "step": 5441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6304160756629473e-05,
+      "loss": 0.6242,
+      "step": 5442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6287821064691112e-05,
+      "loss": 0.5597,
+      "step": 5443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6271487971287954e-05,
+      "loss": 0.5987,
+      "step": 5444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6255161479616855e-05,
+      "loss": 0.6458,
+      "step": 5445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6238841592873444e-05,
+      "loss": 0.6365,
+      "step": 5446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6222528314252035e-05,
+      "loss": 0.6258,
+      "step": 5447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6206221646945674e-05,
+      "loss": 0.7051,
+      "step": 5448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6189921594146046e-05,
+      "loss": 0.6361,
+      "step": 5449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.617362815904363e-05,
+      "loss": 0.6071,
+      "step": 5450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.615734134482752e-05,
+      "loss": 0.7037,
+      "step": 5451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6141061154685577e-05,
+      "loss": 0.6175,
+      "step": 5452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6124787591804352e-05,
+      "loss": 0.6209,
+      "step": 5453
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.610852065936911e-05,
+      "loss": 0.6464,
+      "step": 5454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6092260360563764e-05,
+      "loss": 0.6766,
+      "step": 5455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6076006698571006e-05,
+      "loss": 0.6784,
+      "step": 5456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6059759676572143e-05,
+      "loss": 0.658,
+      "step": 5457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6043519297747285e-05,
+      "loss": 0.7212,
+      "step": 5458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6027285565275112e-05,
+      "loss": 0.5918,
+      "step": 5459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6011058482333162e-05,
+      "loss": 0.5734,
+      "step": 5460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5994838052097522e-05,
+      "loss": 0.5549,
+      "step": 5461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.597862427774308e-05,
+      "loss": 0.6451,
+      "step": 5462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5962417162443357e-05,
+      "loss": 0.6761,
+      "step": 5463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5946216709370615e-05,
+      "loss": 0.6311,
+      "step": 5464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.593002292169577e-05,
+      "loss": 0.6726,
+      "step": 5465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5913835802588467e-05,
+      "loss": 0.6383,
+      "step": 5466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5897655355217038e-05,
+      "loss": 0.6516,
+      "step": 5467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5881481582748525e-05,
+      "loss": 0.3627,
+      "step": 5468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5865314488348603e-05,
+      "loss": 0.6612,
+      "step": 5469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.584915407518172e-05,
+      "loss": 0.6371,
+      "step": 5470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.583300034641095e-05,
+      "loss": 0.5833,
+      "step": 5471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.58168533051981e-05,
+      "loss": 0.5675,
+      "step": 5472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5800712954703644e-05,
+      "loss": 0.6123,
+      "step": 5473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.578457929808679e-05,
+      "loss": 0.645,
+      "step": 5474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5768452338505352e-05,
+      "loss": 0.6318,
+      "step": 5475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5752332079115932e-05,
+      "loss": 0.621,
+      "step": 5476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5736218523073737e-05,
+      "loss": 0.6336,
+      "step": 5477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5720111673532707e-05,
+      "loss": 0.6393,
+      "step": 5478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5704011533645462e-05,
+      "loss": 0.6867,
+      "step": 5479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5687918106563326e-05,
+      "loss": 0.6288,
+      "step": 5480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.567183139543626e-05,
+      "loss": 0.7021,
+      "step": 5481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5655751403412966e-05,
+      "loss": 0.6902,
+      "step": 5482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5639678133640767e-05,
+      "loss": 0.6228,
+      "step": 5483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.562361158926574e-05,
+      "loss": 0.624,
+      "step": 5484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.56075517734326e-05,
+      "loss": 0.6362,
+      "step": 5485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5591498689284783e-05,
+      "loss": 0.6056,
+      "step": 5486
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5575452339964346e-05,
+      "loss": 0.6642,
+      "step": 5487
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5559412728612105e-05,
+      "loss": 0.5721,
+      "step": 5488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5543379858367475e-05,
+      "loss": 0.6186,
+      "step": 5489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5527353732368614e-05,
+      "loss": 0.607,
+      "step": 5490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5511334353752338e-05,
+      "loss": 0.6204,
+      "step": 5491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5495321725654167e-05,
+      "loss": 0.6629,
+      "step": 5492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5479315851208237e-05,
+      "loss": 0.5781,
+      "step": 5493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5463316733547434e-05,
+      "loss": 0.3256,
+      "step": 5494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5447324375803256e-05,
+      "loss": 0.6852,
+      "step": 5495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5431338781105937e-05,
+      "loss": 0.6985,
+      "step": 5496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.541535995258435e-05,
+      "loss": 0.6838,
+      "step": 5497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5399387893366064e-05,
+      "loss": 0.6636,
+      "step": 5498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5383422606577318e-05,
+      "loss": 0.348,
+      "step": 5499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5367464095343004e-05,
+      "loss": 0.6657,
+      "step": 5500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.535151236278673e-05,
+      "loss": 0.6492,
+      "step": 5501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5335567412030716e-05,
+      "loss": 0.6698,
+      "step": 5502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5319629246195917e-05,
+      "loss": 0.6312,
+      "step": 5503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5303697868401924e-05,
+      "loss": 0.6127,
+      "step": 5504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.528777328176703e-05,
+      "loss": 0.5959,
+      "step": 5505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5271855489408148e-05,
+      "loss": 0.7063,
+      "step": 5506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5255944494440927e-05,
+      "loss": 0.6495,
+      "step": 5507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5240040299979608e-05,
+      "loss": 0.6722,
+      "step": 5508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.522414290913718e-05,
+      "loss": 0.6182,
+      "step": 5509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5208252325025212e-05,
+      "loss": 0.6198,
+      "step": 5510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5192368550754061e-05,
+      "loss": 0.6289,
+      "step": 5511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5176491589432628e-05,
+      "loss": 0.6487,
+      "step": 5512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5160621444168571e-05,
+      "loss": 0.6573,
+      "step": 5513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5144758118068142e-05,
+      "loss": 0.5941,
+      "step": 5514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.512890161423633e-05,
+      "loss": 0.6212,
+      "step": 5515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5113051935776711e-05,
+      "loss": 0.5765,
+      "step": 5516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5097209085791598e-05,
+      "loss": 0.6684,
+      "step": 5517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.508137306738192e-05,
+      "loss": 0.5851,
+      "step": 5518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.506554388364731e-05,
+      "loss": 0.6562,
+      "step": 5519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5049721537686002e-05,
+      "loss": 0.56,
+      "step": 5520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5033906032594958e-05,
+      "loss": 0.5406,
+      "step": 5521
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5018097371469741e-05,
+      "loss": 0.6801,
+      "step": 5522
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5002295557404622e-05,
+      "loss": 0.5954,
+      "step": 5523
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4986500593492508e-05,
+      "loss": 0.6674,
+      "step": 5524
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4970712482824989e-05,
+      "loss": 0.588,
+      "step": 5525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.495493122849227e-05,
+      "loss": 0.6672,
+      "step": 5526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4939156833583262e-05,
+      "loss": 0.6339,
+      "step": 5527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.492338930118548e-05,
+      "loss": 0.6458,
+      "step": 5528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.490762863438515e-05,
+      "loss": 0.6341,
+      "step": 5529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4891874836267127e-05,
+      "loss": 0.6174,
+      "step": 5530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4876127909914939e-05,
+      "loss": 0.6409,
+      "step": 5531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.486038785841073e-05,
+      "loss": 0.558,
+      "step": 5532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4844654684835341e-05,
+      "loss": 0.613,
+      "step": 5533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4828928392268237e-05,
+      "loss": 0.6715,
+      "step": 5534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4813208983787553e-05,
+      "loss": 0.6044,
+      "step": 5535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4797496462470073e-05,
+      "loss": 0.6916,
+      "step": 5536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4781790831391251e-05,
+      "loss": 0.6359,
+      "step": 5537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4766092093625139e-05,
+      "loss": 0.6574,
+      "step": 5538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4750400252244511e-05,
+      "loss": 0.681,
+      "step": 5539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4734715310320718e-05,
+      "loss": 0.6596,
+      "step": 5540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4719037270923819e-05,
+      "loss": 0.6726,
+      "step": 5541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4703366137122494e-05,
+      "loss": 0.6609,
+      "step": 5542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4687701911984092e-05,
+      "loss": 0.6105,
+      "step": 5543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4672044598574568e-05,
+      "loss": 0.6907,
+      "step": 5544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.465639419995858e-05,
+      "loss": 0.6443,
+      "step": 5545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4640750719199376e-05,
+      "loss": 0.6617,
+      "step": 5546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4625114159358893e-05,
+      "loss": 0.5829,
+      "step": 5547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4609484523497696e-05,
+      "loss": 0.6584,
+      "step": 5548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4593861814675014e-05,
+      "loss": 0.6635,
+      "step": 5549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4578246035948673e-05,
+      "loss": 0.6885,
+      "step": 5550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.45626371903752e-05,
+      "loss": 0.5822,
+      "step": 5551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4547035281009708e-05,
+      "loss": 0.3601,
+      "step": 5552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4531440310906019e-05,
+      "loss": 0.619,
+      "step": 5553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4515852283116505e-05,
+      "loss": 0.6406,
+      "step": 5554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4500271200692295e-05,
+      "loss": 0.5938,
+      "step": 5555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4484697066683062e-05,
+      "loss": 0.6922,
+      "step": 5556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4469129884137177e-05,
+      "loss": 0.5986,
+      "step": 5557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4453569656101596e-05,
+      "loss": 0.6588,
+      "step": 5558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4438016385621983e-05,
+      "loss": 0.6855,
+      "step": 5559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4422470075742566e-05,
+      "loss": 0.6217,
+      "step": 5560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4406930729506268e-05,
+      "loss": 0.6194,
+      "step": 5561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4391398349954627e-05,
+      "loss": 0.6833,
+      "step": 5562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4375872940127815e-05,
+      "loss": 0.6226,
+      "step": 5563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.436035450306466e-05,
+      "loss": 0.6422,
+      "step": 5564
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4344843041802587e-05,
+      "loss": 0.6909,
+      "step": 5565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4329338559377691e-05,
+      "loss": 0.6116,
+      "step": 5566
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4313841058824674e-05,
+      "loss": 0.6214,
+      "step": 5567
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4298350543176891e-05,
+      "loss": 0.3696,
+      "step": 5568
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4282867015466328e-05,
+      "loss": 0.6174,
+      "step": 5569
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4267390478723614e-05,
+      "loss": 0.6976,
+      "step": 5570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4251920935977959e-05,
+      "loss": 0.6099,
+      "step": 5571
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4236458390257273e-05,
+      "loss": 0.5998,
+      "step": 5572
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4221002844588033e-05,
+      "loss": 0.6092,
+      "step": 5573
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4205554301995393e-05,
+      "loss": 0.6719,
+      "step": 5574
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4190112765503117e-05,
+      "loss": 0.6499,
+      "step": 5575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4174678238133621e-05,
+      "loss": 0.3424,
+      "step": 5576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4159250722907885e-05,
+      "loss": 0.6433,
+      "step": 5577
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4143830222845605e-05,
+      "loss": 0.6554,
+      "step": 5578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4128416740965017e-05,
+      "loss": 0.598,
+      "step": 5579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4113010280283046e-05,
+      "loss": 0.6067,
+      "step": 5580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4097610843815223e-05,
+      "loss": 0.6784,
+      "step": 5581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4082218434575712e-05,
+      "loss": 0.6226,
+      "step": 5582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4066833055577261e-05,
+      "loss": 0.6534,
+      "step": 5583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4051454709831308e-05,
+      "loss": 0.6099,
+      "step": 5584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4036083400347855e-05,
+      "loss": 0.5919,
+      "step": 5585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4020719130135552e-05,
+      "loss": 0.6482,
+      "step": 5586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.400536190220168e-05,
+      "loss": 0.6261,
+      "step": 5587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.399001171955215e-05,
+      "loss": 0.5373,
+      "step": 5588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3974668585191442e-05,
+      "loss": 0.6539,
+      "step": 5589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3959332502122718e-05,
+      "loss": 0.6542,
+      "step": 5590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3944003473347705e-05,
+      "loss": 0.7039,
+      "step": 5591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3928681501866792e-05,
+      "loss": 0.6115,
+      "step": 5592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3913366590678966e-05,
+      "loss": 0.6645,
+      "step": 5593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3898058742781867e-05,
+      "loss": 0.3666,
+      "step": 5594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3882757961171678e-05,
+      "loss": 0.6221,
+      "step": 5595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3867464248843281e-05,
+      "loss": 0.5897,
+      "step": 5596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3852177608790107e-05,
+      "loss": 0.6139,
+      "step": 5597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3836898044004243e-05,
+      "loss": 0.6317,
+      "step": 5598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3821625557476386e-05,
+      "loss": 0.5826,
+      "step": 5599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3806360152195857e-05,
+      "loss": 0.6188,
+      "step": 5600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3791101831150544e-05,
+      "loss": 0.6316,
+      "step": 5601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.377585059732701e-05,
+      "loss": 0.6191,
+      "step": 5602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3760606453710379e-05,
+      "loss": 0.6228,
+      "step": 5603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3745369403284437e-05,
+      "loss": 0.6451,
+      "step": 5604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3730139449031504e-05,
+      "loss": 0.5965,
+      "step": 5605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.371491659393263e-05,
+      "loss": 0.6813,
+      "step": 5606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3699700840967355e-05,
+      "loss": 0.6359,
+      "step": 5607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3684492193113917e-05,
+      "loss": 0.565,
+      "step": 5608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3669290653349093e-05,
+      "loss": 0.6426,
+      "step": 5609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3654096224648333e-05,
+      "loss": 0.572,
+      "step": 5610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3638908909985621e-05,
+      "loss": 0.6038,
+      "step": 5611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3623728712333656e-05,
+      "loss": 0.6718,
+      "step": 5612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3608555634663633e-05,
+      "loss": 0.6343,
+      "step": 5613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3593389679945433e-05,
+      "loss": 0.6544,
+      "step": 5614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3578230851147478e-05,
+      "loss": 0.6602,
+      "step": 5615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3563079151236864e-05,
+      "loss": 0.6176,
+      "step": 5616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3547934583179217e-05,
+      "loss": 0.571,
+      "step": 5617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3532797149938825e-05,
+      "loss": 0.6562,
+      "step": 5618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3517666854478561e-05,
+      "loss": 0.5923,
+      "step": 5619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3502543699759917e-05,
+      "loss": 0.6176,
+      "step": 5620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3487427688742943e-05,
+      "loss": 0.6745,
+      "step": 5621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3472318824386343e-05,
+      "loss": 0.6264,
+      "step": 5622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3457217109647363e-05,
+      "loss": 0.6816,
+      "step": 5623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3442122547481912e-05,
+      "loss": 0.6652,
+      "step": 5624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3427035140844463e-05,
+      "loss": 0.6034,
+      "step": 5625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3411954892688112e-05,
+      "loss": 0.5518,
+      "step": 5626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3396881805964513e-05,
+      "loss": 0.6281,
+      "step": 5627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.338181588362395e-05,
+      "loss": 0.6455,
+      "step": 5628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.336675712861532e-05,
+      "loss": 0.6286,
+      "step": 5629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3351705543886067e-05,
+      "loss": 0.6196,
+      "step": 5630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3336661132382272e-05,
+      "loss": 0.3682,
+      "step": 5631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3321623897048602e-05,
+      "loss": 0.7121,
+      "step": 5632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3306593840828336e-05,
+      "loss": 0.6226,
+      "step": 5633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3291570966663286e-05,
+      "loss": 0.5777,
+      "step": 5634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3276555277493946e-05,
+      "loss": 0.6195,
+      "step": 5635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.326154677625932e-05,
+      "loss": 0.6385,
+      "step": 5636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.324654546589707e-05,
+      "loss": 0.6471,
+      "step": 5637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3231551349343413e-05,
+      "loss": 0.6747,
+      "step": 5638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.321656442953319e-05,
+      "loss": 0.6193,
+      "step": 5639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3201584709399784e-05,
+      "loss": 0.6552,
+      "step": 5640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3186612191875226e-05,
+      "loss": 0.6238,
+      "step": 5641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3171646879890082e-05,
+      "loss": 0.5862,
+      "step": 5642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.315668877637355e-05,
+      "loss": 0.599,
+      "step": 5643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3141737884253397e-05,
+      "loss": 0.6214,
+      "step": 5644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3126794206456005e-05,
+      "loss": 0.6393,
+      "step": 5645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3111857745906286e-05,
+      "loss": 0.5927,
+      "step": 5646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3096928505527811e-05,
+      "loss": 0.5843,
+      "step": 5647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.308200648824267e-05,
+      "loss": 0.6005,
+      "step": 5648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3067091696971583e-05,
+      "loss": 0.5803,
+      "step": 5649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3052184134633855e-05,
+      "loss": 0.5239,
+      "step": 5650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.303728380414737e-05,
+      "loss": 0.6089,
+      "step": 5651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3022390708428572e-05,
+      "loss": 0.6796,
+      "step": 5652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3007504850392527e-05,
+      "loss": 0.6174,
+      "step": 5653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2992626232952842e-05,
+      "loss": 0.6367,
+      "step": 5654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.297775485902175e-05,
+      "loss": 0.6916,
+      "step": 5655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.296289073151004e-05,
+      "loss": 0.6614,
+      "step": 5656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2948033853327113e-05,
+      "loss": 0.646,
+      "step": 5657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.293318422738089e-05,
+      "loss": 0.6654,
+      "step": 5658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2918341856577943e-05,
+      "loss": 0.3911,
+      "step": 5659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2903506743823362e-05,
+      "loss": 0.7032,
+      "step": 5660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2888678892020879e-05,
+      "loss": 0.6016,
+      "step": 5661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2873858304072716e-05,
+      "loss": 0.5922,
+      "step": 5662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2859044982879804e-05,
+      "loss": 0.6387,
+      "step": 5663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2844238931341523e-05,
+      "loss": 0.6044,
+      "step": 5664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2829440152355915e-05,
+      "loss": 0.5898,
+      "step": 5665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2814648648819534e-05,
+      "loss": 0.6136,
+      "step": 5666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2799864423627578e-05,
+      "loss": 0.657,
+      "step": 5667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2785087479673752e-05,
+      "loss": 0.5507,
+      "step": 5668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2770317819850386e-05,
+      "loss": 0.5373,
+      "step": 5669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2755555447048367e-05,
+      "loss": 0.6481,
+      "step": 5670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2740800364157174e-05,
+      "loss": 0.6274,
+      "step": 5671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2726052574064807e-05,
+      "loss": 0.631,
+      "step": 5672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.27113120796579e-05,
+      "loss": 0.6162,
+      "step": 5673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2696578883821614e-05,
+      "loss": 0.6241,
+      "step": 5674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2681852989439703e-05,
+      "loss": 0.6441,
+      "step": 5675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2667134399394497e-05,
+      "loss": 0.5868,
+      "step": 5676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2652423116566898e-05,
+      "loss": 0.6368,
+      "step": 5677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2637719143836335e-05,
+      "loss": 0.602,
+      "step": 5678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.262302248408087e-05,
+      "loss": 0.5888,
+      "step": 5679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2608333140177076e-05,
+      "loss": 0.6271,
+      "step": 5680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2593651115000132e-05,
+      "loss": 0.63,
+      "step": 5681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2578976411423771e-05,
+      "loss": 0.6157,
+      "step": 5682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2564309032320315e-05,
+      "loss": 0.649,
+      "step": 5683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2549648980560602e-05,
+      "loss": 0.6316,
+      "step": 5684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2534996259014087e-05,
+      "loss": 0.6327,
+      "step": 5685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2520350870548746e-05,
+      "loss": 0.6107,
+      "step": 5686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2505712818031162e-05,
+      "loss": 0.6291,
+      "step": 5687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2491082104326457e-05,
+      "loss": 0.5792,
+      "step": 5688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2476458732298335e-05,
+      "loss": 0.576,
+      "step": 5689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2461842704809024e-05,
+      "loss": 0.6212,
+      "step": 5690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2447234024719374e-05,
+      "loss": 0.6121,
+      "step": 5691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2432632694888724e-05,
+      "loss": 0.6351,
+      "step": 5692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.241803871817504e-05,
+      "loss": 0.6947,
+      "step": 5693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2403452097434814e-05,
+      "loss": 0.6931,
+      "step": 5694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2388872835523119e-05,
+      "loss": 0.631,
+      "step": 5695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2374300935293576e-05,
+      "loss": 0.6495,
+      "step": 5696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2359736399598348e-05,
+      "loss": 0.6761,
+      "step": 5697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2345179231288195e-05,
+      "loss": 0.6417,
+      "step": 5698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2330629433212381e-05,
+      "loss": 0.586,
+      "step": 5699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2316087008218784e-05,
+      "loss": 0.5617,
+      "step": 5700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2301551959153813e-05,
+      "loss": 0.6395,
+      "step": 5701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2287024288862442e-05,
+      "loss": 0.6705,
+      "step": 5702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2272504000188173e-05,
+      "loss": 0.6181,
+      "step": 5703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2257991095973104e-05,
+      "loss": 0.7027,
+      "step": 5704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2243485579057845e-05,
+      "loss": 0.6545,
+      "step": 5705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2228987452281598e-05,
+      "loss": 0.6191,
+      "step": 5706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2214496718482093e-05,
+      "loss": 0.6149,
+      "step": 5707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.220001338049565e-05,
+      "loss": 0.5824,
+      "step": 5708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.218553744115708e-05,
+      "loss": 0.6257,
+      "step": 5709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.217106890329981e-05,
+      "loss": 0.3651,
+      "step": 5710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2156607769755762e-05,
+      "loss": 0.6745,
+      "step": 5711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2142154043355452e-05,
+      "loss": 0.6087,
+      "step": 5712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2127707726927905e-05,
+      "loss": 0.6768,
+      "step": 5713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.211326882330076e-05,
+      "loss": 0.6274,
+      "step": 5714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2098837335300133e-05,
+      "loss": 0.6251,
+      "step": 5715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2084413265750749e-05,
+      "loss": 0.6093,
+      "step": 5716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2069996617475821e-05,
+      "loss": 0.6336,
+      "step": 5717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2055587393297169e-05,
+      "loss": 0.6274,
+      "step": 5718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.204118559603511e-05,
+      "loss": 0.626,
+      "step": 5719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2026791228508532e-05,
+      "loss": 0.6226,
+      "step": 5720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2012404293534873e-05,
+      "loss": 0.6251,
+      "step": 5721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1998024793930124e-05,
+      "loss": 0.6395,
+      "step": 5722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1983652732508777e-05,
+      "loss": 0.676,
+      "step": 5723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.196928811208392e-05,
+      "loss": 0.6087,
+      "step": 5724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1954930935467146e-05,
+      "loss": 0.6071,
+      "step": 5725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1940581205468609e-05,
+      "loss": 0.6125,
+      "step": 5726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.192623892489701e-05,
+      "loss": 0.6011,
+      "step": 5727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1911904096559589e-05,
+      "loss": 0.6711,
+      "step": 5728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.189757672326211e-05,
+      "loss": 0.56,
+      "step": 5729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1883256807808906e-05,
+      "loss": 0.6445,
+      "step": 5730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.186894435300282e-05,
+      "loss": 0.6542,
+      "step": 5731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1854639361645254e-05,
+      "loss": 0.6218,
+      "step": 5732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1840341836536145e-05,
+      "loss": 0.611,
+      "step": 5733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.182605178047399e-05,
+      "loss": 0.6016,
+      "step": 5734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1811769196255773e-05,
+      "loss": 0.5944,
+      "step": 5735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1797494086677068e-05,
+      "loss": 0.5743,
+      "step": 5736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1783226454531942e-05,
+      "loss": 0.6908,
+      "step": 5737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1768966302613033e-05,
+      "loss": 0.6501,
+      "step": 5738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1754713633711501e-05,
+      "loss": 0.6191,
+      "step": 5739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1740468450617054e-05,
+      "loss": 0.7064,
+      "step": 5740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1726230756117901e-05,
+      "loss": 0.6567,
+      "step": 5741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1712000553000824e-05,
+      "loss": 0.5777,
+      "step": 5742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1697777844051105e-05,
+      "loss": 0.6577,
+      "step": 5743
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1683562632052586e-05,
+      "loss": 0.6349,
+      "step": 5744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1669354919787629e-05,
+      "loss": 0.6081,
+      "step": 5745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1655154710037153e-05,
+      "loss": 0.6358,
+      "step": 5746
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1640962005580546e-05,
+      "loss": 0.6514,
+      "step": 5747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1626776809195805e-05,
+      "loss": 0.659,
+      "step": 5748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.161259912365939e-05,
+      "loss": 0.6241,
+      "step": 5749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1598428951746332e-05,
+      "loss": 0.5886,
+      "step": 5750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1584266296230179e-05,
+      "loss": 0.5744,
+      "step": 5751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1570111159883028e-05,
+      "loss": 0.3873,
+      "step": 5752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1555963545475452e-05,
+      "loss": 0.6215,
+      "step": 5753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1541823455776618e-05,
+      "loss": 0.6134,
+      "step": 5754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1527690893554156e-05,
+      "loss": 0.5977,
+      "step": 5755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1513565861574277e-05,
+      "loss": 0.6364,
+      "step": 5756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1499448362601655e-05,
+      "loss": 0.7151,
+      "step": 5757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1485338399399576e-05,
+      "loss": 0.6409,
+      "step": 5758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.14712359747298e-05,
+      "loss": 0.6494,
+      "step": 5759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1457141091352591e-05,
+      "loss": 0.3592,
+      "step": 5760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.144305375202679e-05,
+      "loss": 0.3911,
+      "step": 5761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1428973959509693e-05,
+      "loss": 0.6219,
+      "step": 5762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1414901716557202e-05,
+      "loss": 0.6214,
+      "step": 5763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1400837025923649e-05,
+      "loss": 0.6553,
+      "step": 5764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1386779890361998e-05,
+      "loss": 0.652,
+      "step": 5765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1372730312623626e-05,
+      "loss": 0.6725,
+      "step": 5766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1358688295458503e-05,
+      "loss": 0.631,
+      "step": 5767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1344653841615072e-05,
+      "loss": 0.6297,
+      "step": 5768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1330626953840345e-05,
+      "loss": 0.5621,
+      "step": 5769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1316607634879783e-05,
+      "loss": 0.6117,
+      "step": 5770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1302595887477463e-05,
+      "loss": 0.5931,
+      "step": 5771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1288591714375884e-05,
+      "loss": 0.5839,
+      "step": 5772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1274595118316134e-05,
+      "loss": 0.6314,
+      "step": 5773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1260606102037758e-05,
+      "loss": 0.6371,
+      "step": 5774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.124662466827887e-05,
+      "loss": 0.6465,
+      "step": 5775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1232650819776052e-05,
+      "loss": 0.6378,
+      "step": 5776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1218684559264448e-05,
+      "loss": 0.6443,
+      "step": 5777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1204725889477685e-05,
+      "loss": 0.6667,
+      "step": 5778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1190774813147937e-05,
+      "loss": 0.676,
+      "step": 5779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1176831333005832e-05,
+      "loss": 0.6732,
+      "step": 5780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.116289545178058e-05,
+      "loss": 0.6205,
+      "step": 5781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1148967172199848e-05,
+      "loss": 0.6194,
+      "step": 5782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1135046496989843e-05,
+      "loss": 0.657,
+      "step": 5783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1121133428875292e-05,
+      "loss": 0.664,
+      "step": 5784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1107227970579426e-05,
+      "loss": 0.7001,
+      "step": 5785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1093330124823954e-05,
+      "loss": 0.617,
+      "step": 5786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1079439894329152e-05,
+      "loss": 0.6515,
+      "step": 5787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1065557281813749e-05,
+      "loss": 0.7226,
+      "step": 5788
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.105168228999502e-05,
+      "loss": 0.6147,
+      "step": 5789
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1037814921588741e-05,
+      "loss": 0.3711,
+      "step": 5790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1023955179309203e-05,
+      "loss": 0.6506,
+      "step": 5791
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1010103065869171e-05,
+      "loss": 0.3791,
+      "step": 5792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0996258583979962e-05,
+      "loss": 0.6763,
+      "step": 5793
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0982421736351356e-05,
+      "loss": 0.6361,
+      "step": 5794
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0968592525691673e-05,
+      "loss": 0.5924,
+      "step": 5795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0954770954707716e-05,
+      "loss": 0.5946,
+      "step": 5796
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0940957026104825e-05,
+      "loss": 0.6753,
+      "step": 5797
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.092715074258679e-05,
+      "loss": 0.6297,
+      "step": 5798
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0913352106855962e-05,
+      "loss": 0.6865,
+      "step": 5799
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0899561121613145e-05,
+      "loss": 0.6215,
+      "step": 5800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0885777789557678e-05,
+      "loss": 0.5841,
+      "step": 5801
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0872002113387397e-05,
+      "loss": 0.6661,
+      "step": 5802
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0858234095798647e-05,
+      "loss": 0.376,
+      "step": 5803
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0844473739486238e-05,
+      "loss": 0.5719,
+      "step": 5804
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0830721047143526e-05,
+      "loss": 0.6667,
+      "step": 5805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0816976021462334e-05,
+      "loss": 0.6909,
+      "step": 5806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0803238665132987e-05,
+      "loss": 0.6232,
+      "step": 5807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0789508980844337e-05,
+      "loss": 0.6669,
+      "step": 5808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0775786971283725e-05,
+      "loss": 0.6371,
+      "step": 5809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0762072639136943e-05,
+      "loss": 0.6441,
+      "step": 5810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.074836598708836e-05,
+      "loss": 0.6542,
+      "step": 5811
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0734667017820766e-05,
+      "loss": 0.6324,
+      "step": 5812
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0720975734015504e-05,
+      "loss": 0.3833,
+      "step": 5813
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0707292138352354e-05,
+      "loss": 0.6157,
+      "step": 5814
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0693616233509673e-05,
+      "loss": 0.641,
+      "step": 5815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0679948022164237e-05,
+      "loss": 0.5994,
+      "step": 5816
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0666287506991368e-05,
+      "loss": 0.6012,
+      "step": 5817
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.065263469066483e-05,
+      "loss": 0.6208,
+      "step": 5818
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0638989575856939e-05,
+      "loss": 0.584,
+      "step": 5819
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0625352165238444e-05,
+      "loss": 0.6142,
+      "step": 5820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0611722461478629e-05,
+      "loss": 0.621,
+      "step": 5821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0598100467245254e-05,
+      "loss": 0.6217,
+      "step": 5822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0584486185204584e-05,
+      "loss": 0.6665,
+      "step": 5823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0570879618021362e-05,
+      "loss": 0.6078,
+      "step": 5824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0557280768358801e-05,
+      "loss": 0.6462,
+      "step": 5825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0543689638878656e-05,
+      "loss": 0.6989,
+      "step": 5826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0530106232241099e-05,
+      "loss": 0.6197,
+      "step": 5827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0516530551104859e-05,
+      "loss": 0.664,
+      "step": 5828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0502962598127114e-05,
+      "loss": 0.6126,
+      "step": 5829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0489402375963558e-05,
+      "loss": 0.631,
+      "step": 5830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.047584988726833e-05,
+      "loss": 0.3866,
+      "step": 5831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0462305134694095e-05,
+      "loss": 0.6197,
+      "step": 5832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0448768120891966e-05,
+      "loss": 0.5975,
+      "step": 5833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.043523884851158e-05,
+      "loss": 0.6503,
+      "step": 5834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0421717320201041e-05,
+      "loss": 0.5194,
+      "step": 5835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0408203538606948e-05,
+      "loss": 0.6583,
+      "step": 5836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0394697506374345e-05,
+      "loss": 0.5741,
+      "step": 5837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.038119922614682e-05,
+      "loss": 0.6533,
+      "step": 5838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0367708700566381e-05,
+      "loss": 0.614,
+      "step": 5839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0354225932273564e-05,
+      "loss": 0.6501,
+      "step": 5840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0340750923907371e-05,
+      "loss": 0.5791,
+      "step": 5841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0327283678105304e-05,
+      "loss": 0.6033,
+      "step": 5842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.031382419750329e-05,
+      "loss": 0.6485,
+      "step": 5843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.030037248473581e-05,
+      "loss": 0.3744,
+      "step": 5844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0286928542435754e-05,
+      "loss": 0.5531,
+      "step": 5845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0273492373234538e-05,
+      "loss": 0.5974,
+      "step": 5846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0260063979762047e-05,
+      "loss": 0.3421,
+      "step": 5847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0246643364646658e-05,
+      "loss": 0.564,
+      "step": 5848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0233230530515176e-05,
+      "loss": 0.5813,
+      "step": 5849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.021982547999294e-05,
+      "loss": 0.6588,
+      "step": 5850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0206428215703712e-05,
+      "loss": 0.5591,
+      "step": 5851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0193038740269783e-05,
+      "loss": 0.6346,
+      "step": 5852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0179657056311886e-05,
+      "loss": 0.5827,
+      "step": 5853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0166283166449253e-05,
+      "loss": 0.6909,
+      "step": 5854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.015291707329955e-05,
+      "loss": 0.6389,
+      "step": 5855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.013955877947897e-05,
+      "loss": 0.5456,
+      "step": 5856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0126208287602117e-05,
+      "loss": 0.6269,
+      "step": 5857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0112865600282128e-05,
+      "loss": 0.653,
+      "step": 5858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0099530720130584e-05,
+      "loss": 0.6691,
+      "step": 5859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0086203649757547e-05,
+      "loss": 0.6432,
+      "step": 5860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0072884391771525e-05,
+      "loss": 0.5778,
+      "step": 5861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0059572948779539e-05,
+      "loss": 0.582,
+      "step": 5862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0046269323387036e-05,
+      "loss": 0.6948,
+      "step": 5863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0032973518197974e-05,
+      "loss": 0.5919,
+      "step": 5864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0019685535814727e-05,
+      "loss": 0.5727,
+      "step": 5865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0006405378838223e-05,
+      "loss": 0.638,
+      "step": 5866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.99313304986777e-06,
+      "loss": 0.6878,
+      "step": 5867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.979868551501199e-06,
+      "loss": 0.6055,
+      "step": 5868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.96661188633477e-06,
+      "loss": 0.6357,
+      "step": 5869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.953363056963254e-06,
+      "loss": 0.6319,
+      "step": 5870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.94012206597984e-06,
+      "loss": 0.3814,
+      "step": 5871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.926888915976219e-06,
+      "loss": 0.6605,
+      "step": 5872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.913663609542534e-06,
+      "loss": 0.6855,
+      "step": 5873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.900446149267412e-06,
+      "loss": 0.6371,
+      "step": 5874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.887236537737893e-06,
+      "loss": 0.6189,
+      "step": 5875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.874034777539542e-06,
+      "loss": 0.607,
+      "step": 5876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.860840871256338e-06,
+      "loss": 0.6268,
+      "step": 5877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.847654821470753e-06,
+      "loss": 0.6408,
+      "step": 5878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.834476630763711e-06,
+      "loss": 0.6571,
+      "step": 5879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.821306301714616e-06,
+      "loss": 0.6169,
+      "step": 5880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.808143836901284e-06,
+      "loss": 0.7091,
+      "step": 5881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.79498923890006e-06,
+      "loss": 0.6127,
+      "step": 5882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.781842510285677e-06,
+      "loss": 0.6289,
+      "step": 5883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.768703653631383e-06,
+      "loss": 0.7061,
+      "step": 5884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.755572671508868e-06,
+      "loss": 0.6193,
+      "step": 5885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.742449566488282e-06,
+      "loss": 0.3664,
+      "step": 5886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.729334341138208e-06,
+      "loss": 0.6216,
+      "step": 5887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.71622699802573e-06,
+      "loss": 0.5879,
+      "step": 5888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.703127539716367e-06,
+      "loss": 0.385,
+      "step": 5889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.69003596877408e-06,
+      "loss": 0.6272,
+      "step": 5890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.676952287761315e-06,
+      "loss": 0.6372,
+      "step": 5891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.663876499238956e-06,
+      "loss": 0.5913,
+      "step": 5892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.650808605766354e-06,
+      "loss": 0.6209,
+      "step": 5893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.637748609901292e-06,
+      "loss": 0.6217,
+      "step": 5894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.624696514200048e-06,
+      "loss": 0.6438,
+      "step": 5895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.611652321217296e-06,
+      "loss": 0.6387,
+      "step": 5896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.598616033506213e-06,
+      "loss": 0.6805,
+      "step": 5897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.585587653618405e-06,
+      "loss": 0.6183,
+      "step": 5898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.572567184103958e-06,
+      "loss": 0.6584,
+      "step": 5899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.559554627511352e-06,
+      "loss": 0.6356,
+      "step": 5900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.546549986387594e-06,
+      "loss": 0.7027,
+      "step": 5901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.533553263278061e-06,
+      "loss": 0.64,
+      "step": 5902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.52056446072664e-06,
+      "loss": 0.6288,
+      "step": 5903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.507583581275647e-06,
+      "loss": 0.5599,
+      "step": 5904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.494610627465873e-06,
+      "loss": 0.6453,
+      "step": 5905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.481645601836491e-06,
+      "loss": 0.5309,
+      "step": 5906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.468688506925199e-06,
+      "loss": 0.6123,
+      "step": 5907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.455739345268088e-06,
+      "loss": 0.6324,
+      "step": 5908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.442798119399721e-06,
+      "loss": 0.5865,
+      "step": 5909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.429864831853108e-06,
+      "loss": 0.651,
+      "step": 5910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.41693948515971e-06,
+      "loss": 0.3571,
+      "step": 5911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.4040220818494e-06,
+      "loss": 0.578,
+      "step": 5912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.39111262445055e-06,
+      "loss": 0.6237,
+      "step": 5913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.37821111548991e-06,
+      "loss": 0.6788,
+      "step": 5914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.365317557492754e-06,
+      "loss": 0.6346,
+      "step": 5915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.35243195298271e-06,
+      "loss": 0.6194,
+      "step": 5916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.339554304481951e-06,
+      "loss": 0.6146,
+      "step": 5917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.326684614510995e-06,
+      "loss": 0.5819,
+      "step": 5918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.313822885588875e-06,
+      "loss": 0.5951,
+      "step": 5919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.300969120233011e-06,
+      "loss": 0.6494,
+      "step": 5920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.288123320959308e-06,
+      "loss": 0.4033,
+      "step": 5921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.275285490282082e-06,
+      "loss": 0.6553,
+      "step": 5922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.2624556307141e-06,
+      "loss": 0.6341,
+      "step": 5923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.249633744766577e-06,
+      "loss": 0.639,
+      "step": 5924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.236819834949173e-06,
+      "loss": 0.6211,
+      "step": 5925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.224013903769946e-06,
+      "loss": 0.6185,
+      "step": 5926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.211215953735446e-06,
+      "loss": 0.6052,
+      "step": 5927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.198425987350606e-06,
+      "loss": 0.6321,
+      "step": 5928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.185644007118844e-06,
+      "loss": 0.673,
+      "step": 5929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.172870015541995e-06,
+      "loss": 0.6586,
+      "step": 5930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.160104015120341e-06,
+      "loss": 0.351,
+      "step": 5931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.147346008352564e-06,
+      "loss": 0.6185,
+      "step": 5932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.13459599773584e-06,
+      "loss": 0.563,
+      "step": 5933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.12185398576571e-06,
+      "loss": 0.5918,
+      "step": 5934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.109119974936214e-06,
+      "loss": 0.6423,
+      "step": 5935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.096393967739791e-06,
+      "loss": 0.6141,
+      "step": 5936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.083675966667333e-06,
+      "loss": 0.5965,
+      "step": 5937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.070965974208135e-06,
+      "loss": 0.6441,
+      "step": 5938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.058263992849964e-06,
+      "loss": 0.6038,
+      "step": 5939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.045570025078965e-06,
+      "loss": 0.6429,
+      "step": 5940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.032884073379771e-06,
+      "loss": 0.6424,
+      "step": 5941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.02020614023542e-06,
+      "loss": 0.3697,
+      "step": 5942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.007536228127394e-06,
+      "loss": 0.7505,
+      "step": 5943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.994874339535569e-06,
+      "loss": 0.6401,
+      "step": 5944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.982220476938296e-06,
+      "loss": 0.3843,
+      "step": 5945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.969574642812318e-06,
+      "loss": 0.6707,
+      "step": 5946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.95693683963283e-06,
+      "loss": 0.6557,
+      "step": 5947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.944307069873448e-06,
+      "loss": 0.5851,
+      "step": 5948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.931685336006229e-06,
+      "loss": 0.6733,
+      "step": 5949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.91907164050162e-06,
+      "loss": 0.6574,
+      "step": 5950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.906465985828538e-06,
+      "loss": 0.5861,
+      "step": 5951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.893868374454295e-06,
+      "loss": 0.6539,
+      "step": 5952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.881278808844634e-06,
+      "loss": 0.6644,
+      "step": 5953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.868697291463746e-06,
+      "loss": 0.6024,
+      "step": 5954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.85612382477422e-06,
+      "loss": 0.6418,
+      "step": 5955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.843558411237101e-06,
+      "loss": 0.6663,
+      "step": 5956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.831001053311805e-06,
+      "loss": 0.6852,
+      "step": 5957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.818451753456225e-06,
+      "loss": 0.6039,
+      "step": 5958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.805910514126631e-06,
+      "loss": 0.6117,
+      "step": 5959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.793377337777759e-06,
+      "loss": 0.6322,
+      "step": 5960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.780852226862735e-06,
+      "loss": 0.699,
+      "step": 5961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.768335183833132e-06,
+      "loss": 0.577,
+      "step": 5962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.755826211138913e-06,
+      "loss": 0.6602,
+      "step": 5963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.743325311228495e-06,
+      "loss": 0.6853,
+      "step": 5964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.73083248654868e-06,
+      "loss": 0.6122,
+      "step": 5965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.718347739544718e-06,
+      "loss": 0.638,
+      "step": 5966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.70587107266025e-06,
+      "loss": 0.3665,
+      "step": 5967
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.693402488337387e-06,
+      "loss": 0.5572,
+      "step": 5968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.680941989016588e-06,
+      "loss": 0.6237,
+      "step": 5969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.668489577136797e-06,
+      "loss": 0.6379,
+      "step": 5970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.656045255135314e-06,
+      "loss": 0.6341,
+      "step": 5971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.643609025447908e-06,
+      "loss": 0.6673,
+      "step": 5972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.63118089050871e-06,
+      "loss": 0.6413,
+      "step": 5973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.61876085275034e-06,
+      "loss": 0.6378,
+      "step": 5974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.606348914603756e-06,
+      "loss": 0.7048,
+      "step": 5975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.59394507849839e-06,
+      "loss": 0.661,
+      "step": 5976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.581549346862038e-06,
+      "loss": 0.5444,
+      "step": 5977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.56916172212096e-06,
+      "loss": 0.6408,
+      "step": 5978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.556782206699781e-06,
+      "loss": 0.6529,
+      "step": 5979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.544410803021574e-06,
+      "loss": 0.6484,
+      "step": 5980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.532047513507813e-06,
+      "loss": 0.558,
+      "step": 5981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.519692340578394e-06,
+      "loss": 0.6451,
+      "step": 5982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.507345286651591e-06,
+      "loss": 0.6153,
+      "step": 5983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.49500635414413e-06,
+      "loss": 0.611,
+      "step": 5984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.482675545471108e-06,
+      "loss": 0.626,
+      "step": 5985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.470352863046066e-06,
+      "loss": 0.6032,
+      "step": 5986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.458038309280941e-06,
+      "loss": 0.632,
+      "step": 5987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.445731886586095e-06,
+      "loss": 0.6118,
+      "step": 5988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.43343359737025e-06,
+      "loss": 0.6221,
+      "step": 5989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.421143444040591e-06,
+      "loss": 0.6629,
+      "step": 5990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.408861429002674e-06,
+      "loss": 0.6664,
+      "step": 5991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.396587554660484e-06,
+      "loss": 0.5952,
+      "step": 5992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.384321823416403e-06,
+      "loss": 0.7058,
+      "step": 5993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.372064237671235e-06,
+      "loss": 0.5982,
+      "step": 5994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.359814799824151e-06,
+      "loss": 0.6486,
+      "step": 5995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.34757351227277e-06,
+      "loss": 0.602,
+      "step": 5996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.335340377413086e-06,
+      "loss": 0.6376,
+      "step": 5997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.323115397639513e-06,
+      "loss": 0.5572,
+      "step": 5998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.310898575344861e-06,
+      "loss": 0.5951,
+      "step": 5999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.298689912920366e-06,
+      "loss": 0.5742,
+      "step": 6000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.28648941275562e-06,
+      "loss": 0.5285,
+      "step": 6001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.274297077238674e-06,
+      "loss": 0.6775,
+      "step": 6002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.262112908755921e-06,
+      "loss": 0.5683,
+      "step": 6003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.249936909692202e-06,
+      "loss": 0.3493,
+      "step": 6004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.237769082430745e-06,
+      "loss": 0.5603,
+      "step": 6005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.225609429353187e-06,
+      "loss": 0.6492,
+      "step": 6006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.213457952839527e-06,
+      "loss": 0.609,
+      "step": 6007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.201314655268222e-06,
+      "loss": 0.6104,
+      "step": 6008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.189179539016072e-06,
+      "loss": 0.6253,
+      "step": 6009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.177052606458313e-06,
+      "loss": 0.6857,
+      "step": 6010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.16493385996857e-06,
+      "loss": 0.6409,
+      "step": 6011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.152823301918872e-06,
+      "loss": 0.5955,
+      "step": 6012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.140720934679613e-06,
+      "loss": 0.6342,
+      "step": 6013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.128626760619633e-06,
+      "loss": 0.3719,
+      "step": 6014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.116540782106119e-06,
+      "loss": 0.6024,
+      "step": 6015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.104463001504697e-06,
+      "loss": 0.6433,
+      "step": 6016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.09239342117934e-06,
+      "loss": 0.617,
+      "step": 6017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.08033204349249e-06,
+      "loss": 0.6126,
+      "step": 6018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.068278870804902e-06,
+      "loss": 0.3379,
+      "step": 6019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.056233905475774e-06,
+      "loss": 0.5894,
+      "step": 6020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.044197149862692e-06,
+      "loss": 0.6304,
+      "step": 6021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.032168606321616e-06,
+      "loss": 0.603,
+      "step": 6022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.020148277206924e-06,
+      "loss": 0.6655,
+      "step": 6023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.008136164871338e-06,
+      "loss": 0.666,
+      "step": 6024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.996132271666062e-06,
+      "loss": 0.6557,
+      "step": 6025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.984136599940594e-06,
+      "loss": 0.6583,
+      "step": 6026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.97214915204289e-06,
+      "loss": 0.6124,
+      "step": 6027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.96016993031925e-06,
+      "loss": 0.59,
+      "step": 6028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.948198937114404e-06,
+      "loss": 0.6169,
+      "step": 6029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.936236174771427e-06,
+      "loss": 0.3748,
+      "step": 6030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.92428164563183e-06,
+      "loss": 0.5746,
+      "step": 6031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.912335352035483e-06,
+      "loss": 0.3773,
+      "step": 6032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.900397296320661e-06,
+      "loss": 0.6201,
+      "step": 6033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.888467480824003e-06,
+      "loss": 0.5821,
+      "step": 6034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.876545907880567e-06,
+      "loss": 0.6227,
+      "step": 6035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.864632579823761e-06,
+      "loss": 0.6332,
+      "step": 6036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.852727498985407e-06,
+      "loss": 0.6045,
+      "step": 6037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.840830667695703e-06,
+      "loss": 0.6702,
+      "step": 6038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.828942088283248e-06,
+      "loss": 0.6822,
+      "step": 6039
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.817061763074989e-06,
+      "loss": 0.6311,
+      "step": 6040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.805189694396298e-06,
+      "loss": 0.6491,
+      "step": 6041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.793325884570895e-06,
+      "loss": 0.6291,
+      "step": 6042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.78147033592091e-06,
+      "loss": 0.3524,
+      "step": 6043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.76962305076685e-06,
+      "loss": 0.6069,
+      "step": 6044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.757784031427613e-06,
+      "loss": 0.5971,
+      "step": 6045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.745953280220435e-06,
+      "loss": 0.6518,
+      "step": 6046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.734130799461003e-06,
+      "loss": 0.5773,
+      "step": 6047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.722316591463325e-06,
+      "loss": 0.6206,
+      "step": 6048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.710510658539816e-06,
+      "loss": 0.6351,
+      "step": 6049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.698713003001273e-06,
+      "loss": 0.5416,
+      "step": 6050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.686923627156883e-06,
+      "loss": 0.605,
+      "step": 6051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.675142533314172e-06,
+      "loss": 0.6173,
+      "step": 6052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.663369723779096e-06,
+      "loss": 0.6834,
+      "step": 6053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.651605200855939e-06,
+      "loss": 0.5904,
+      "step": 6054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.639848966847401e-06,
+      "loss": 0.6115,
+      "step": 6055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.628101024054552e-06,
+      "loss": 0.6677,
+      "step": 6056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.616361374776842e-06,
+      "loss": 0.6389,
+      "step": 6057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.604630021312065e-06,
+      "loss": 0.6207,
+      "step": 6058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.592906965956442e-06,
+      "loss": 0.7115,
+      "step": 6059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.581192211004523e-06,
+      "loss": 0.6471,
+      "step": 6060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.569485758749262e-06,
+      "loss": 0.6215,
+      "step": 6061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.557787611481981e-06,
+      "loss": 0.5906,
+      "step": 6062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.546097771492394e-06,
+      "loss": 0.6153,
+      "step": 6063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.534416241068537e-06,
+      "loss": 0.6594,
+      "step": 6064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.522743022496881e-06,
+      "loss": 0.5767,
+      "step": 6065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.511078118062226e-06,
+      "loss": 0.6291,
+      "step": 6066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.499421530047773e-06,
+      "loss": 0.3687,
+      "step": 6067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.487773260735059e-06,
+      "loss": 0.5752,
+      "step": 6068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.476133312404054e-06,
+      "loss": 0.5818,
+      "step": 6069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4645016873330366e-06,
+      "loss": 0.6444,
+      "step": 6070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.452878387798695e-06,
+      "loss": 0.6262,
+      "step": 6071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.441263416076067e-06,
+      "loss": 0.6582,
+      "step": 6072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.429656774438576e-06,
+      "loss": 0.6028,
+      "step": 6073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.41805846515799e-06,
+      "loss": 0.614,
+      "step": 6074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.406468490504481e-06,
+      "loss": 0.3555,
+      "step": 6075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.394886852746563e-06,
+      "loss": 0.625,
+      "step": 6076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3833135541511445e-06,
+      "loss": 0.693,
+      "step": 6077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3717485969834575e-06,
+      "loss": 0.5653,
+      "step": 6078
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.360191983507153e-06,
+      "loss": 0.6196,
+      "step": 6079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3486437159842045e-06,
+      "loss": 0.6412,
+      "step": 6080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.337103796674977e-06,
+      "loss": 0.6763,
+      "step": 6081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.325572227838201e-06,
+      "loss": 0.3798,
+      "step": 6082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.314049011730983e-06,
+      "loss": 0.647,
+      "step": 6083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.302534150608742e-06,
+      "loss": 0.6375,
+      "step": 6084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2910276467253235e-06,
+      "loss": 0.6387,
+      "step": 6085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.279529502332921e-06,
+      "loss": 0.6578,
+      "step": 6086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.268039719682063e-06,
+      "loss": 0.6619,
+      "step": 6087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.256558301021665e-06,
+      "loss": 0.643,
+      "step": 6088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.245085248599009e-06,
+      "loss": 0.5604,
+      "step": 6089
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.233620564659743e-06,
+      "loss": 0.6739,
+      "step": 6090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.222164251447838e-06,
+      "loss": 0.6067,
+      "step": 6091
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.21071631120569e-06,
+      "loss": 0.6438,
+      "step": 6092
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.199276746173988e-06,
+      "loss": 0.6581,
+      "step": 6093
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.187845558591832e-06,
+      "loss": 0.5741,
+      "step": 6094
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.176422750696665e-06,
+      "loss": 0.6339,
+      "step": 6095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.165008324724298e-06,
+      "loss": 0.3834,
+      "step": 6096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.153602282908877e-06,
+      "loss": 0.564,
+      "step": 6097
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.142204627482946e-06,
+      "loss": 0.6253,
+      "step": 6098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.130815360677351e-06,
+      "loss": 0.6438,
+      "step": 6099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1194344847213546e-06,
+      "loss": 0.6079,
+      "step": 6100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.108062001842547e-06,
+      "loss": 0.6425,
+      "step": 6101
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0966979142668925e-06,
+      "loss": 0.5855,
+      "step": 6102
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.085342224218683e-06,
+      "loss": 0.5947,
+      "step": 6103
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0739949339206025e-06,
+      "loss": 0.6602,
+      "step": 6104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.062656045593652e-06,
+      "loss": 0.6762,
+      "step": 6105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.051325561457217e-06,
+      "loss": 0.686,
+      "step": 6106
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.040003483729035e-06,
+      "loss": 0.6054,
+      "step": 6107
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.028689814625206e-06,
+      "loss": 0.6039,
+      "step": 6108
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.01738455636014e-06,
+      "loss": 0.6179,
+      "step": 6109
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0060877111466635e-06,
+      "loss": 0.6415,
+      "step": 6110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.9947992811958965e-06,
+      "loss": 0.5472,
+      "step": 6111
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.983519268717353e-06,
+      "loss": 0.6455,
+      "step": 6112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.972247675918891e-06,
+      "loss": 0.6288,
+      "step": 6113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.960984505006724e-06,
+      "loss": 0.5937,
+      "step": 6114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.949729758185386e-06,
+      "loss": 0.5601,
+      "step": 6115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.938483437657811e-06,
+      "loss": 0.6118,
+      "step": 6116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.927245545625227e-06,
+      "loss": 0.629,
+      "step": 6117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.91601608428728e-06,
+      "loss": 0.5514,
+      "step": 6118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.904795055841884e-06,
+      "loss": 0.6314,
+      "step": 6119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.893582462485399e-06,
+      "loss": 0.6324,
+      "step": 6120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.882378306412446e-06,
+      "loss": 0.6217,
+      "step": 6121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.87118258981605e-06,
+      "loss": 0.6212,
+      "step": 6122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.859995314887552e-06,
+      "loss": 0.6328,
+      "step": 6123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.848816483816672e-06,
+      "loss": 0.6382,
+      "step": 6124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.837646098791433e-06,
+      "loss": 0.6043,
+      "step": 6125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.826484161998248e-06,
+      "loss": 0.6417,
+      "step": 6126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.815330675621856e-06,
+      "loss": 0.6442,
+      "step": 6127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8041856418453566e-06,
+      "loss": 0.6542,
+      "step": 6128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.793049062850166e-06,
+      "loss": 0.675,
+      "step": 6129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.781920940816084e-06,
+      "loss": 0.6677,
+      "step": 6130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7708012779212075e-06,
+      "loss": 0.5821,
+      "step": 6131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.759690076342023e-06,
+      "loss": 0.3436,
+      "step": 6132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.748587338253337e-06,
+      "loss": 0.5592,
+      "step": 6133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.737493065828321e-06,
+      "loss": 0.6946,
+      "step": 6134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.726407261238443e-06,
+      "loss": 0.6252,
+      "step": 6135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.715329926653574e-06,
+      "loss": 0.6077,
+      "step": 6136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.704261064241873e-06,
+      "loss": 0.5373,
+      "step": 6137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.693200676169881e-06,
+      "loss": 0.6249,
+      "step": 6138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.682148764602452e-06,
+      "loss": 0.6271,
+      "step": 6139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.671105331702821e-06,
+      "loss": 0.348,
+      "step": 6140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.660070379632494e-06,
+      "loss": 0.6496,
+      "step": 6141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.649043910551394e-06,
+      "loss": 0.6037,
+      "step": 6142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.638025926617725e-06,
+      "loss": 0.6881,
+      "step": 6143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.627016429988064e-06,
+      "loss": 0.5993,
+      "step": 6144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.616015422817312e-06,
+      "loss": 0.6128,
+      "step": 6145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.605022907258729e-06,
+      "loss": 0.6185,
+      "step": 6146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.594038885463866e-06,
+      "loss": 0.607,
+      "step": 6147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.583063359582675e-06,
+      "loss": 0.5929,
+      "step": 6148
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.572096331763383e-06,
+      "loss": 0.6924,
+      "step": 6149
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.561137804152595e-06,
+      "loss": 0.6144,
+      "step": 6150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.550187778895245e-06,
+      "loss": 0.6408,
+      "step": 6151
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.53924625813459e-06,
+      "loss": 0.6525,
+      "step": 6152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.528313244012246e-06,
+      "loss": 0.6776,
+      "step": 6153
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5173887386681236e-06,
+      "loss": 0.6432,
+      "step": 6154
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.506472744240516e-06,
+      "loss": 0.6225,
+      "step": 6155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4955652628660045e-06,
+      "loss": 0.5669,
+      "step": 6156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.484666296679542e-06,
+      "loss": 0.6026,
+      "step": 6157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.47377584781439e-06,
+      "loss": 0.6571,
+      "step": 6158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.462893918402174e-06,
+      "loss": 0.6243,
+      "step": 6159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.452020510572798e-06,
+      "loss": 0.3704,
+      "step": 6160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.441155626454559e-06,
+      "loss": 0.6254,
+      "step": 6161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.430299268174039e-06,
+      "loss": 0.6235,
+      "step": 6162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4194514378561725e-06,
+      "loss": 0.6098,
+      "step": 6163
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.40861213762422e-06,
+      "loss": 0.5799,
+      "step": 6164
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.397781369599792e-06,
+      "loss": 0.6492,
+      "step": 6165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.386959135902787e-06,
+      "loss": 0.5644,
+      "step": 6166
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3761454386514764e-06,
+      "loss": 0.6137,
+      "step": 6167
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3653402799624174e-06,
+      "loss": 0.6993,
+      "step": 6168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.354543661950552e-06,
+      "loss": 0.7256,
+      "step": 6169
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.343755586729072e-06,
+      "loss": 0.6003,
+      "step": 6170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.332976056409601e-06,
+      "loss": 0.3736,
+      "step": 6171
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.322205073101989e-06,
+      "loss": 0.6538,
+      "step": 6172
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.311442638914477e-06,
+      "loss": 0.6875,
+      "step": 6173
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3006887559536046e-06,
+      "loss": 0.6026,
+      "step": 6174
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.289943426324252e-06,
+      "loss": 0.5848,
+      "step": 6175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2792066521295985e-06,
+      "loss": 0.3611,
+      "step": 6176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.268478435471204e-06,
+      "loss": 0.685,
+      "step": 6177
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.257758778448886e-06,
+      "loss": 0.7686,
+      "step": 6178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.24704768316084e-06,
+      "loss": 0.663,
+      "step": 6179
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.236345151703549e-06,
+      "loss": 0.5741,
+      "step": 6180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.225651186171855e-06,
+      "loss": 0.637,
+      "step": 6181
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.214965788658883e-06,
+      "loss": 0.6385,
+      "step": 6182
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.204288961256105e-06,
+      "loss": 0.5564,
+      "step": 6183
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.193620706053321e-06,
+      "loss": 0.6161,
+      "step": 6184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.182961025138656e-06,
+      "loss": 0.6267,
+      "step": 6185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.172309920598512e-06,
+      "loss": 0.6651,
+      "step": 6186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1616673945176836e-06,
+      "loss": 0.6248,
+      "step": 6187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.151033448979215e-06,
+      "loss": 0.6228,
+      "step": 6188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.14040808606452e-06,
+      "loss": 0.7467,
+      "step": 6189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.129791307853311e-06,
+      "loss": 0.5936,
+      "step": 6190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1191831164236435e-06,
+      "loss": 0.65,
+      "step": 6191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.108583513851851e-06,
+      "loss": 0.6543,
+      "step": 6192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.097992502212635e-06,
+      "loss": 0.6527,
+      "step": 6193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0874100835789595e-06,
+      "loss": 0.6126,
+      "step": 6194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0768362600221506e-06,
+      "loss": 0.5975,
+      "step": 6195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0662710336118425e-06,
+      "loss": 0.6303,
+      "step": 6196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.055714406415997e-06,
+      "loss": 0.6401,
+      "step": 6197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.045166380500844e-06,
+      "loss": 0.6208,
+      "step": 6198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.034626957930994e-06,
+      "loss": 0.6877,
+      "step": 6199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.024096140769325e-06,
+      "loss": 0.3741,
+      "step": 6200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.01357393107706e-06,
+      "loss": 0.6066,
+      "step": 6201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.003060330913718e-06,
+      "loss": 0.3537,
+      "step": 6202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.992555342337164e-06,
+      "loss": 0.608,
+      "step": 6203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9820589674035254e-06,
+      "loss": 0.6281,
+      "step": 6204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.971571208167298e-06,
+      "loss": 0.5797,
+      "step": 6205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.961092066681245e-06,
+      "loss": 0.5918,
+      "step": 6206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.950621544996471e-06,
+      "loss": 0.6087,
+      "step": 6207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.940159645162396e-06,
+      "loss": 0.5216,
+      "step": 6208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9297063692267455e-06,
+      "loss": 0.6148,
+      "step": 6209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.919261719235541e-06,
+      "loss": 0.6408,
+      "step": 6210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9088256972331446e-06,
+      "loss": 0.3874,
+      "step": 6211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8983983052621925e-06,
+      "loss": 0.6043,
+      "step": 6212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.88797954536367e-06,
+      "loss": 0.6288,
+      "step": 6213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.87756941957685e-06,
+      "loss": 0.6391,
+      "step": 6214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.867167929939327e-06,
+      "loss": 0.673,
+      "step": 6215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.856775078487009e-06,
+      "loss": 0.661,
+      "step": 6216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.846390867254087e-06,
+      "loss": 0.6684,
+      "step": 6217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.83601529827309e-06,
+      "loss": 0.6109,
+      "step": 6218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.825648373574833e-06,
+      "loss": 0.6563,
+      "step": 6219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.815290095188452e-06,
+      "loss": 0.5967,
+      "step": 6220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.804940465141395e-06,
+      "loss": 0.6251,
+      "step": 6221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.794599485459412e-06,
+      "loss": 0.6751,
+      "step": 6222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.784267158166551e-06,
+      "loss": 0.5823,
+      "step": 6223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.773943485285182e-06,
+      "loss": 0.6118,
+      "step": 6224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.763628468835958e-06,
+      "loss": 0.5914,
+      "step": 6225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.753322110837872e-06,
+      "loss": 0.6222,
+      "step": 6226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.743024413308174e-06,
+      "loss": 0.6558,
+      "step": 6227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.732735378262482e-06,
+      "loss": 0.5868,
+      "step": 6228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.722455007714667e-06,
+      "loss": 0.3733,
+      "step": 6229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.712183303676927e-06,
+      "loss": 0.6351,
+      "step": 6230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.701920268159744e-06,
+      "loss": 0.639,
+      "step": 6231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.691665903171939e-06,
+      "loss": 0.6462,
+      "step": 6232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.681420210720595e-06,
+      "loss": 0.6211,
+      "step": 6233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.671183192811125e-06,
+      "loss": 0.6546,
+      "step": 6234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.660954851447231e-06,
+      "loss": 0.6233,
+      "step": 6235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.650735188630935e-06,
+      "loss": 0.6273,
+      "step": 6236
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.64052420636253e-06,
+      "loss": 0.693,
+      "step": 6237
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6303219066406455e-06,
+      "loss": 0.6784,
+      "step": 6238
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.620128291462173e-06,
+      "loss": 0.3594,
+      "step": 6239
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.609943362822328e-06,
+      "loss": 0.6286,
+      "step": 6240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.599767122714627e-06,
+      "loss": 0.5854,
+      "step": 6241
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5895995731308935e-06,
+      "loss": 0.5554,
+      "step": 6242
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.579440716061213e-06,
+      "loss": 0.6968,
+      "step": 6243
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.569290553494017e-06,
+      "loss": 0.638,
+      "step": 6244
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.559149087415988e-06,
+      "loss": 0.5334,
+      "step": 6245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.54901631981214e-06,
+      "loss": 0.6306,
+      "step": 6246
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.538892252665778e-06,
+      "loss": 0.6255,
+      "step": 6247
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.528776887958509e-06,
+      "loss": 0.7138,
+      "step": 6248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.518670227670209e-06,
+      "loss": 0.6727,
+      "step": 6249
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.508572273779089e-06,
+      "loss": 0.6022,
+      "step": 6250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4984830282616195e-06,
+      "loss": 0.575,
+      "step": 6251
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.488402493092593e-06,
+      "loss": 0.6239,
+      "step": 6252
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.478330670245086e-06,
+      "loss": 0.6226,
+      "step": 6253
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.468267561690477e-06,
+      "loss": 0.6113,
+      "step": 6254
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.458213169398424e-06,
+      "loss": 0.6994,
+      "step": 6255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.448167495336897e-06,
+      "loss": 0.7393,
+      "step": 6256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4381305414721415e-06,
+      "loss": 0.6522,
+      "step": 6257
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4281023097687065e-06,
+      "loss": 0.6566,
+      "step": 6258
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.418082802189434e-06,
+      "loss": 0.6064,
+      "step": 6259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.4080720206954705e-06,
+      "loss": 0.6578,
+      "step": 6260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.398069967246222e-06,
+      "loss": 0.6437,
+      "step": 6261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.388076643799422e-06,
+      "loss": 0.5676,
+      "step": 6262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3780920523110625e-06,
+      "loss": 0.6252,
+      "step": 6263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.368116194735451e-06,
+      "loss": 0.5734,
+      "step": 6264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.358149073025176e-06,
+      "loss": 0.607,
+      "step": 6265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.348190689131122e-06,
+      "loss": 0.5818,
+      "step": 6266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.338241045002451e-06,
+      "loss": 0.6549,
+      "step": 6267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.328300142586629e-06,
+      "loss": 0.6004,
+      "step": 6268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.318367983829392e-06,
+      "loss": 0.6357,
+      "step": 6269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.308444570674792e-06,
+      "loss": 0.6638,
+      "step": 6270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.298529905065125e-06,
+      "loss": 0.6985,
+      "step": 6271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.288623988941044e-06,
+      "loss": 0.6301,
+      "step": 6272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.27872682424142e-06,
+      "loss": 0.6484,
+      "step": 6273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.268838412903454e-06,
+      "loss": 0.6599,
+      "step": 6274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.258958756862603e-06,
+      "loss": 0.6196,
+      "step": 6275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.249087858052648e-06,
+      "loss": 0.5721,
+      "step": 6276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.239225718405616e-06,
+      "loss": 0.7035,
+      "step": 6277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.229372339851846e-06,
+      "loss": 0.6245,
+      "step": 6278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.219527724319956e-06,
+      "loss": 0.4115,
+      "step": 6279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.209691873736844e-06,
+      "loss": 0.6288,
+      "step": 6280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1998647900277085e-06,
+      "loss": 0.6513,
+      "step": 6281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.190046475116001e-06,
+      "loss": 0.6383,
+      "step": 6282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.180236930923488e-06,
+      "loss": 0.6436,
+      "step": 6283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.17043615937019e-06,
+      "loss": 0.6072,
+      "step": 6284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.160644162374445e-06,
+      "loss": 0.6434,
+      "step": 6285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.150860941852842e-06,
+      "loss": 0.6212,
+      "step": 6286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.141086499720282e-06,
+      "loss": 0.5284,
+      "step": 6287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1313208378899136e-06,
+      "loss": 0.3303,
+      "step": 6288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.121563958273195e-06,
+      "loss": 0.6268,
+      "step": 6289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1118158627798455e-06,
+      "loss": 0.6256,
+      "step": 6290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.10207655331788e-06,
+      "loss": 0.3699,
+      "step": 6291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.092346031793588e-06,
+      "loss": 0.58,
+      "step": 6292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.082624300111544e-06,
+      "loss": 0.6367,
+      "step": 6293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.072911360174587e-06,
+      "loss": 0.6955,
+      "step": 6294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0632072138838584e-06,
+      "loss": 0.6151,
+      "step": 6295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0535118631387494e-06,
+      "loss": 0.6058,
+      "step": 6296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0438253098369525e-06,
+      "loss": 0.6777,
+      "step": 6297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.03414755587443e-06,
+      "loss": 0.5879,
+      "step": 6298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0244786031454325e-06,
+      "loss": 0.4004,
+      "step": 6299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.014818453542469e-06,
+      "loss": 0.6517,
+      "step": 6300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.005167108956343e-06,
+      "loss": 0.6551,
+      "step": 6301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.995524571276111e-06,
+      "loss": 0.7018,
+      "step": 6302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.985890842389129e-06,
+      "loss": 0.6465,
+      "step": 6303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9762659241810285e-06,
+      "loss": 0.3479,
+      "step": 6304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.966649818535707e-06,
+      "loss": 0.6676,
+      "step": 6305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.957042527335332e-06,
+      "loss": 0.6121,
+      "step": 6306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9474440524603645e-06,
+      "loss": 0.616,
+      "step": 6307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.937854395789515e-06,
+      "loss": 0.6999,
+      "step": 6308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9282735591997806e-06,
+      "loss": 0.5715,
+      "step": 6309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.918701544566445e-06,
+      "loss": 0.5968,
+      "step": 6310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9091383537630545e-06,
+      "loss": 0.6107,
+      "step": 6311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.89958398866141e-06,
+      "loss": 0.5727,
+      "step": 6312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8900384511316234e-06,
+      "loss": 0.6429,
+      "step": 6313
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.880501743042032e-06,
+      "loss": 0.5714,
+      "step": 6314
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.870973866259287e-06,
+      "loss": 0.6305,
+      "step": 6315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.861454822648287e-06,
+      "loss": 0.5904,
+      "step": 6316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8519446140722235e-06,
+      "loss": 0.6389,
+      "step": 6317
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.842443242392519e-06,
+      "loss": 0.6266,
+      "step": 6318
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.832950709468914e-06,
+      "loss": 0.6278,
+      "step": 6319
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8234670171593775e-06,
+      "loss": 0.6044,
+      "step": 6320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.813992167320186e-06,
+      "loss": 0.6151,
+      "step": 6321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.804526161805833e-06,
+      "loss": 0.6502,
+      "step": 6322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.795069002469161e-06,
+      "loss": 0.6225,
+      "step": 6323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7856206911611965e-06,
+      "loss": 0.6934,
+      "step": 6324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.776181229731297e-06,
+      "loss": 0.5966,
+      "step": 6325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.766750620027033e-06,
+      "loss": 0.6297,
+      "step": 6326
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.757328863894306e-06,
+      "loss": 0.7278,
+      "step": 6327
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.747915963177218e-06,
+      "loss": 0.6268,
+      "step": 6328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738511919718186e-06,
+      "loss": 0.5569,
+      "step": 6329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.72911673535788e-06,
+      "loss": 0.5746,
+      "step": 6330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7197304119352355e-06,
+      "loss": 0.5625,
+      "step": 6331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.710352951287439e-06,
+      "loss": 0.5777,
+      "step": 6332
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.700984355249971e-06,
+      "loss": 0.6196,
+      "step": 6333
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.691624625656544e-06,
+      "loss": 0.6569,
+      "step": 6334
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.682273764339162e-06,
+      "loss": 0.6115,
+      "step": 6335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.672931773128075e-06,
+      "loss": 0.5969,
+      "step": 6336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.663598653851825e-06,
+      "loss": 0.6343,
+      "step": 6337
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6542744083371725e-06,
+      "loss": 0.6619,
+      "step": 6338
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.644959038409186e-06,
+      "loss": 0.6253,
+      "step": 6339
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.635652545891156e-06,
+      "loss": 0.6094,
+      "step": 6340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6263549326046704e-06,
+      "loss": 0.6077,
+      "step": 6341
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6170662003695575e-06,
+      "loss": 0.627,
+      "step": 6342
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.607786351003934e-06,
+      "loss": 0.6381,
+      "step": 6343
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.598515386324131e-06,
+      "loss": 0.6748,
+      "step": 6344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.589253308144787e-06,
+      "loss": 0.6001,
+      "step": 6345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.580000118278782e-06,
+      "loss": 0.6611,
+      "step": 6346
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5707558185372425e-06,
+      "loss": 0.6217,
+      "step": 6347
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.561520410729575e-06,
+      "loss": 0.5469,
+      "step": 6348
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.552293896663451e-06,
+      "loss": 0.6325,
+      "step": 6349
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.543076278144787e-06,
+      "loss": 0.6711,
+      "step": 6350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.533867556977744e-06,
+      "loss": 0.5636,
+      "step": 6351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5246677349647825e-06,
+      "loss": 0.6335,
+      "step": 6352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.515476813906572e-06,
+      "loss": 0.6599,
+      "step": 6353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.506294795602084e-06,
+      "loss": 0.6902,
+      "step": 6354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.497121681848521e-06,
+      "loss": 0.5991,
+      "step": 6355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4879574744413674e-06,
+      "loss": 0.605,
+      "step": 6356
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.478802175174318e-06,
+      "loss": 0.6245,
+      "step": 6357
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.469655785839377e-06,
+      "loss": 0.6576,
+      "step": 6358
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.460518308226763e-06,
+      "loss": 0.6695,
+      "step": 6359
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.451389744124978e-06,
+      "loss": 0.6499,
+      "step": 6360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4422700953207666e-06,
+      "loss": 0.624,
+      "step": 6361
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.433159363599143e-06,
+      "loss": 0.6692,
+      "step": 6362
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4240575507433494e-06,
+      "loss": 0.6371,
+      "step": 6363
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.414964658534903e-06,
+      "loss": 0.3826,
+      "step": 6364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.405880688753566e-06,
+      "loss": 0.6714,
+      "step": 6365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.396805643177354e-06,
+      "loss": 0.5938,
+      "step": 6366
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.387739523582551e-06,
+      "loss": 0.582,
+      "step": 6367
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3786823317436836e-06,
+      "loss": 0.6224,
+      "step": 6368
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.369634069433509e-06,
+      "loss": 0.5837,
+      "step": 6369
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.360594738423085e-06,
+      "loss": 0.638,
+      "step": 6370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.351564340481668e-06,
+      "loss": 0.5636,
+      "step": 6371
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.342542877376809e-06,
+      "loss": 0.6293,
+      "step": 6372
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.333530350874265e-06,
+      "loss": 0.5617,
+      "step": 6373
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.324526762738112e-06,
+      "loss": 0.627,
+      "step": 6374
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.315532114730608e-06,
+      "loss": 0.6551,
+      "step": 6375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.306546408612306e-06,
+      "loss": 0.6287,
+      "step": 6376
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.297569646141969e-06,
+      "loss": 0.6788,
+      "step": 6377
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.288601829076655e-06,
+      "loss": 0.6292,
+      "step": 6378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.27964295917162e-06,
+      "loss": 0.5829,
+      "step": 6379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.270693038180434e-06,
+      "loss": 0.5444,
+      "step": 6380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.261752067854851e-06,
+      "loss": 0.6168,
+      "step": 6381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2528200499449165e-06,
+      "loss": 0.6055,
+      "step": 6382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.243896986198898e-06,
+      "loss": 0.6,
+      "step": 6383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.234982878363325e-06,
+      "loss": 0.6389,
+      "step": 6384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.226077728182959e-06,
+      "loss": 0.6321,
+      "step": 6385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.217181537400821e-06,
+      "loss": 0.6548,
+      "step": 6386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2082943077581875e-06,
+      "loss": 0.6733,
+      "step": 6387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.19941604099457e-06,
+      "loss": 0.669,
+      "step": 6388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.190546738847711e-06,
+      "loss": 0.6381,
+      "step": 6389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.181686403053625e-06,
+      "loss": 0.7169,
+      "step": 6390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.172835035346539e-06,
+      "loss": 0.6966,
+      "step": 6391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.163992637458964e-06,
+      "loss": 0.5613,
+      "step": 6392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.155159211121629e-06,
+      "loss": 0.6392,
+      "step": 6393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1463347580635195e-06,
+      "loss": 0.6237,
+      "step": 6394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.137519280011848e-06,
+      "loss": 0.6554,
+      "step": 6395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.128712778692096e-06,
+      "loss": 0.6321,
+      "step": 6396
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.119915255827955e-06,
+      "loss": 0.6035,
+      "step": 6397
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1111267131413886e-06,
+      "loss": 0.6284,
+      "step": 6398
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.102347152352587e-06,
+      "loss": 0.6407,
+      "step": 6399
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.093576575179997e-06,
+      "loss": 0.6469,
+      "step": 6400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0848149833402795e-06,
+      "loss": 0.7143,
+      "step": 6401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.076062378548373e-06,
+      "loss": 0.7015,
+      "step": 6402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0673187625174195e-06,
+      "loss": 0.6159,
+      "step": 6403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0585841369588265e-06,
+      "loss": 0.648,
+      "step": 6404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.049858503582238e-06,
+      "loss": 0.5831,
+      "step": 6405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.041141864095549e-06,
+      "loss": 0.6839,
+      "step": 6406
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0324342202048504e-06,
+      "loss": 0.6225,
+      "step": 6407
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.023735573614529e-06,
+      "loss": 0.5956,
+      "step": 6408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0150459260271665e-06,
+      "loss": 0.6232,
+      "step": 6409
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.006365279143604e-06,
+      "loss": 0.6498,
+      "step": 6410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.997693634662925e-06,
+      "loss": 0.6457,
+      "step": 6411
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.989030994282434e-06,
+      "loss": 0.5938,
+      "step": 6412
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.980377359697701e-06,
+      "loss": 0.5596,
+      "step": 6413
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.97173273260249e-06,
+      "loss": 0.6023,
+      "step": 6414
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.963097114688846e-06,
+      "loss": 0.6395,
+      "step": 6415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.954470507647018e-06,
+      "loss": 0.3358,
+      "step": 6416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.945852913165504e-06,
+      "loss": 0.6464,
+      "step": 6417
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.937244332931051e-06,
+      "loss": 0.5947,
+      "step": 6418
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.928644768628631e-06,
+      "loss": 0.6333,
+      "step": 6419
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.920054221941433e-06,
+      "loss": 0.6079,
+      "step": 6420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.911472694550916e-06,
+      "loss": 0.6779,
+      "step": 6421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.902900188136732e-06,
+      "loss": 0.6199,
+      "step": 6422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8943367043768105e-06,
+      "loss": 0.6474,
+      "step": 6423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.885782244947283e-06,
+      "loss": 0.6002,
+      "step": 6424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8772368115225445e-06,
+      "loss": 0.6568,
+      "step": 6425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.868700405775183e-06,
+      "loss": 0.5957,
+      "step": 6426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.860173029376063e-06,
+      "loss": 0.6148,
+      "step": 6427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.851654683994238e-06,
+      "loss": 0.6334,
+      "step": 6428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.843145371297036e-06,
+      "loss": 0.6405,
+      "step": 6429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.834645092949973e-06,
+      "loss": 0.66,
+      "step": 6430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.826153850616859e-06,
+      "loss": 0.6636,
+      "step": 6431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.817671645959664e-06,
+      "loss": 0.6554,
+      "step": 6432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.809198480638643e-06,
+      "loss": 0.6586,
+      "step": 6433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8007343563122464e-06,
+      "loss": 0.6175,
+      "step": 6434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.792279274637184e-06,
+      "loss": 0.6465,
+      "step": 6435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7838332372683695e-06,
+      "loss": 0.6163,
+      "step": 6436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.775396245858959e-06,
+      "loss": 0.6605,
+      "step": 6437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7669683020603474e-06,
+      "loss": 0.3305,
+      "step": 6438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.758549407522144e-06,
+      "loss": 0.599,
+      "step": 6439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7501395638921852e-06,
+      "loss": 0.6407,
+      "step": 6440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.741738772816561e-06,
+      "loss": 0.5482,
+      "step": 6441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.733347035939544e-06,
+      "loss": 0.6115,
+      "step": 6442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.724964354903676e-06,
+      "loss": 0.714,
+      "step": 6443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7165907313497116e-06,
+      "loss": 0.6016,
+      "step": 6444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7082261669166398e-06,
+      "loss": 0.6443,
+      "step": 6445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6998706632416504e-06,
+      "loss": 0.6616,
+      "step": 6446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.691524221960202e-06,
+      "loss": 0.525,
+      "step": 6447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6831868447059324e-06,
+      "loss": 0.6689,
+      "step": 6448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.674858533110742e-06,
+      "loss": 0.5838,
+      "step": 6449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.666539288804738e-06,
+      "loss": 0.6656,
+      "step": 6450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6582291134162727e-06,
+      "loss": 0.6561,
+      "step": 6451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.64992800857189e-06,
+      "loss": 0.6564,
+      "step": 6452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6416359758963956e-06,
+      "loss": 0.6149,
+      "step": 6453
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.633353017012786e-06,
+      "loss": 0.6475,
+      "step": 6454
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6250791335422974e-06,
+      "loss": 0.6609,
+      "step": 6455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.616814327104401e-06,
+      "loss": 0.6637,
+      "step": 6456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6085585993167805e-06,
+      "loss": 0.5723,
+      "step": 6457
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.600311951795321e-06,
+      "loss": 0.6282,
+      "step": 6458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5920743861541815e-06,
+      "loss": 0.6109,
+      "step": 6459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5838459040056836e-06,
+      "loss": 0.6369,
+      "step": 6460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5756265069604167e-06,
+      "loss": 0.6461,
+      "step": 6461
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5674161966271714e-06,
+      "loss": 0.6169,
+      "step": 6462
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5592149746129732e-06,
+      "loss": 0.617,
+      "step": 6463
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.551022842523044e-06,
+      "loss": 0.7153,
+      "step": 6464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5428398019608564e-06,
+      "loss": 0.6478,
+      "step": 6465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5346658545280795e-06,
+      "loss": 0.6684,
+      "step": 6466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5265010018246113e-06,
+      "loss": 0.6257,
+      "step": 6467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5183452454485734e-06,
+      "loss": 0.6183,
+      "step": 6468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.510198586996316e-06,
+      "loss": 0.6115,
+      "step": 6469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5020610280623755e-06,
+      "loss": 0.5959,
+      "step": 6470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4939325702395497e-06,
+      "loss": 0.5463,
+      "step": 6471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4858132151188107e-06,
+      "loss": 0.6566,
+      "step": 6472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4777029642893976e-06,
+      "loss": 0.6556,
+      "step": 6473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4696018193387083e-06,
+      "loss": 0.6477,
+      "step": 6474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4615097818524235e-06,
+      "loss": 0.6244,
+      "step": 6475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4534268534144042e-06,
+      "loss": 0.659,
+      "step": 6476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4453530356067243e-06,
+      "loss": 0.6441,
+      "step": 6477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.437288330009697e-06,
+      "loss": 0.6158,
+      "step": 6478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.429232738201821e-06,
+      "loss": 0.3605,
+      "step": 6479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4211862617598457e-06,
+      "loss": 0.6504,
+      "step": 6480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4131489022587005e-06,
+      "loss": 0.6105,
+      "step": 6481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4051206612715716e-06,
+      "loss": 0.6434,
+      "step": 6482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.397101540369829e-06,
+      "loss": 0.6068,
+      "step": 6483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.389091541123074e-06,
+      "loss": 0.6306,
+      "step": 6484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3810906650991016e-06,
+      "loss": 0.6601,
+      "step": 6485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.373098913863948e-06,
+      "loss": 0.5969,
+      "step": 6486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.36511628898184e-06,
+      "loss": 0.6237,
+      "step": 6487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3571427920152388e-06,
+      "loss": 0.6135,
+      "step": 6488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3491784245248014e-06,
+      "loss": 0.6986,
+      "step": 6489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3412231880694145e-06,
+      "loss": 0.6405,
+      "step": 6490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3332770842061545e-06,
+      "loss": 0.6141,
+      "step": 6491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.325340114490344e-06,
+      "loss": 0.588,
+      "step": 6492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3174122804754738e-06,
+      "loss": 0.591,
+      "step": 6493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3094935837132855e-06,
+      "loss": 0.6151,
+      "step": 6494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.301584025753712e-06,
+      "loss": 0.6094,
+      "step": 6495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2936836081449207e-06,
+      "loss": 0.6518,
+      "step": 6496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2857923324332464e-06,
+      "loss": 0.6434,
+      "step": 6497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2779102001632755e-06,
+      "loss": 0.6039,
+      "step": 6498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.27003721287778e-06,
+      "loss": 0.6122,
+      "step": 6499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2621733721177548e-06,
+      "loss": 0.5969,
+      "step": 6500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2543186794224077e-06,
+      "loss": 0.6198,
+      "step": 6501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.246473136329148e-06,
+      "loss": 0.5942,
+      "step": 6502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2386367443735852e-06,
+      "loss": 0.5725,
+      "step": 6503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.230809505089566e-06,
+      "loss": 0.6415,
+      "step": 6504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2229914200091083e-06,
+      "loss": 0.5575,
+      "step": 6505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.215182490662466e-06,
+      "loss": 0.6663,
+      "step": 6506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.207382718578089e-06,
+      "loss": 0.6545,
+      "step": 6507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.199592105282656e-06,
+      "loss": 0.5894,
+      "step": 6508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.191810652301008e-06,
+      "loss": 0.6179,
+      "step": 6509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.184038361156244e-06,
+      "loss": 0.6384,
+      "step": 6510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1762752333696297e-06,
+      "loss": 0.3752,
+      "step": 6511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1685212704606615e-06,
+      "loss": 0.615,
+      "step": 6512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1607764739470302e-06,
+      "loss": 0.6142,
+      "step": 6513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.153040845344646e-06,
+      "loss": 0.5735,
+      "step": 6514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1453143861676025e-06,
+      "loss": 0.5741,
+      "step": 6515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1375970979282187e-06,
+      "loss": 0.5375,
+      "step": 6516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.129888982137008e-06,
+      "loss": 0.6485,
+      "step": 6517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.122190040302686e-06,
+      "loss": 0.5979,
+      "step": 6518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1145002739321916e-06,
+      "loss": 0.6065,
+      "step": 6519
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1068196845306487e-06,
+      "loss": 0.5944,
+      "step": 6520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.099148273601382e-06,
+      "loss": 0.6421,
+      "step": 6521
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.091486042645947e-06,
+      "loss": 0.631,
+      "step": 6522
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.08383299316406e-06,
+      "loss": 0.5868,
+      "step": 6523
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0761891266536837e-06,
+      "loss": 0.6511,
+      "step": 6524
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0685544446109382e-06,
+      "loss": 0.6034,
+      "step": 6525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0609289485302064e-06,
+      "loss": 0.6729,
+      "step": 6526
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0533126399040157e-06,
+      "loss": 0.6795,
+      "step": 6527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.045705520223124e-06,
+      "loss": 0.5886,
+      "step": 6528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0381075909764744e-06,
+      "loss": 0.6663,
+      "step": 6529
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0305188536512376e-06,
+      "loss": 0.6295,
+      "step": 6530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.022939309732753e-06,
+      "loss": 0.6339,
+      "step": 6531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0153689607045845e-06,
+      "loss": 0.6225,
+      "step": 6532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.007807808048485e-06,
+      "loss": 0.6685,
+      "step": 6533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0002558532444157e-06,
+      "loss": 0.6294,
+      "step": 6534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.992713097770522e-06,
+      "loss": 0.5811,
+      "step": 6535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.985179543103167e-06,
+      "loss": 0.6327,
+      "step": 6536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9776551907168992e-06,
+      "loss": 0.6381,
+      "step": 6537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9701400420844737e-06,
+      "loss": 0.5877,
+      "step": 6538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.962634098676842e-06,
+      "loss": 0.6645,
+      "step": 6539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9551373619631508e-06,
+      "loss": 0.6733,
+      "step": 6540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.947649833410765e-06,
+      "loss": 0.624,
+      "step": 6541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9401715144852017e-06,
+      "loss": 0.5671,
+      "step": 6542
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.932702406650223e-06,
+      "loss": 0.6102,
+      "step": 6543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9252425113677594e-06,
+      "loss": 0.5973,
+      "step": 6544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.917791830097949e-06,
+      "loss": 0.6285,
+      "step": 6545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9103503642991247e-06,
+      "loss": 0.5967,
+      "step": 6546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9029181154278274e-06,
+      "loss": 0.6771,
+      "step": 6547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.89549508493876e-06,
+      "loss": 0.6526,
+      "step": 6548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.888081274284865e-06,
+      "loss": 0.6643,
+      "step": 6549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.880676684917244e-06,
+      "loss": 0.6408,
+      "step": 6550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.873281318285209e-06,
+      "loss": 0.6376,
+      "step": 6551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.865895175836275e-06,
+      "loss": 0.6044,
+      "step": 6552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8585182590161462e-06,
+      "loss": 0.6561,
+      "step": 6553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.851150569268696e-06,
+      "loss": 0.6153,
+      "step": 6554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.843792108036042e-06,
+      "loss": 0.6597,
+      "step": 6555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.836442876758438e-06,
+      "loss": 0.5561,
+      "step": 6556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.829102876874379e-06,
+      "loss": 0.6442,
+      "step": 6557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8217721098205252e-06,
+      "loss": 0.6302,
+      "step": 6558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.814450577031752e-06,
+      "loss": 0.653,
+      "step": 6559
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.807138279941096e-06,
+      "loss": 0.6281,
+      "step": 6560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7998352199798183e-06,
+      "loss": 0.6363,
+      "step": 6561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7925413985773474e-06,
+      "loss": 0.7064,
+      "step": 6562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7852568171613134e-06,
+      "loss": 0.6621,
+      "step": 6563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7779814771575485e-06,
+      "loss": 0.5941,
+      "step": 6564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.770715379990069e-06,
+      "loss": 0.6253,
+      "step": 6565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.763458527081059e-06,
+      "loss": 0.6128,
+      "step": 6566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.756210919850932e-06,
+      "loss": 0.6492,
+      "step": 6567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.74897255971826e-06,
+      "loss": 0.6045,
+      "step": 6568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7417434480998207e-06,
+      "loss": 0.599,
+      "step": 6569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7345235864105866e-06,
+      "loss": 0.6245,
+      "step": 6570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.727312976063712e-06,
+      "loss": 0.36,
+      "step": 6571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7201116184705277e-06,
+      "loss": 0.6278,
+      "step": 6572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7129195150405852e-06,
+      "loss": 0.6434,
+      "step": 6573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7057366671815856e-06,
+      "loss": 0.6472,
+      "step": 6574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6985630762994484e-06,
+      "loss": 0.5943,
+      "step": 6575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.691398743798268e-06,
+      "loss": 0.6294,
+      "step": 6576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6842436710803387e-06,
+      "loss": 0.5983,
+      "step": 6577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6770978595461238e-06,
+      "loss": 0.6226,
+      "step": 6578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6699613105942935e-06,
+      "loss": 0.6001,
+      "step": 6579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.66283402562168e-06,
+      "loss": 0.5785,
+      "step": 6580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6557160060233346e-06,
+      "loss": 0.6195,
+      "step": 6581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.648607253192459e-06,
+      "loss": 0.6379,
+      "step": 6582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.641507768520479e-06,
+      "loss": 0.6175,
+      "step": 6583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6344175533969728e-06,
+      "loss": 0.6684,
+      "step": 6584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.627336609209735e-06,
+      "loss": 0.6604,
+      "step": 6585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.620264937344713e-06,
+      "loss": 0.3768,
+      "step": 6586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.613202539186066e-06,
+      "loss": 0.6801,
+      "step": 6587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.606149416116116e-06,
+      "loss": 0.6159,
+      "step": 6588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.599105569515392e-06,
+      "loss": 0.5726,
+      "step": 6589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5920710007625914e-06,
+      "loss": 0.6379,
+      "step": 6590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5850457112346126e-06,
+      "loss": 0.6243,
+      "step": 6591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5780297023065057e-06,
+      "loss": 0.64,
+      "step": 6592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5710229753515446e-06,
+      "loss": 0.6373,
+      "step": 6593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5640255317411486e-06,
+      "loss": 0.6006,
+      "step": 6594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5570373728449504e-06,
+      "loss": 0.6948,
+      "step": 6595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5500585000307496e-06,
+      "loss": 0.5896,
+      "step": 6596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.543088914664532e-06,
+      "loss": 0.6239,
+      "step": 6597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5361286181104616e-06,
+      "loss": 0.3719,
+      "step": 6598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5291776117308984e-06,
+      "loss": 0.6844,
+      "step": 6599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5222358968863547e-06,
+      "loss": 0.6506,
+      "step": 6600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5153034749355487e-06,
+      "loss": 0.7145,
+      "step": 6601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.508380347235384e-06,
+      "loss": 0.6441,
+      "step": 6602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.501466515140932e-06,
+      "loss": 0.5852,
+      "step": 6603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4945619800054387e-06,
+      "loss": 0.3526,
+      "step": 6604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.487666743180345e-06,
+      "loss": 0.6006,
+      "step": 6605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4807808060152704e-06,
+      "loss": 0.6122,
+      "step": 6606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.473904169857999e-06,
+      "loss": 0.5653,
+      "step": 6607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.467036836054509e-06,
+      "loss": 0.3727,
+      "step": 6608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4601788059489537e-06,
+      "loss": 0.6527,
+      "step": 6609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4533300808836757e-06,
+      "loss": 0.6758,
+      "step": 6610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4464906621991747e-06,
+      "loss": 0.6261,
+      "step": 6611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4396605512341464e-06,
+      "loss": 0.6661,
+      "step": 6612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4328397493254494e-06,
+      "loss": 0.5848,
+      "step": 6613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.426028257808133e-06,
+      "loss": 0.6515,
+      "step": 6614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.419226078015424e-06,
+      "loss": 0.5925,
+      "step": 6615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4124332112787364e-06,
+      "loss": 0.5772,
+      "step": 6616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4056496589276224e-06,
+      "loss": 0.6104,
+      "step": 6617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3988754222898534e-06,
+      "loss": 0.6269,
+      "step": 6618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3921105026913527e-06,
+      "loss": 0.6009,
+      "step": 6619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.385354901456227e-06,
+      "loss": 0.7325,
+      "step": 6620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.378608619906769e-06,
+      "loss": 0.6195,
+      "step": 6621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3718716593634393e-06,
+      "loss": 0.6437,
+      "step": 6622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3651440211448663e-06,
+      "loss": 0.604,
+      "step": 6623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3584257065678628e-06,
+      "loss": 0.5832,
+      "step": 6624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.351716716947411e-06,
+      "loss": 0.6017,
+      "step": 6625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3450170535966766e-06,
+      "loss": 0.6216,
+      "step": 6626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3383267178269887e-06,
+      "loss": 0.6176,
+      "step": 6627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3316457109478716e-06,
+      "loss": 0.3859,
+      "step": 6628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3249740342669914e-06,
+      "loss": 0.6295,
+      "step": 6629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3183116890902135e-06,
+      "loss": 0.572,
+      "step": 6630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3116586767215677e-06,
+      "loss": 0.6122,
+      "step": 6631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3050149984632562e-06,
+      "loss": 0.6245,
+      "step": 6632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2983806556156508e-06,
+      "loss": 0.6267,
+      "step": 6633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2917556494773227e-06,
+      "loss": 0.6395,
+      "step": 6634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.285139981344969e-06,
+      "loss": 0.6462,
+      "step": 6635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2785336525135035e-06,
+      "loss": 0.5995,
+      "step": 6636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2719366642759754e-06,
+      "loss": 0.5879,
+      "step": 6637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2653490179236403e-06,
+      "loss": 0.3702,
+      "step": 6638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2587707147458893e-06,
+      "loss": 0.6555,
+      "step": 6639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2522017560303198e-06,
+      "loss": 0.613,
+      "step": 6640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.245642143062676e-06,
+      "loss": 0.5991,
+      "step": 6641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2390918771268863e-06,
+      "loss": 0.6389,
+      "step": 6642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2325509595050363e-06,
+      "loss": 0.6303,
+      "step": 6643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.226019391477402e-06,
+      "loss": 0.5778,
+      "step": 6644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2194971743224047e-06,
+      "loss": 0.6167,
+      "step": 6645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.212984309316646e-06,
+      "loss": 0.6745,
+      "step": 6646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2064807977349058e-06,
+      "loss": 0.698,
+      "step": 6647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1999866408501326e-06,
+      "loss": 0.638,
+      "step": 6648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1935018399334263e-06,
+      "loss": 0.6263,
+      "step": 6649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1870263962540714e-06,
+      "loss": 0.6187,
+      "step": 6650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.18056031107951e-06,
+      "loss": 0.6802,
+      "step": 6651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1741035856753634e-06,
+      "loss": 0.6335,
+      "step": 6652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.167656221305414e-06,
+      "loss": 0.6127,
+      "step": 6653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1612182192316257e-06,
+      "loss": 0.5938,
+      "step": 6654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1547895807141004e-06,
+      "loss": 0.6191,
+      "step": 6655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.148370307011144e-06,
+      "loss": 0.709,
+      "step": 6656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.141960399379189e-06,
+      "loss": 0.5961,
+      "step": 6657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.135559859072872e-06,
+      "loss": 0.6297,
+      "step": 6658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1291686873449735e-06,
+      "loss": 0.6328,
+      "step": 6659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1227868854464537e-06,
+      "loss": 0.6701,
+      "step": 6660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1164144546264252e-06,
+      "loss": 0.6079,
+      "step": 6661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.110051396132179e-06,
+      "loss": 0.6811,
+      "step": 6662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.103697711209157e-06,
+      "loss": 0.3643,
+      "step": 6663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0973534011009823e-06,
+      "loss": 0.6219,
+      "step": 6664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0910184670494326e-06,
+      "loss": 0.6617,
+      "step": 6665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.084692910294467e-06,
+      "loss": 0.5833,
+      "step": 6666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.078376732074178e-06,
+      "loss": 0.6518,
+      "step": 6667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.072069933624854e-06,
+      "loss": 0.6437,
+      "step": 6668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.065772516180925e-06,
+      "loss": 0.6175,
+      "step": 6669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0594844809749938e-06,
+      "loss": 0.3608,
+      "step": 6670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0532058292378307e-06,
+      "loss": 0.6559,
+      "step": 6671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0469365621983706e-06,
+      "loss": 0.6229,
+      "step": 6672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.040676681083703e-06,
+      "loss": 0.606,
+      "step": 6673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0344261871190816e-06,
+      "loss": 0.6056,
+      "step": 6674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.028185081527928e-06,
+      "loss": 0.5954,
+      "step": 6675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0219533655318147e-06,
+      "loss": 0.3593,
+      "step": 6676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0157310403504882e-06,
+      "loss": 0.6872,
+      "step": 6677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.009518107201858e-06,
+      "loss": 0.6246,
+      "step": 6678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0033145673019947e-06,
+      "loss": 0.6135,
+      "step": 6679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9971204218651173e-06,
+      "loss": 0.6374,
+      "step": 6680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9909356721036223e-06,
+      "loss": 0.6162,
+      "step": 6681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9847603192280515e-06,
+      "loss": 0.6426,
+      "step": 6682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9785943644471227e-06,
+      "loss": 0.5716,
+      "step": 6683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9724378089676978e-06,
+      "loss": 0.7037,
+      "step": 6684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9662906539948288e-06,
+      "loss": 0.6222,
+      "step": 6685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9601529007316875e-06,
+      "loss": 0.6518,
+      "step": 6686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9540245503796408e-06,
+      "loss": 0.6475,
+      "step": 6687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.947905604138178e-06,
+      "loss": 0.5682,
+      "step": 6688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.941796063204998e-06,
+      "loss": 0.6773,
+      "step": 6689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.93569592877591e-06,
+      "loss": 0.6605,
+      "step": 6690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.929605202044904e-06,
+      "loss": 0.6378,
+      "step": 6691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9235238842041312e-06,
+      "loss": 0.698,
+      "step": 6692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.917451976443907e-06,
+      "loss": 0.6555,
+      "step": 6693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.91138947995268e-06,
+      "loss": 0.6528,
+      "step": 6694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.905336395917079e-06,
+      "loss": 0.6514,
+      "step": 6695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8992927255218783e-06,
+      "loss": 0.5578,
+      "step": 6696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8932584699500156e-06,
+      "loss": 0.5946,
+      "step": 6697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8872336303825899e-06,
+      "loss": 0.6913,
+      "step": 6698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8812182079988528e-06,
+      "loss": 0.6833,
+      "step": 6699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.875212203976201e-06,
+      "loss": 0.5797,
+      "step": 6700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.869215619490211e-06,
+      "loss": 0.6438,
+      "step": 6701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.863228455714594e-06,
+      "loss": 0.6159,
+      "step": 6702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8572507138212292e-06,
+      "loss": 0.6699,
+      "step": 6703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8512823949801472e-06,
+      "loss": 0.6381,
+      "step": 6704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8453235003595471e-06,
+      "loss": 0.6005,
+      "step": 6705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8393740311257567e-06,
+      "loss": 0.5935,
+      "step": 6706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8334339884432893e-06,
+      "loss": 0.6487,
+      "step": 6707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.827503373474787e-06,
+      "loss": 0.6653,
+      "step": 6708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8215821873810601e-06,
+      "loss": 0.6756,
+      "step": 6709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8156704313210704e-06,
+      "loss": 0.5987,
+      "step": 6710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.809768106451948e-06,
+      "loss": 0.6549,
+      "step": 6711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.803875213928946e-06,
+      "loss": 0.6409,
+      "step": 6712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.797991754905498e-06,
+      "loss": 0.695,
+      "step": 6713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.792117730533177e-06,
+      "loss": 0.6658,
+      "step": 6714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.786253141961719e-06,
+      "loss": 0.6033,
+      "step": 6715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7803979903390055e-06,
+      "loss": 0.607,
+      "step": 6716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.774552276811081e-06,
+      "loss": 0.605,
+      "step": 6717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.768716002522125e-06,
+      "loss": 0.7325,
+      "step": 6718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7628891686144955e-06,
+      "loss": 0.5578,
+      "step": 6719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7570717762286692e-06,
+      "loss": 0.5996,
+      "step": 6720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.751263826503302e-06,
+      "loss": 0.6035,
+      "step": 6721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7454653205751903e-06,
+      "loss": 0.6024,
+      "step": 6722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7396762595792981e-06,
+      "loss": 0.6862,
+      "step": 6723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7338966446487027e-06,
+      "loss": 0.6177,
+      "step": 6724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.728126476914682e-06,
+      "loss": 0.6676,
+      "step": 6725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7223657575066165e-06,
+      "loss": 0.6521,
+      "step": 6726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7166144875520763e-06,
+      "loss": 0.6608,
+      "step": 6727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7108726681767495e-06,
+      "loss": 0.6389,
+      "step": 6728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7051403005045153e-06,
+      "loss": 0.6202,
+      "step": 6729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6994173856573592e-06,
+      "loss": 0.6219,
+      "step": 6730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6937039247554465e-06,
+      "loss": 0.6289,
+      "step": 6731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6879999189170714e-06,
+      "loss": 0.6527,
+      "step": 6732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6823053692587021e-06,
+      "loss": 0.6603,
+      "step": 6733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6766202768949246e-06,
+      "loss": 0.6183,
+      "step": 6734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6709446429384878e-06,
+      "loss": 0.3641,
+      "step": 6735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6652784685003197e-06,
+      "loss": 0.5739,
+      "step": 6736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6596217546894443e-06,
+      "loss": 0.6392,
+      "step": 6737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6539745026130704e-06,
+      "loss": 0.57,
+      "step": 6738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.648336713376536e-06,
+      "loss": 0.6647,
+      "step": 6739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6427083880833416e-06,
+      "loss": 0.6354,
+      "step": 6740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6370895278351118e-06,
+      "loss": 0.6468,
+      "step": 6741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6314801337316554e-06,
+      "loss": 0.3785,
+      "step": 6742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625880206870889e-06,
+      "loss": 0.6144,
+      "step": 6743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6202897483489133e-06,
+      "loss": 0.669,
+      "step": 6744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.614708759259942e-06,
+      "loss": 0.6158,
+      "step": 6745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6091372406963623e-06,
+      "loss": 0.6493,
+      "step": 6746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6035751937486798e-06,
+      "loss": 0.6323,
+      "step": 6747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5980226195055736e-06,
+      "loss": 0.6651,
+      "step": 6748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5924795190538577e-06,
+      "loss": 0.626,
+      "step": 6749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.586945893478492e-06,
+      "loss": 0.6538,
+      "step": 6750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5814217438625766e-06,
+      "loss": 0.5979,
+      "step": 6751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5759070712873635e-06,
+      "loss": 0.5793,
+      "step": 6752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5704018768322449e-06,
+      "loss": 0.591,
+      "step": 6753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.564906161574764e-06,
+      "loss": 0.588,
+      "step": 6754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.559419926590605e-06,
+      "loss": 0.6366,
+      "step": 6755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5539431729536036e-06,
+      "loss": 0.6623,
+      "step": 6756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5484759017357186e-06,
+      "loss": 0.65,
+      "step": 6757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5430181140070833e-06,
+      "loss": 0.6262,
+      "step": 6758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5375698108359426e-06,
+      "loss": 0.6879,
+      "step": 6759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.532130993288705e-06,
+      "loss": 0.6233,
+      "step": 6760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5267016624299247e-06,
+      "loss": 0.6257,
+      "step": 6761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5212818193222956e-06,
+      "loss": 0.6136,
+      "step": 6762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5158714650266414e-06,
+      "loss": 0.6849,
+      "step": 6763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5104706006019487e-06,
+      "loss": 0.626,
+      "step": 6764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5050792271053272e-06,
+      "loss": 0.609,
+      "step": 6765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4996973455920439e-06,
+      "loss": 0.6679,
+      "step": 6766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4943249571154949e-06,
+      "loss": 0.5793,
+      "step": 6767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.488962062727245e-06,
+      "loss": 0.6081,
+      "step": 6768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4836086634769596e-06,
+      "loss": 0.7026,
+      "step": 6769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4782647604124845e-06,
+      "loss": 0.5924,
+      "step": 6770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4729303545797768e-06,
+      "loss": 0.5866,
+      "step": 6771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4676054470229517e-06,
+      "loss": 0.6757,
+      "step": 6772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4622900387842642e-06,
+      "loss": 0.6771,
+      "step": 6773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4569841309041099e-06,
+      "loss": 0.6232,
+      "step": 6774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4516877244210137e-06,
+      "loss": 0.3722,
+      "step": 6775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4464008203716573e-06,
+      "loss": 0.6333,
+      "step": 6776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4411234197908463e-06,
+      "loss": 0.6278,
+      "step": 6777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.435855523711538e-06,
+      "loss": 0.618,
+      "step": 6778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4305971331648237e-06,
+      "loss": 0.5702,
+      "step": 6779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4253482491799474e-06,
+      "loss": 0.6174,
+      "step": 6780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4201088727842648e-06,
+      "loss": 0.6014,
+      "step": 6781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4148790050033e-06,
+      "loss": 0.5919,
+      "step": 6782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4096586468606898e-06,
+      "loss": 0.6125,
+      "step": 6783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4044477993782334e-06,
+      "loss": 0.5937,
+      "step": 6784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3992464635758483e-06,
+      "loss": 0.638,
+      "step": 6785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3940546404716092e-06,
+      "loss": 0.6448,
+      "step": 6786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3888723310817142e-06,
+      "loss": 0.6011,
+      "step": 6787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3836995364205129e-06,
+      "loss": 0.6131,
+      "step": 6788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3785362575004679e-06,
+      "loss": 0.3908,
+      "step": 6789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.373382495332215e-06,
+      "loss": 0.6083,
+      "step": 6790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3682382509244862e-06,
+      "loss": 0.6591,
+      "step": 6791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3631035252841873e-06,
+      "loss": 0.6105,
+      "step": 6792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.357978319416342e-06,
+      "loss": 0.6176,
+      "step": 6793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.35286263432412e-06,
+      "loss": 0.6446,
+      "step": 6794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3477564710088098e-06,
+      "loss": 0.5996,
+      "step": 6795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3426598304698613e-06,
+      "loss": 0.6023,
+      "step": 6796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3375727137048321e-06,
+      "loss": 0.608,
+      "step": 6797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3324951217094483e-06,
+      "loss": 0.6635,
+      "step": 6798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3274270554775425e-06,
+      "loss": 0.6284,
+      "step": 6799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.322368516001099e-06,
+      "loss": 0.6159,
+      "step": 6800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3173195042702369e-06,
+      "loss": 0.6483,
+      "step": 6801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.312280021273199e-06,
+      "loss": 0.6472,
+      "step": 6802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3072500679963795e-06,
+      "loss": 0.6482,
+      "step": 6803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.302229645424291e-06,
+      "loss": 0.6602,
+      "step": 6804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.297218754539592e-06,
+      "loss": 0.6093,
+      "step": 6805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.29221739632307e-06,
+      "loss": 0.5933,
+      "step": 6806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.287225571753653e-06,
+      "loss": 0.5714,
+      "step": 6807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.282243281808393e-06,
+      "loss": 0.6806,
+      "step": 6808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2772705274624818e-06,
+      "loss": 0.6149,
+      "step": 6809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2723073096892468e-06,
+      "loss": 0.5814,
+      "step": 6810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.267353629460144e-06,
+      "loss": 0.3306,
+      "step": 6811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.262409487744759e-06,
+      "loss": 0.5825,
+      "step": 6812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2574748855108342e-06,
+      "loss": 0.6041,
+      "step": 6813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2525498237242029e-06,
+      "loss": 0.5908,
+      "step": 6814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.247634303348877e-06,
+      "loss": 0.6918,
+      "step": 6815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.242728325346959e-06,
+      "loss": 0.6568,
+      "step": 6816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2378318906787145e-06,
+      "loss": 0.5964,
+      "step": 6817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2329450003025323e-06,
+      "loss": 0.6669,
+      "step": 6818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2280676551749248e-06,
+      "loss": 0.6207,
+      "step": 6819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2231998562505397e-06,
+      "loss": 0.6054,
+      "step": 6820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2183416044821705e-06,
+      "loss": 0.5767,
+      "step": 6821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2134929008207118e-06,
+      "loss": 0.6257,
+      "step": 6822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2086537462152214e-06,
+      "loss": 0.5923,
+      "step": 6823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2038241416128692e-06,
+      "loss": 0.6073,
+      "step": 6824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1990040879589715e-06,
+      "loss": 0.3513,
+      "step": 6825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1941935861969455e-06,
+      "loss": 0.6405,
+      "step": 6826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1893926372683772e-06,
+      "loss": 0.6225,
+      "step": 6827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1846012421129481e-06,
+      "loss": 0.6395,
+      "step": 6828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1798194016684861e-06,
+      "loss": 0.6684,
+      "step": 6829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1750471168709587e-06,
+      "loss": 0.5815,
+      "step": 6830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1702843886544523e-06,
+      "loss": 0.7101,
+      "step": 6831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1655312179511714e-06,
+      "loss": 0.6188,
+      "step": 6832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1607876056914713e-06,
+      "loss": 0.6123,
+      "step": 6833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1560535528038153e-06,
+      "loss": 0.666,
+      "step": 6834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1513290602148174e-06,
+      "loss": 0.6538,
+      "step": 6835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1466141288491994e-06,
+      "loss": 0.601,
+      "step": 6836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1419087596298394e-06,
+      "loss": 0.6303,
+      "step": 6837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1372129534777064e-06,
+      "loss": 0.6212,
+      "step": 6838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.132526711311932e-06,
+      "loss": 0.6285,
+      "step": 6839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1278500340497543e-06,
+      "loss": 0.5845,
+      "step": 6840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1231829226065472e-06,
+      "loss": 0.6113,
+      "step": 6841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1185253778958072e-06,
+      "loss": 0.6057,
+      "step": 6842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.113877400829172e-06,
+      "loss": 0.5815,
+      "step": 6843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1092389923163915e-06,
+      "loss": 0.6787,
+      "step": 6844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1046101532653507e-06,
+      "loss": 0.6282,
+      "step": 6845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.099990884582053e-06,
+      "loss": 0.649,
+      "step": 6846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0953811871706466e-06,
+      "loss": 0.6238,
+      "step": 6847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0907810619333825e-06,
+      "loss": 0.3825,
+      "step": 6848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0861905097706516e-06,
+      "loss": 0.685,
+      "step": 6849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0816095315809739e-06,
+      "loss": 0.6069,
+      "step": 6850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0770381282609931e-06,
+      "loss": 0.6772,
+      "step": 6851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0724763007054716e-06,
+      "loss": 0.6512,
+      "step": 6852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0679240498073118e-06,
+      "loss": 0.6225,
+      "step": 6853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0633813764575173e-06,
+      "loss": 0.6144,
+      "step": 6854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0588482815452438e-06,
+      "loss": 0.6545,
+      "step": 6855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0543247659577537e-06,
+      "loss": 0.5634,
+      "step": 6856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0498108305804499e-06,
+      "loss": 0.574,
+      "step": 6857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0453064762968422e-06,
+      "loss": 0.5874,
+      "step": 6858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0408117039885868e-06,
+      "loss": 0.6278,
+      "step": 6859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0363265145354406e-06,
+      "loss": 0.627,
+      "step": 6860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0318509088153016e-06,
+      "loss": 0.6087,
+      "step": 6861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0273848877041802e-06,
+      "loss": 0.6438,
+      "step": 6862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0229284520762328e-06,
+      "loss": 0.6533,
+      "step": 6863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0184816028037115e-06,
+      "loss": 0.6435,
+      "step": 6864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0140443407570034e-06,
+      "loss": 0.6414,
+      "step": 6865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0096166668046302e-06,
+      "loss": 0.6769,
+      "step": 6866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.005198581813216e-06,
+      "loss": 0.6343,
+      "step": 6867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0007900866475294e-06,
+      "loss": 0.6379,
+      "step": 6868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.963911821704475e-07,
+      "loss": 0.6117,
+      "step": 6869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.920018692429755e-07,
+      "loss": 0.6304,
+      "step": 6870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.876221487242322e-07,
+      "loss": 0.6004,
+      "step": 6871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.832520214714813e-07,
+      "loss": 0.6237,
+      "step": 6872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.788914883400834e-07,
+      "loss": 0.5931,
+      "step": 6873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.745405501835337e-07,
+      "loss": 0.6586,
+      "step": 6874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.701992078534505e-07,
+      "loss": 0.6023,
+      "step": 6875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.658674621995712e-07,
+      "loss": 0.6561,
+      "step": 6876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.615453140697562e-07,
+      "loss": 0.6701,
+      "step": 6877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.572327643099844e-07,
+      "loss": 0.5532,
+      "step": 6878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.529298137643527e-07,
+      "loss": 0.6215,
+      "step": 6879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.486364632750878e-07,
+      "loss": 0.5702,
+      "step": 6880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.443527136825336e-07,
+      "loss": 0.6677,
+      "step": 6881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.400785658251587e-07,
+      "loss": 0.5889,
+      "step": 6882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.358140205395438e-07,
+      "loss": 0.5753,
+      "step": 6883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.31559078660399e-07,
+      "loss": 0.6637,
+      "step": 6884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.273137410205413e-07,
+      "loss": 0.5946,
+      "step": 6885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.230780084509283e-07,
+      "loss": 0.6098,
+      "step": 6886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.188518817806135e-07,
+      "loss": 0.5834,
+      "step": 6887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.146353618367964e-07,
+      "loss": 0.5901,
+      "step": 6888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.104284494447779e-07,
+      "loss": 0.6399,
+      "step": 6889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.062311454279826e-07,
+      "loss": 0.6575,
+      "step": 6890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.020434506079534e-07,
+      "loss": 0.5505,
+      "step": 6891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.978653658043567e-07,
+      "loss": 0.6444,
+      "step": 6892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.93696891834972e-07,
+      "loss": 0.619,
+      "step": 6893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.895380295157018e-07,
+      "loss": 0.6268,
+      "step": 6894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.853887796605675e-07,
+      "loss": 0.5597,
+      "step": 6895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.812491430817083e-07,
+      "loss": 0.6326,
+      "step": 6896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.771191205893813e-07,
+      "loss": 0.6159,
+      "step": 6897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.729987129919682e-07,
+      "loss": 0.5946,
+      "step": 6898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.688879210959455e-07,
+      "loss": 0.622,
+      "step": 6899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.647867457059422e-07,
+      "loss": 0.5491,
+      "step": 6900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.606951876246771e-07,
+      "loss": 0.3716,
+      "step": 6901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.566132476530042e-07,
+      "loss": 0.7122,
+      "step": 6902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.525409265898842e-07,
+      "loss": 0.616,
+      "step": 6903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.484782252323964e-07,
+      "loss": 0.6083,
+      "step": 6904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.444251443757434e-07,
+      "loss": 0.5504,
+      "step": 6905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.40381684813235e-07,
+      "loss": 0.6085,
+      "step": 6906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.363478473363107e-07,
+      "loss": 0.6342,
+      "step": 6907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.323236327345163e-07,
+      "loss": 0.6502,
+      "step": 6908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.283090417955108e-07,
+      "loss": 0.6344,
+      "step": 6909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.243040753050935e-07,
+      "loss": 0.6698,
+      "step": 6910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.203087340471427e-07,
+      "loss": 0.5875,
+      "step": 6911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.163230188036831e-07,
+      "loss": 0.5959,
+      "step": 6912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.123469303548458e-07,
+      "loss": 0.687,
+      "step": 6913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.083804694788754e-07,
+      "loss": 0.6065,
+      "step": 6914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.044236369521285e-07,
+      "loss": 0.683,
+      "step": 6915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.004764335490856e-07,
+      "loss": 0.6246,
+      "step": 6916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.96538860042334e-07,
+      "loss": 0.6043,
+      "step": 6917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.926109172025853e-07,
+      "loss": 0.6769,
+      "step": 6918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.886926057986633e-07,
+      "loss": 0.6673,
+      "step": 6919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.847839265975099e-07,
+      "loss": 0.3725,
+      "step": 6920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.808848803641578e-07,
+      "loss": 0.6931,
+      "step": 6921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.769954678617963e-07,
+      "loss": 0.6913,
+      "step": 6922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.731156898516834e-07,
+      "loss": 0.5801,
+      "step": 6923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.692455470932336e-07,
+      "loss": 0.5894,
+      "step": 6924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.653850403439411e-07,
+      "loss": 0.6183,
+      "step": 6925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.615341703594403e-07,
+      "loss": 0.6175,
+      "step": 6926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.57692937893456e-07,
+      "loss": 0.6107,
+      "step": 6927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.538613436978537e-07,
+      "loss": 0.6248,
+      "step": 6928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.500393885225776e-07,
+      "loss": 0.6476,
+      "step": 6929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.46227073115724e-07,
+      "loss": 0.6178,
+      "step": 6930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.424243982234624e-07,
+      "loss": 0.5777,
+      "step": 6931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.386313645901144e-07,
+      "loss": 0.6288,
+      "step": 6932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.348479729580915e-07,
+      "loss": 0.5855,
+      "step": 6933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.31074224067918e-07,
+      "loss": 0.6616,
+      "step": 6934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.273101186582421e-07,
+      "loss": 0.5816,
+      "step": 6935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.235556574658131e-07,
+      "loss": 0.678,
+      "step": 6936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.198108412254933e-07,
+      "loss": 0.6464,
+      "step": 6937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.160756706702632e-07,
+      "loss": 0.6489,
+      "step": 6938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.123501465312266e-07,
+      "loss": 0.653,
+      "step": 6939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.086342695375669e-07,
+      "loss": 0.6141,
+      "step": 6940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.049280404166137e-07,
+      "loss": 0.6433,
+      "step": 6941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.012314598937752e-07,
+      "loss": 0.6166,
+      "step": 6942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.975445286926063e-07,
+      "loss": 0.5991,
+      "step": 6943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.938672475347463e-07,
+      "loss": 0.6203,
+      "step": 6944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.901996171399583e-07,
+      "loss": 0.6068,
+      "step": 6945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.865416382261069e-07,
+      "loss": 0.62,
+      "step": 6946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.828933115091863e-07,
+      "loss": 0.6176,
+      "step": 6947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.792546377032749e-07,
+      "loss": 0.6497,
+      "step": 6948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.756256175205867e-07,
+      "loss": 0.6277,
+      "step": 6949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.720062516714309e-07,
+      "loss": 0.665,
+      "step": 6950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.683965408642245e-07,
+      "loss": 0.6159,
+      "step": 6951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.647964858055133e-07,
+      "loss": 0.5849,
+      "step": 6952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.612060871999392e-07,
+      "loss": 0.6841,
+      "step": 6953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.576253457502457e-07,
+      "loss": 0.595,
+      "step": 6954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.540542621573109e-07,
+      "loss": 0.6245,
+      "step": 6955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.50492837120098e-07,
+      "loss": 0.5999,
+      "step": 6956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.469410713356938e-07,
+      "loss": 0.662,
+      "step": 6957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.433989654992923e-07,
+      "loss": 0.622,
+      "step": 6958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.398665203041942e-07,
+      "loss": 0.6243,
+      "step": 6959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.363437364418079e-07,
+      "loss": 0.6956,
+      "step": 6960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.328306146016593e-07,
+      "loss": 0.5724,
+      "step": 6961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.293271554713709e-07,
+      "loss": 0.6979,
+      "step": 6962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.258333597366773e-07,
+      "loss": 0.6435,
+      "step": 6963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.223492280814313e-07,
+      "loss": 0.577,
+      "step": 6964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.188747611875934e-07,
+      "loss": 0.6073,
+      "step": 6965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.154099597352081e-07,
+      "loss": 0.5874,
+      "step": 6966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.119548244024609e-07,
+      "loss": 0.6292,
+      "step": 6967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.085093558656164e-07,
+      "loss": 0.6378,
+      "step": 6968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.050735547990794e-07,
+      "loss": 0.6361,
+      "step": 6969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.016474218753288e-07,
+      "loss": 0.6431,
+      "step": 6970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.98230957764978e-07,
+      "loss": 0.374,
+      "step": 6971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.948241631367313e-07,
+      "loss": 0.6419,
+      "step": 6972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.91427038657405e-07,
+      "loss": 0.4103,
+      "step": 6973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.88039584991923e-07,
+      "loss": 0.593,
+      "step": 6974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.846618028033213e-07,
+      "loss": 0.6581,
+      "step": 6975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.812936927527324e-07,
+      "loss": 0.6185,
+      "step": 6976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.779352554994066e-07,
+      "loss": 0.598,
+      "step": 6977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.745864917006904e-07,
+      "loss": 0.5506,
+      "step": 6978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.712474020120484e-07,
+      "loss": 0.5552,
+      "step": 6979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.67917987087041e-07,
+      "loss": 0.6634,
+      "step": 6980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.645982475773415e-07,
+      "loss": 0.6525,
+      "step": 6981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6128818413273e-07,
+      "loss": 0.628,
+      "step": 6982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.579877974010883e-07,
+      "loss": 0.6621,
+      "step": 6983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.546970880283997e-07,
+      "loss": 0.6758,
+      "step": 6984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.514160566587711e-07,
+      "loss": 0.5986,
+      "step": 6985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.48144703934389e-07,
+      "loss": 0.5718,
+      "step": 6986
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.448830304955798e-07,
+      "loss": 0.5764,
+      "step": 6987
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.416310369807331e-07,
+      "loss": 0.702,
+      "step": 6988
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.383887240263841e-07,
+      "loss": 0.3765,
+      "step": 6989
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.351560922671528e-07,
+      "loss": 0.6353,
+      "step": 6990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.31933142335761e-07,
+      "loss": 0.5958,
+      "step": 6991
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.287198748630428e-07,
+      "loss": 0.6179,
+      "step": 6992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.255162904779398e-07,
+      "loss": 0.6373,
+      "step": 6993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.223223898074892e-07,
+      "loss": 0.6283,
+      "step": 6994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.191381734768408e-07,
+      "loss": 0.6421,
+      "step": 6995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.159636421092462e-07,
+      "loss": 0.6003,
+      "step": 6996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.127987963260583e-07,
+      "loss": 0.6015,
+      "step": 6997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.096436367467483e-07,
+      "loss": 0.6232,
+      "step": 6998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.06498163988861e-07,
+      "loss": 0.5696,
+      "step": 6999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.033623786680874e-07,
+      "loss": 0.7008,
+      "step": 7000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.002362813981754e-07,
+      "loss": 0.6192,
+      "step": 7001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.971198727910187e-07,
+      "loss": 0.6187,
+      "step": 7002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.940131534565851e-07,
+      "loss": 0.5797,
+      "step": 7003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.909161240029714e-07,
+      "loss": 0.6018,
+      "step": 7004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.878287850363483e-07,
+      "loss": 0.605,
+      "step": 7005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.847511371610159e-07,
+      "loss": 0.6029,
+      "step": 7006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.816831809793587e-07,
+      "loss": 0.6299,
+      "step": 7007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.786249170918799e-07,
+      "loss": 0.6744,
+      "step": 7008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7557634609717295e-07,
+      "loss": 0.6042,
+      "step": 7009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.725374685919437e-07,
+      "loss": 0.6068,
+      "step": 7010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.695082851709942e-07,
+      "loss": 0.6761,
+      "step": 7011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.66488796427228e-07,
+      "loss": 0.6074,
+      "step": 7012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.634790029516556e-07,
+      "loss": 0.6302,
+      "step": 7013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6047890533339466e-07,
+      "loss": 0.5647,
+      "step": 7014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5748850415964774e-07,
+      "loss": 0.6824,
+      "step": 7015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.54507800015741e-07,
+      "loss": 0.6239,
+      "step": 7016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5153679348508004e-07,
+      "loss": 0.686,
+      "step": 7017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4857548514919966e-07,
+      "loss": 0.6468,
+      "step": 7018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.456238755877085e-07,
+      "loss": 0.6428,
+      "step": 7019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4268196537833317e-07,
+      "loss": 0.6129,
+      "step": 7020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3974975509690207e-07,
+      "loss": 0.6276,
+      "step": 7021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3682724531733945e-07,
+      "loss": 0.3876,
+      "step": 7022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3391443661167097e-07,
+      "loss": 0.5915,
+      "step": 7023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3101132955002396e-07,
+      "loss": 0.6879,
+      "step": 7024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2811792470062726e-07,
+      "loss": 0.372,
+      "step": 7025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.252342226298167e-07,
+      "loss": 0.6008,
+      "step": 7026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2236022390201856e-07,
+      "loss": 0.6184,
+      "step": 7027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1949592907977173e-07,
+      "loss": 0.6233,
+      "step": 7028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.166413387236945e-07,
+      "loss": 0.6688,
+      "step": 7029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.137964533925398e-07,
+      "loss": 0.7012,
+      "step": 7030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.109612736431234e-07,
+      "loss": 0.6887,
+      "step": 7031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0813580003039034e-07,
+      "loss": 0.6526,
+      "step": 7032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0532003310736475e-07,
+      "loss": 0.5531,
+      "step": 7033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.025139734251948e-07,
+      "loss": 0.6789,
+      "step": 7034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9971762153310775e-07,
+      "loss": 0.6299,
+      "step": 7035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9693097797843247e-07,
+      "loss": 0.6462,
+      "step": 7036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9415404330661044e-07,
+      "loss": 0.5878,
+      "step": 7037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.913868180611735e-07,
+      "loss": 0.5658,
+      "step": 7038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8862930278374954e-07,
+      "loss": 0.6475,
+      "step": 7039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.85881498014079e-07,
+      "loss": 0.6432,
+      "step": 7040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8314340428999284e-07,
+      "loss": 0.6479,
+      "step": 7041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.804150221474179e-07,
+      "loss": 0.636,
+      "step": 7042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7769635212038244e-07,
+      "loss": 0.6455,
+      "step": 7043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.749873947410221e-07,
+      "loss": 0.6048,
+      "step": 7044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7228815053956255e-07,
+      "loss": 0.6648,
+      "step": 7045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.695986200443313e-07,
+      "loss": 0.6217,
+      "step": 7046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.669188037817517e-07,
+      "loss": 0.6125,
+      "step": 7047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.642487022763541e-07,
+      "loss": 0.5719,
+      "step": 7048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.615883160507538e-07,
+      "loss": 0.6482,
+      "step": 7049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5893764562568433e-07,
+      "loss": 0.5827,
+      "step": 7050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5629669151994725e-07,
+      "loss": 0.5526,
+      "step": 7051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.536654542504736e-07,
+      "loss": 0.5747,
+      "step": 7052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.510439343322791e-07,
+      "loss": 0.5643,
+      "step": 7053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4843213227847003e-07,
+      "loss": 0.6523,
+      "step": 7054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4583004860026523e-07,
+      "loss": 0.5948,
+      "step": 7055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4323768380697396e-07,
+      "loss": 0.5995,
+      "step": 7056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4065503840600145e-07,
+      "loss": 0.6904,
+      "step": 7057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.380821129028489e-07,
+      "loss": 0.6173,
+      "step": 7058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3551890780113004e-07,
+      "loss": 0.6604,
+      "step": 7059
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3296542360253256e-07,
+      "loss": 0.6427,
+      "step": 7060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.304216608068622e-07,
+      "loss": 0.5692,
+      "step": 7061
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.278876199120096e-07,
+      "loss": 0.6492,
+      "step": 7062
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2536330141396677e-07,
+      "loss": 0.682,
+      "step": 7063
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.228487058068275e-07,
+      "loss": 0.3449,
+      "step": 7064
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2034383358276465e-07,
+      "loss": 0.5537,
+      "step": 7065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1784868523206945e-07,
+      "loss": 0.598,
+      "step": 7066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1536326124312896e-07,
+      "loss": 0.6395,
+      "step": 7067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.128875621024041e-07,
+      "loss": 0.6513,
+      "step": 7068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1042158829447385e-07,
+      "loss": 0.5776,
+      "step": 7069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.079653403020022e-07,
+      "loss": 0.6407,
+      "step": 7070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0551881860576556e-07,
+      "loss": 0.7305,
+      "step": 7071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0308202368461434e-07,
+      "loss": 0.6112,
+      "step": 7072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.006549560155114e-07,
+      "loss": 0.6181,
+      "step": 7073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9823761607350455e-07,
+      "loss": 0.6456,
+      "step": 7074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.958300043317541e-07,
+      "loss": 0.5756,
+      "step": 7075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9343212126149434e-07,
+      "loss": 0.6437,
+      "step": 7076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9104396733207194e-07,
+      "loss": 0.6008,
+      "step": 7077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8866554301091866e-07,
+      "loss": 0.6233,
+      "step": 7078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8629684876357867e-07,
+      "loss": 0.3386,
+      "step": 7079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8393788505367e-07,
+      "loss": 0.6311,
+      "step": 7080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8158865234291785e-07,
+      "loss": 0.3583,
+      "step": 7081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7924915109114324e-07,
+      "loss": 0.6234,
+      "step": 7082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7691938175625764e-07,
+      "loss": 0.6548,
+      "step": 7083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7459934479427406e-07,
+      "loss": 0.5823,
+      "step": 7084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7228904065929595e-07,
+      "loss": 0.6506,
+      "step": 7085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.699884698035171e-07,
+      "loss": 0.6774,
+      "step": 7086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6769763267723845e-07,
+      "loss": 0.6111,
+      "step": 7087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6541652972884576e-07,
+      "loss": 0.6496,
+      "step": 7088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.631451614048208e-07,
+      "loss": 0.6183,
+      "step": 7089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6088352814974127e-07,
+      "loss": 0.6254,
+      "step": 7090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.586316304062919e-07,
+      "loss": 0.6558,
+      "step": 7091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.563894686152257e-07,
+      "loss": 0.6137,
+      "step": 7092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.541570432154139e-07,
+      "loss": 0.648,
+      "step": 7093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5193435464380134e-07,
+      "loss": 0.6167,
+      "step": 7094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4972140333545114e-07,
+      "loss": 0.6629,
+      "step": 7095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4751818972350016e-07,
+      "loss": 0.6677,
+      "step": 7096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.453247142391868e-07,
+      "loss": 0.6142,
+      "step": 7097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.431409773118454e-07,
+      "loss": 0.6316,
+      "step": 7098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4096697936890066e-07,
+      "loss": 0.6342,
+      "step": 7099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.388027208358734e-07,
+      "loss": 0.6541,
+      "step": 7100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3664820213638028e-07,
+      "loss": 0.6331,
+      "step": 7101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3450342369212287e-07,
+      "loss": 0.7008,
+      "step": 7102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3236838592289867e-07,
+      "loss": 0.7404,
+      "step": 7103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.302430892466123e-07,
+      "loss": 0.6383,
+      "step": 7104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.281275340792477e-07,
+      "loss": 0.6655,
+      "step": 7105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2602172083487916e-07,
+      "loss": 0.6251,
+      "step": 7106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.239256499256881e-07,
+      "loss": 0.6001,
+      "step": 7107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.218393217619408e-07,
+      "loss": 0.6053,
+      "step": 7108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1976273675199387e-07,
+      "loss": 0.3866,
+      "step": 7109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1769589530230006e-07,
+      "loss": 0.6574,
+      "step": 7110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1563879781741347e-07,
+      "loss": 0.5709,
+      "step": 7111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1359144469996205e-07,
+      "loss": 0.6374,
+      "step": 7112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1155383635068636e-07,
+      "loss": 0.6847,
+      "step": 7113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.095259731684007e-07,
+      "loss": 0.5896,
+      "step": 7114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.075078555500265e-07,
+      "loss": 0.6353,
+      "step": 7115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0549948389058104e-07,
+      "loss": 0.5803,
+      "step": 7116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0350085858315548e-07,
+      "loss": 0.6225,
+      "step": 7117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0151198001894245e-07,
+      "loss": 0.6349,
+      "step": 7118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9953284858723608e-07,
+      "loss": 0.6581,
+      "step": 7119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.975634646754154e-07,
+      "loss": 0.5955,
+      "step": 7120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9560382866893878e-07,
+      "loss": 0.6507,
+      "step": 7121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9365394095138268e-07,
+      "loss": 0.6736,
+      "step": 7122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.917138019043918e-07,
+      "loss": 0.6122,
+      "step": 7123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8978341190771797e-07,
+      "loss": 0.6834,
+      "step": 7124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8786277133919782e-07,
+      "loss": 0.6588,
+      "step": 7125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8595188057475843e-07,
+      "loss": 0.5539,
+      "step": 7126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8405073998842836e-07,
+      "loss": 0.6057,
+      "step": 7127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8215934995231e-07,
+      "loss": 0.6359,
+      "step": 7128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8027771083661826e-07,
+      "loss": 0.3528,
+      "step": 7129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7840582300964192e-07,
+      "loss": 0.6608,
+      "step": 7130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7654368683777678e-07,
+      "loss": 0.6593,
+      "step": 7131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7469130268549238e-07,
+      "loss": 0.6685,
+      "step": 7132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7284867091535982e-07,
+      "loss": 0.621,
+      "step": 7133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7101579188804616e-07,
+      "loss": 0.6854,
+      "step": 7134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6919266596229777e-07,
+      "loss": 0.6043,
+      "step": 7135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6737929349496805e-07,
+      "loss": 0.6053,
+      "step": 7136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6557567484097315e-07,
+      "loss": 0.6931,
+      "step": 7137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6378181035335282e-07,
+      "loss": 0.5994,
+      "step": 7138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.619977003832207e-07,
+      "loss": 0.6663,
+      "step": 7139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6022334527977522e-07,
+      "loss": 0.6447,
+      "step": 7140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5845874539032192e-07,
+      "loss": 0.6314,
+      "step": 7141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5670390106024557e-07,
+      "loss": 0.6316,
+      "step": 7142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5495881263302148e-07,
+      "loss": 0.6398,
+      "step": 7143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5322348045022638e-07,
+      "loss": 0.5874,
+      "step": 7144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5149790485151083e-07,
+      "loss": 0.6227,
+      "step": 7145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4978208617463242e-07,
+      "loss": 0.6031,
+      "step": 7146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4807602475542249e-07,
+      "loss": 0.6422,
+      "step": 7147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.463797209278195e-07,
+      "loss": 0.6249,
+      "step": 7148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4469317502383562e-07,
+      "loss": 0.6544,
+      "step": 7149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4301638737358459e-07,
+      "loss": 0.5996,
+      "step": 7150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4134935830527052e-07,
+      "loss": 0.6069,
+      "step": 7151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3969208814517687e-07,
+      "loss": 0.673,
+      "step": 7152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3804457721768304e-07,
+      "loss": 0.5908,
+      "step": 7153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3640682584526443e-07,
+      "loss": 0.6392,
+      "step": 7154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3477883434848126e-07,
+      "loss": 0.6163,
+      "step": 7155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.331606030459731e-07,
+      "loss": 0.6574,
+      "step": 7156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3155213225449215e-07,
+      "loss": 0.5616,
+      "step": 7157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2995342228885877e-07,
+      "loss": 0.6207,
+      "step": 7158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.283644734619893e-07,
+      "loss": 0.6246,
+      "step": 7159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.267852860848906e-07,
+      "loss": 0.6809,
+      "step": 7160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2521586046666533e-07,
+      "loss": 0.641,
+      "step": 7161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2365619691450114e-07,
+      "loss": 0.5878,
+      "step": 7162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2210629573366495e-07,
+      "loss": 0.6875,
+      "step": 7163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2056615722752518e-07,
+      "loss": 0.6381,
+      "step": 7164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.190357816975407e-07,
+      "loss": 0.5806,
+      "step": 7165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.175151694432386e-07,
+      "loss": 0.7066,
+      "step": 7166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1600432076226964e-07,
+      "loss": 0.6193,
+      "step": 7167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1450323595034174e-07,
+      "loss": 0.5961,
+      "step": 7168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1301191530126986e-07,
+      "loss": 0.6599,
+      "step": 7169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1153035910695387e-07,
+      "loss": 0.6145,
+      "step": 7170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1005856765737288e-07,
+      "loss": 0.621,
+      "step": 7171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0859654124061313e-07,
+      "loss": 0.6258,
+      "step": 7172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0714428014282907e-07,
+      "loss": 0.6184,
+      "step": 7173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0570178464828217e-07,
+      "loss": 0.7208,
+      "step": 7174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0426905503931328e-07,
+      "loss": 0.5607,
+      "step": 7175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0284609159634806e-07,
+      "loss": 0.659,
+      "step": 7176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0143289459790816e-07,
+      "loss": 0.628,
+      "step": 7177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0002946432060012e-07,
+      "loss": 0.6653,
+      "step": 7178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.863580103912084e-08,
+      "loss": 0.6073,
+      "step": 7179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.725190502625215e-08,
+      "loss": 0.5659,
+      "step": 7180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.587777655286623e-08,
+      "loss": 0.5759,
+      "step": 7181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.451341588792573e-08,
+      "loss": 0.619,
+      "step": 7182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.315882329847259e-08,
+      "loss": 0.6444,
+      "step": 7183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.181399904965582e-08,
+      "loss": 0.6195,
+      "step": 7184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.04789434046871e-08,
+      "loss": 0.6189,
+      "step": 7185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.915365662488518e-08,
+      "loss": 0.6629,
+      "step": 7186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.783813896964809e-08,
+      "loss": 0.6199,
+      "step": 7187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.653239069646435e-08,
+      "loss": 0.633,
+      "step": 7188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.523641206091282e-08,
+      "loss": 0.6251,
+      "step": 7189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.395020331665172e-08,
+      "loss": 0.643,
+      "step": 7190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.26737647154352e-08,
+      "loss": 0.6795,
+      "step": 7191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.140709650710787e-08,
+      "loss": 0.6313,
+      "step": 7192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.015019893959364e-08,
+      "loss": 0.6758,
+      "step": 7193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.890307225890126e-08,
+      "loss": 0.6487,
+      "step": 7194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.766571670913547e-08,
+      "loss": 0.609,
+      "step": 7195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.643813253249143e-08,
+      "loss": 0.5899,
+      "step": 7196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.522031996924362e-08,
+      "loss": 0.3889,
+      "step": 7197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.401227925774578e-08,
+      "loss": 0.6936,
+      "step": 7198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.281401063446435e-08,
+      "loss": 0.6364,
+      "step": 7199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.162551433392839e-08,
+      "loss": 0.6337,
+      "step": 7200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.044679058877402e-08,
+      "loss": 0.6773,
+      "step": 7201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.927783962970002e-08,
+      "loss": 0.6889,
+      "step": 7202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.811866168552338e-08,
+      "loss": 0.5967,
+      "step": 7203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.696925698311817e-08,
+      "loss": 0.3452,
+      "step": 7204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.582962574746554e-08,
+      "loss": 0.3916,
+      "step": 7205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.469976820162593e-08,
+      "loss": 0.6494,
+      "step": 7206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.357968456675023e-08,
+      "loss": 0.6286,
+      "step": 7207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.246937506206863e-08,
+      "loss": 0.613,
+      "step": 7208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.136883990491282e-08,
+      "loss": 0.5746,
+      "step": 7209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.02780793106772e-08,
+      "loss": 0.6402,
+      "step": 7210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.919709349287428e-08,
+      "loss": 0.6974,
+      "step": 7211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.812588266307928e-08,
+      "loss": 0.5853,
+      "step": 7212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.706444703096336e-08,
+      "loss": 0.7505,
+      "step": 7213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.601278680427702e-08,
+      "loss": 0.5799,
+      "step": 7214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4970902188872245e-08,
+      "loss": 0.6146,
+      "step": 7215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.39387933886748e-08,
+      "loss": 0.6986,
+      "step": 7216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.291646060570088e-08,
+      "loss": 0.6222,
+      "step": 7217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.190390404005152e-08,
+      "loss": 0.6336,
+      "step": 7218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.090112388992374e-08,
+      "loss": 0.6156,
+      "step": 7219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9908120351577215e-08,
+      "loss": 0.572,
+      "step": 7220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.892489361938979e-08,
+      "loss": 0.6984,
+      "step": 7221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7951443885807524e-08,
+      "loss": 0.5846,
+      "step": 7222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.698777134135579e-08,
+      "loss": 0.5996,
+      "step": 7223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.603387617466148e-08,
+      "loss": 0.6173,
+      "step": 7224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.508975857243636e-08,
+      "loss": 0.6397,
+      "step": 7225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4155418719465936e-08,
+      "loss": 0.6191,
+      "step": 7226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3230856798637255e-08,
+      "loss": 0.6584,
+      "step": 7227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.231607299091112e-08,
+      "loss": 0.6327,
+      "step": 7228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.141106747533874e-08,
+      "loss": 0.6388,
+      "step": 7229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.051584042906176e-08,
+      "loss": 0.6606,
+      "step": 7230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.963039202730112e-08,
+      "loss": 0.5665,
+      "step": 7231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.875472244337375e-08,
+      "loss": 0.6474,
+      "step": 7232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.788883184866476e-08,
+      "loss": 0.583,
+      "step": 7233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.703272041266637e-08,
+      "loss": 0.637,
+      "step": 7234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.618638830294452e-08,
+      "loss": 0.6862,
+      "step": 7235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5349835685144496e-08,
+      "loss": 0.651,
+      "step": 7236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.452306272301864e-08,
+      "loss": 0.6176,
+      "step": 7237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.370606957838196e-08,
+      "loss": 0.3672,
+      "step": 7238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.289885641115653e-08,
+      "loss": 0.6767,
+      "step": 7239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.210142337932709e-08,
+      "loss": 0.5994,
+      "step": 7240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.131377063898544e-08,
+      "loss": 0.6821,
+      "step": 7241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.05358983442916e-08,
+      "loss": 0.5806,
+      "step": 7242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9767806647512662e-08,
+      "loss": 0.3345,
+      "step": 7243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9009495698972823e-08,
+      "loss": 0.6057,
+      "step": 7244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8260965647108895e-08,
+      "loss": 0.6523,
+      "step": 7245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7522216638431463e-08,
+      "loss": 0.5829,
+      "step": 7246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6793248817530426e-08,
+      "loss": 0.3998,
+      "step": 7247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.607406232709164e-08,
+      "loss": 0.6286,
+      "step": 7248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.536465730788029e-08,
+      "loss": 0.6333,
+      "step": 7249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4665033898751965e-08,
+      "loss": 0.6042,
+      "step": 7250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3975192236647125e-08,
+      "loss": 0.6433,
+      "step": 7251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3295132456579995e-08,
+      "loss": 0.6619,
+      "step": 7252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.262485469167186e-08,
+      "loss": 0.6178,
+      "step": 7253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.196435907310668e-08,
+      "loss": 0.6156,
+      "step": 7254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1313645730169917e-08,
+      "loss": 0.606,
+      "step": 7255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0672714790226366e-08,
+      "loss": 0.6599,
+      "step": 7256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0041566378725675e-08,
+      "loss": 0.6434,
+      "step": 7257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9420200619207907e-08,
+      "loss": 0.5837,
+      "step": 7258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8808617633281344e-08,
+      "loss": 0.6074,
+      "step": 7259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8206817540666887e-08,
+      "loss": 0.5877,
+      "step": 7260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7614800459148094e-08,
+      "loss": 0.6017,
+      "step": 7261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7032566504598946e-08,
+      "loss": 0.6418,
+      "step": 7262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.646011579098383e-08,
+      "loss": 0.6402,
+      "step": 7263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5897448430352014e-08,
+      "loss": 0.6416,
+      "step": 7264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.534456453283206e-08,
+      "loss": 0.6734,
+      "step": 7265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4801464206642967e-08,
+      "loss": 0.5882,
+      "step": 7266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4268147558088585e-08,
+      "loss": 0.6864,
+      "step": 7267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3744614691546531e-08,
+      "loss": 0.6687,
+      "step": 7268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3230865709501494e-08,
+      "loss": 0.6516,
+      "step": 7269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2726900712495271e-08,
+      "loss": 0.6801,
+      "step": 7270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.223271979918228e-08,
+      "loss": 0.67,
+      "step": 7271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.174832306628515e-08,
+      "loss": 0.5918,
+      "step": 7272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1273710608616928e-08,
+      "loss": 0.6029,
+      "step": 7273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0808882519069974e-08,
+      "loss": 0.6034,
+      "step": 7274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0353838888632617e-08,
+      "loss": 0.6217,
+      "step": 7275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.908579806361396e-09,
+      "loss": 0.6345,
+      "step": 7276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.47310535941992e-09,
+      "loss": 0.6574,
+      "step": 7277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.047415633034462e-09,
+      "loss": 0.3559,
+      "step": 7278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.631510710527257e-09,
+      "loss": 0.6304,
+      "step": 7279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.22539067330541e-09,
+      "loss": 0.6902,
+      "step": 7280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.829055600860891e-09,
+      "loss": 0.7161,
+      "step": 7281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.4425055707705304e-09,
+      "loss": 0.7125,
+      "step": 7282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.065740658690478e-09,
+      "loss": 0.39,
+      "step": 7283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.6987609383617475e-09,
+      "loss": 0.5595,
+      "step": 7284
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.341566481626871e-09,
+      "loss": 0.6082,
+      "step": 7285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.994157358385488e-09,
+      "loss": 0.6169,
+      "step": 7286
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.656533636644312e-09,
+      "loss": 0.6276,
+      "step": 7287
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.328695382489368e-09,
+      "loss": 0.6286,
+      "step": 7288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.010642660080444e-09,
+      "loss": 0.5617,
+      "step": 7289
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.702375531678849e-09,
+      "loss": 0.6217,
+      "step": 7290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.403894057614099e-09,
+      "loss": 0.6484,
+      "step": 7291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.115198296317235e-09,
+      "loss": 0.6285,
+      "step": 7292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.836288304287505e-09,
+      "loss": 0.688,
+      "step": 7293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.567164136120127e-09,
+      "loss": 0.5711,
+      "step": 7294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3078258444951826e-09,
+      "loss": 0.7037,
+      "step": 7295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.058273480160967e-09,
+      "loss": 0.668,
+      "step": 7296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8185070919783953e-09,
+      "loss": 0.6403,
+      "step": 7297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.588526726865492e-09,
+      "loss": 0.5525,
+      "step": 7298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.368332429841802e-09,
+      "loss": 0.6265,
+      "step": 7299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1579242440006307e-09,
+      "loss": 0.6289,
+      "step": 7300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9573022105368042e-09,
+      "loss": 0.6521,
+      "step": 7301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7664663687078086e-09,
+      "loss": 0.6171,
+      "step": 7302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5854167558670975e-09,
+      "loss": 0.6463,
+      "step": 7303
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4141534074585405e-09,
+      "loss": 0.6229,
+      "step": 7304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2526763569997712e-09,
+      "loss": 0.6564,
+      "step": 7305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1009856360988392e-09,
+      "loss": 0.5543,
+      "step": 7306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.590812744431076e-10,
+      "loss": 0.6204,
+      "step": 7307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.26963299804806e-10,
+      "loss": 0.632,
+      "step": 7308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.046317380521306e-10,
+      "loss": 0.6242,
+      "step": 7309
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.920866131270409e-10,
+      "loss": 0.6092,
+      "step": 7310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.893279470508105e-10,
+      "loss": 0.6664,
+      "step": 7311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9635575994623196e-10,
+      "loss": 0.624,
+      "step": 7312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.131700700043094e-10,
+      "loss": 0.631,
+      "step": 7313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3977089351201467e-10,
+      "loss": 0.6405,
+      "step": 7314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7615824483008248e-10,
+      "loss": 0.633,
+      "step": 7315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2233213641521525e-10,
+      "loss": 0.6453,
+      "step": 7316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.829257879787832e-11,
+      "loss": 0.6612,
+      "step": 7317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.403958060450464e-11,
+      "loss": 0.6424,
+      "step": 7318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9573148529739015e-11,
+      "loss": 0.6077,
+      "step": 7319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.893287369744925e-12,
+      "loss": 0.3802,
+      "step": 7320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.4105,
+      "step": 7321
+    },
+    {
+      "epoch": 1.0,
+      "step": 7321,
+      "total_flos": 3.320275842419917e+16,
+      "train_loss": 0.6619515117889316,
+      "train_runtime": 159508.9616,
+      "train_samples_per_second": 5.875,
+      "train_steps_per_second": 0.046
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7321,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 3.320275842419917e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}