diff --git "a/Chinese_tiny_llm-zh_105k_en_26k/trainer_state.json" "b/Chinese_tiny_llm-zh_105k_en_26k/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/Chinese_tiny_llm-zh_105k_en_26k/trainer_state.json"
@@ -0,0 +1,18489 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.997808619430241,
+  "eval_steps": 500,
+  "global_step": 3078,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.298701298701299e-07,
+      "loss": 2.2406,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.597402597402598e-07,
+      "loss": 2.4113,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.896103896103896e-07,
+      "loss": 2.2121,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.194805194805196e-07,
+      "loss": 2.3546,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.493506493506493e-07,
+      "loss": 2.1397,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.792207792207792e-07,
+      "loss": 2.4503,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 2.2442,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0389610389610392e-06,
+      "loss": 2.2086,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.168831168831169e-06,
+      "loss": 2.2294,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2987012987012986e-06,
+      "loss": 2.5848,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 2.0445,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5584415584415584e-06,
+      "loss": 2.2522,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6883116883116885e-06,
+      "loss": 1.9668,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 2.528,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9480519480519483e-06,
+      "loss": 2.1507,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0779220779220784e-06,
+      "loss": 2.3257,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.207792207792208e-06,
+      "loss": 1.9621,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.337662337662338e-06,
+      "loss": 2.1586,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4675324675324676e-06,
+      "loss": 1.9939,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.597402597402597e-06,
+      "loss": 2.1574,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 2.1186,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.99,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9870129870129873e-06,
+      "loss": 2.1305,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.116883116883117e-06,
+      "loss": 1.9739,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.246753246753247e-06,
+      "loss": 2.1008,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.376623376623377e-06,
+      "loss": 2.1598,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.506493506493507e-06,
+      "loss": 1.9269,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 2.0129,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7662337662337666e-06,
+      "loss": 1.8318,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.896103896103897e-06,
+      "loss": 1.8116,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 1.9135,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.155844155844157e-06,
+      "loss": 1.9253,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 1.7538,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.415584415584416e-06,
+      "loss": 1.8115,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 1.6623,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.675324675324676e-06,
+      "loss": 1.9018,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.805194805194806e-06,
+      "loss": 1.9084,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935064935064935e-06,
+      "loss": 1.566,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.064935064935065e-06,
+      "loss": 1.9235,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.194805194805194e-06,
+      "loss": 1.9382,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.324675324675325e-06,
+      "loss": 2.0483,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 1.9039,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.584415584415585e-06,
+      "loss": 1.9822,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.9103,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.844155844155844e-06,
+      "loss": 1.7934,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9740259740259746e-06,
+      "loss": 1.9374,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.103896103896104e-06,
+      "loss": 1.9101,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.233766233766234e-06,
+      "loss": 1.9438,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 2.075,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.493506493506494e-06,
+      "loss": 1.7601,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.623376623376624e-06,
+      "loss": 1.6741,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.753246753246754e-06,
+      "loss": 1.9251,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.8831168831168835e-06,
+      "loss": 1.8838,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.012987012987014e-06,
+      "loss": 1.87,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 1.8502,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.7868,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.402597402597404e-06,
+      "loss": 1.896,
+      "step": 57
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.532467532467533e-06,
+      "loss": 2.0506,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.662337662337663e-06,
+      "loss": 1.8279,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.792207792207793e-06,
+      "loss": 1.8367,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.922077922077924e-06,
+      "loss": 1.7444,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.051948051948052e-06,
+      "loss": 1.8889,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 1.7638,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.311688311688313e-06,
+      "loss": 1.874,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.441558441558442e-06,
+      "loss": 2.1087,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.8607,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.701298701298701e-06,
+      "loss": 2.1034,
+      "step": 67
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.831168831168832e-06,
+      "loss": 1.8915,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.96103896103896e-06,
+      "loss": 1.8137,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 1.9592,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.220779220779221e-06,
+      "loss": 1.6891,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.350649350649352e-06,
+      "loss": 1.9853,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.48051948051948e-06,
+      "loss": 1.9133,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.610389610389611e-06,
+      "loss": 1.7308,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.740259740259742e-06,
+      "loss": 1.8777,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87012987012987e-06,
+      "loss": 2.0242,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 1.6466,
+      "step": 77
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.012987012987013e-05,
+      "loss": 1.7858,
+      "step": 78
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.025974025974026e-05,
+      "loss": 1.9504,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0389610389610389e-05,
+      "loss": 1.9337,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0519480519480521e-05,
+      "loss": 1.8321,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.064935064935065e-05,
+      "loss": 1.883,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.077922077922078e-05,
+      "loss": 1.7848,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 1.912,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1038961038961041e-05,
+      "loss": 2.003,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.116883116883117e-05,
+      "loss": 1.9783,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.12987012987013e-05,
+      "loss": 2.0743,
+      "step": 87
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.8282,
+      "step": 88
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.155844155844156e-05,
+      "loss": 1.7178,
+      "step": 89
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1688311688311688e-05,
+      "loss": 1.8958,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.181818181818182e-05,
+      "loss": 1.6794,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1948051948051949e-05,
+      "loss": 1.827,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.207792207792208e-05,
+      "loss": 1.9995,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2207792207792208e-05,
+      "loss": 1.8033,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2337662337662339e-05,
+      "loss": 1.8054,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2467532467532468e-05,
+      "loss": 1.9788,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.25974025974026e-05,
+      "loss": 1.7173,
+      "step": 97
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 1.7239,
+      "step": 98
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 1.813,
+      "step": 99
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2987012987012988e-05,
+      "loss": 1.8267,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.311688311688312e-05,
+      "loss": 1.6709,
+      "step": 101
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3246753246753249e-05,
+      "loss": 1.7046,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3376623376623377e-05,
+      "loss": 1.9265,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3506493506493508e-05,
+      "loss": 1.5906,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 1.5866,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 1.5902,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3896103896103896e-05,
+      "loss": 1.773,
+      "step": 107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4025974025974028e-05,
+      "loss": 1.6751,
+      "step": 108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4155844155844157e-05,
+      "loss": 1.8722,
+      "step": 109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 1.7465,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4415584415584416e-05,
+      "loss": 1.6948,
+      "step": 111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 1.6096,
+      "step": 112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4675324675324675e-05,
+      "loss": 1.8475,
+      "step": 113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4805194805194807e-05,
+      "loss": 1.7179,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4935064935064936e-05,
+      "loss": 1.6387,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5064935064935066e-05,
+      "loss": 1.9,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5194805194805195e-05,
+      "loss": 1.8845,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5324675324675326e-05,
+      "loss": 2.0451,
+      "step": 118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 1.6355,
+      "step": 119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5584415584415587e-05,
+      "loss": 1.5209,
+      "step": 120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 1.6665,
+      "step": 121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5844155844155847e-05,
+      "loss": 1.6792,
+      "step": 122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5974025974025976e-05,
+      "loss": 1.8535,
+      "step": 123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6103896103896105e-05,
+      "loss": 1.8401,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6233766233766234e-05,
+      "loss": 1.7943,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 1.9218,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6493506493506495e-05,
+      "loss": 1.6924,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6623376623376627e-05,
+      "loss": 1.7824,
+      "step": 128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6753246753246756e-05,
+      "loss": 1.8186,
+      "step": 129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6883116883116884e-05,
+      "loss": 1.6247,
+      "step": 130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 1.8899,
+      "step": 131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 1.9079,
+      "step": 132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 1.781,
+      "step": 133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7402597402597403e-05,
+      "loss": 1.8074,
+      "step": 134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7532467532467535e-05,
+      "loss": 1.9233,
+      "step": 135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7662337662337664e-05,
+      "loss": 1.6956,
+      "step": 136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7792207792207792e-05,
+      "loss": 1.585,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.792207792207792e-05,
+      "loss": 1.9442,
+      "step": 138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8051948051948053e-05,
+      "loss": 1.6524,
+      "step": 139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 1.7103,
+      "step": 140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8311688311688314e-05,
+      "loss": 1.9247,
+      "step": 141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8441558441558443e-05,
+      "loss": 1.8547,
+      "step": 142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 1.8386,
+      "step": 143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8701298701298704e-05,
+      "loss": 2.0386,
+      "step": 144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8831168831168833e-05,
+      "loss": 2.2138,
+      "step": 145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.896103896103896e-05,
+      "loss": 1.8287,
+      "step": 146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9090909090909094e-05,
+      "loss": 2.0138,
+      "step": 147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9220779220779222e-05,
+      "loss": 1.5662,
+      "step": 148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9350649350649354e-05,
+      "loss": 1.6956,
+      "step": 149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9480519480519483e-05,
+      "loss": 1.5349,
+      "step": 150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9610389610389612e-05,
+      "loss": 1.8099,
+      "step": 151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.974025974025974e-05,
+      "loss": 1.9258,
+      "step": 152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9870129870129873e-05,
+      "loss": 1.8413,
+      "step": 153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 1.9972,
+      "step": 154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999422815086e-05,
+      "loss": 1.6756,
+      "step": 155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999976912610102e-05,
+      "loss": 1.7794,
+      "step": 156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999948053397712e-05,
+      "loss": 1.9985,
+      "step": 157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999907650547006e-05,
+      "loss": 1.8532,
+      "step": 158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999855704104626e-05,
+      "loss": 1.5329,
+      "step": 159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999792214130535e-05,
+      "loss": 1.9029,
+      "step": 160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999717180698027e-05,
+      "loss": 1.8229,
+      "step": 161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999963060389371e-05,
+      "loss": 1.7951,
+      "step": 162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999532483817537e-05,
+      "loss": 1.73,
+      "step": 163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999942282058277e-05,
+      "loss": 1.7605,
+      "step": 164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999301614315994e-05,
+      "loss": 1.8647,
+      "step": 165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999168865157137e-05,
+      "loss": 1.6823,
+      "step": 166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999024573259432e-05,
+      "loss": 1.6987,
+      "step": 167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999886873878945e-05,
+      "loss": 1.7789,
+      "step": 168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9998701361927083e-05,
+      "loss": 1.777,
+      "step": 169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999852244286554e-05,
+      "loss": 1.7297,
+      "step": 170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999833198181137e-05,
+      "loss": 1.6872,
+      "step": 171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999812997898442e-05,
+      "loss": 1.7286,
+      "step": 172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997916434617888e-05,
+      "loss": 1.9051,
+      "step": 173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997691348958278e-05,
+      "loss": 1.9275,
+      "step": 174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997454722265428e-05,
+      "loss": 1.8241,
+      "step": 175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997206554812485e-05,
+      "loss": 1.5531,
+      "step": 176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999694684688593e-05,
+      "loss": 1.6342,
+      "step": 177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999667559878556e-05,
+      "loss": 1.8113,
+      "step": 178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99963928108245e-05,
+      "loss": 1.8144,
+      "step": 179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999609848332919e-05,
+      "loss": 2.0104,
+      "step": 180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999579261663939e-05,
+      "loss": 2.0157,
+      "step": 181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9995475211108183e-05,
+      "loss": 1.7675,
+      "step": 182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999514626710198e-05,
+      "loss": 1.8599,
+      "step": 183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9994805785000493e-05,
+      "loss": 1.8244,
+      "step": 184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9994453765196776e-05,
+      "loss": 1.9681,
+      "step": 185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9994090208097176e-05,
+      "loss": 1.6771,
+      "step": 186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9993715114121385e-05,
+      "loss": 1.6934,
+      "step": 187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9993328483702393e-05,
+      "loss": 1.6775,
+      "step": 188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999293031728652e-05,
+      "loss": 1.6941,
+      "step": 189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9992520615333393e-05,
+      "loss": 1.996,
+      "step": 190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999209937831596e-05,
+      "loss": 1.822,
+      "step": 191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9991666606720485e-05,
+      "loss": 1.7635,
+      "step": 192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9991222301046547e-05,
+      "loss": 1.6179,
+      "step": 193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9990766461807037e-05,
+      "loss": 2.0408,
+      "step": 194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9990299089528165e-05,
+      "loss": 1.7237,
+      "step": 195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998982018474945e-05,
+      "loss": 1.9914,
+      "step": 196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9989329748023728e-05,
+      "loss": 1.918,
+      "step": 197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9988827779917138e-05,
+      "loss": 1.6953,
+      "step": 198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998831428100914e-05,
+      "loss": 1.7982,
+      "step": 199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9987789251892503e-05,
+      "loss": 1.6849,
+      "step": 200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9987252693173305e-05,
+      "loss": 1.6538,
+      "step": 201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9986704605470932e-05,
+      "loss": 1.6328,
+      "step": 202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998614498941808e-05,
+      "loss": 1.7342,
+      "step": 203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998557384566075e-05,
+      "loss": 2.0404,
+      "step": 204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998499117485826e-05,
+      "loss": 1.8988,
+      "step": 205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9984396977683223e-05,
+      "loss": 1.6775,
+      "step": 206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998379125482156e-05,
+      "loss": 1.6944,
+      "step": 207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9983174006972502e-05,
+      "loss": 1.5753,
+      "step": 208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9982545234848586e-05,
+      "loss": 1.8604,
+      "step": 209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998190493917564e-05,
+      "loss": 1.7701,
+      "step": 210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9981253120692806e-05,
+      "loss": 1.5181,
+      "step": 211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9980589780152523e-05,
+      "loss": 1.9073,
+      "step": 212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9979914918320527e-05,
+      "loss": 1.7607,
+      "step": 213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9979228535975866e-05,
+      "loss": 1.7895,
+      "step": 214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9978530633910873e-05,
+      "loss": 1.7136,
+      "step": 215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997782121293119e-05,
+      "loss": 1.8586,
+      "step": 216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9977100273855746e-05,
+      "loss": 1.662,
+      "step": 217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9976367817516773e-05,
+      "loss": 1.7398,
+      "step": 218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99756238447598e-05,
+      "loss": 1.6921,
+      "step": 219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997486835644364e-05,
+      "loss": 1.7455,
+      "step": 220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9974101353440413e-05,
+      "loss": 1.9404,
+      "step": 221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9973322836635517e-05,
+      "loss": 2.0421,
+      "step": 222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9972532806927655e-05,
+      "loss": 1.7439,
+      "step": 223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9971731265228813e-05,
+      "loss": 1.8198,
+      "step": 224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9970918212464257e-05,
+      "loss": 1.7568,
+      "step": 225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9970093649572567e-05,
+      "loss": 1.6897,
+      "step": 226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9969257577505577e-05,
+      "loss": 1.8382,
+      "step": 227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9968409997228434e-05,
+      "loss": 1.8496,
+      "step": 228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9967550909719557e-05,
+      "loss": 1.8856,
+      "step": 229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9966680315970647e-05,
+      "loss": 1.7844,
+      "step": 230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9965798216986692e-05,
+      "loss": 1.7792,
+      "step": 231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9964904613785966e-05,
+      "loss": 1.7076,
+      "step": 232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9963999507400014e-05,
+      "loss": 1.7722,
+      "step": 233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996308289887366e-05,
+      "loss": 1.6297,
+      "step": 234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9962154789265013e-05,
+      "loss": 1.5994,
+      "step": 235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9961215179645458e-05,
+      "loss": 2.1654,
+      "step": 236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996026407109964e-05,
+      "loss": 1.7753,
+      "step": 237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9959301464725507e-05,
+      "loss": 1.6402,
+      "step": 238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9958327361634248e-05,
+      "loss": 1.7012,
+      "step": 239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9957341762950346e-05,
+      "loss": 1.6705,
+      "step": 240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9956344669811548e-05,
+      "loss": 1.8943,
+      "step": 241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995533608336886e-05,
+      "loss": 1.6194,
+      "step": 242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9954316004786566e-05,
+      "loss": 1.6886,
+      "step": 243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9953284435242222e-05,
+      "loss": 2.0129,
+      "step": 244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9952241375926636e-05,
+      "loss": 1.9811,
+      "step": 245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995118682804388e-05,
+      "loss": 1.3275,
+      "step": 246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9950120792811294e-05,
+      "loss": 1.8802,
+      "step": 247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9949043271459483e-05,
+      "loss": 1.6943,
+      "step": 248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9947954265232298e-05,
+      "loss": 1.9114,
+      "step": 249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9946853775386857e-05,
+      "loss": 1.9585,
+      "step": 250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9945741803193534e-05,
+      "loss": 1.5431,
+      "step": 251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9944618349935956e-05,
+      "loss": 1.922,
+      "step": 252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9943483416911e-05,
+      "loss": 1.6552,
+      "step": 253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9942337005428805e-05,
+      "loss": 1.9619,
+      "step": 254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.994117911681275e-05,
+      "loss": 1.7433,
+      "step": 255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9940009752399462e-05,
+      "loss": 1.9205,
+      "step": 256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9938828913538826e-05,
+      "loss": 1.731,
+      "step": 257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9937636601593965e-05,
+      "loss": 1.4673,
+      "step": 258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.993643281794125e-05,
+      "loss": 1.8056,
+      "step": 259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.993521756397029e-05,
+      "loss": 1.8455,
+      "step": 260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.993399084108394e-05,
+      "loss": 1.8556,
+      "step": 261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9932752650698285e-05,
+      "loss": 1.8286,
+      "step": 262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9931502994242665e-05,
+      "loss": 1.7359,
+      "step": 263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9930241873159638e-05,
+      "loss": 1.7251,
+      "step": 264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9928969288905005e-05,
+      "loss": 1.8468,
+      "step": 265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9927685242947804e-05,
+      "loss": 1.5952,
+      "step": 266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9926389736770294e-05,
+      "loss": 1.759,
+      "step": 267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.992508277186797e-05,
+      "loss": 1.8307,
+      "step": 268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.992376434974955e-05,
+      "loss": 1.9021,
+      "step": 269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9922434471936987e-05,
+      "loss": 1.9823,
+      "step": 270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9921093139965448e-05,
+      "loss": 1.793,
+      "step": 271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9919740355383323e-05,
+      "loss": 1.8502,
+      "step": 272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991837611975223e-05,
+      "loss": 1.7116,
+      "step": 273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9917000434647e-05,
+      "loss": 1.7944,
+      "step": 274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991561330165568e-05,
+      "loss": 1.6988,
+      "step": 275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9914214722379545e-05,
+      "loss": 1.6164,
+      "step": 276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991280469843306e-05,
+      "loss": 1.8396,
+      "step": 277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991138323144392e-05,
+      "loss": 1.6016,
+      "step": 278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990995032305302e-05,
+      "loss": 1.6787,
+      "step": 279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9908505974914473e-05,
+      "loss": 1.9456,
+      "step": 280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9907050188695584e-05,
+      "loss": 1.7995,
+      "step": 281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990558296607687e-05,
+      "loss": 1.8331,
+      "step": 282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9904104308752053e-05,
+      "loss": 1.7004,
+      "step": 283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990261421842805e-05,
+      "loss": 1.6715,
+      "step": 284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9901112696824967e-05,
+      "loss": 1.6701,
+      "step": 285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9899599745676123e-05,
+      "loss": 1.5317,
+      "step": 286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9898075366728024e-05,
+      "loss": 1.8454,
+      "step": 287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9896539561740367e-05,
+      "loss": 1.8331,
+      "step": 288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9894992332486032e-05,
+      "loss": 1.9444,
+      "step": 289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9893433680751105e-05,
+      "loss": 1.6893,
+      "step": 290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9891863608334835e-05,
+      "loss": 1.7873,
+      "step": 291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9890282117049673e-05,
+      "loss": 1.8925,
+      "step": 292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9888689208721246e-05,
+      "loss": 1.8094,
+      "step": 293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9887084885188354e-05,
+      "loss": 1.7978,
+      "step": 294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9885469148302984e-05,
+      "loss": 1.782,
+      "step": 295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9883841999930292e-05,
+      "loss": 1.6562,
+      "step": 296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9882203441948608e-05,
+      "loss": 1.7716,
+      "step": 297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9880553476249437e-05,
+      "loss": 1.5763,
+      "step": 298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9878892104737445e-05,
+      "loss": 1.7692,
+      "step": 299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9877219329330474e-05,
+      "loss": 1.7107,
+      "step": 300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9875535151959526e-05,
+      "loss": 1.5842,
+      "step": 301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9873839574568756e-05,
+      "loss": 1.8934,
+      "step": 302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9872132599115497e-05,
+      "loss": 1.6879,
+      "step": 303
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9870414227570225e-05,
+      "loss": 1.7817,
+      "step": 304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9868684461916578e-05,
+      "loss": 1.9728,
+      "step": 305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9866943304151346e-05,
+      "loss": 1.722,
+      "step": 306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9865190756284467e-05,
+      "loss": 1.7428,
+      "step": 307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.986342682033903e-05,
+      "loss": 1.8,
+      "step": 308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9861651498351268e-05,
+      "loss": 1.7522,
+      "step": 309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9859864792370565e-05,
+      "loss": 1.7214,
+      "step": 310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9858066704459432e-05,
+      "loss": 1.7796,
+      "step": 311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9856257236693536e-05,
+      "loss": 1.6853,
+      "step": 312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9854436391161664e-05,
+      "loss": 1.9685,
+      "step": 313
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.985260416996575e-05,
+      "loss": 1.6717,
+      "step": 314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9850760575220854e-05,
+      "loss": 1.75,
+      "step": 315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9848905609055164e-05,
+      "loss": 1.4076,
+      "step": 316
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9847039273610004e-05,
+      "loss": 1.8508,
+      "step": 317
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9845161571039805e-05,
+      "loss": 1.8255,
+      "step": 318
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9843272503512136e-05,
+      "loss": 1.4479,
+      "step": 319
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.984137207320768e-05,
+      "loss": 1.8653,
+      "step": 320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9839460282320234e-05,
+      "loss": 1.7947,
+      "step": 321
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.983753713305672e-05,
+      "loss": 1.5169,
+      "step": 322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9835602627637146e-05,
+      "loss": 1.8413,
+      "step": 323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.983365676829466e-05,
+      "loss": 1.6661,
+      "step": 324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9831699557275504e-05,
+      "loss": 1.7901,
+      "step": 325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.982973099683902e-05,
+      "loss": 1.797,
+      "step": 326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.982775108925765e-05,
+      "loss": 1.742,
+      "step": 327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9825759836816948e-05,
+      "loss": 1.7826,
+      "step": 328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.982375724181555e-05,
+      "loss": 1.7252,
+      "step": 329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.98217433065652e-05,
+      "loss": 1.7326,
+      "step": 330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9819718033390708e-05,
+      "loss": 1.9651,
+      "step": 331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.981768142463e-05,
+      "loss": 1.5211,
+      "step": 332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9815633482634075e-05,
+      "loss": 1.5237,
+      "step": 333
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9813574209767013e-05,
+      "loss": 1.8086,
+      "step": 334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9811503608405978e-05,
+      "loss": 1.7453,
+      "step": 335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.980942168094121e-05,
+      "loss": 1.872,
+      "step": 336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.980732842977602e-05,
+      "loss": 1.5766,
+      "step": 337
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9805223857326794e-05,
+      "loss": 1.7682,
+      "step": 338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9803107966022993e-05,
+      "loss": 1.7214,
+      "step": 339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9800980758307128e-05,
+      "loss": 1.8943,
+      "step": 340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9798842236634797e-05,
+      "loss": 1.8683,
+      "step": 341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9796692403474632e-05,
+      "loss": 1.6865,
+      "step": 342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9794531261308343e-05,
+      "loss": 1.7859,
+      "step": 343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9792358812630686e-05,
+      "loss": 1.6847,
+      "step": 344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.979017505994947e-05,
+      "loss": 1.7756,
+      "step": 345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9787980005785553e-05,
+      "loss": 1.8554,
+      "step": 346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9785773652672837e-05,
+      "loss": 1.9206,
+      "step": 347
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9783556003158272e-05,
+      "loss": 1.7285,
+      "step": 348
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9781327059801844e-05,
+      "loss": 1.5388,
+      "step": 349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.977908682517658e-05,
+      "loss": 1.6945,
+      "step": 350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.977683530186854e-05,
+      "loss": 1.5777,
+      "step": 351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9774572492476813e-05,
+      "loss": 1.8297,
+      "step": 352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9772298399613517e-05,
+      "loss": 1.7059,
+      "step": 353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9770013025903797e-05,
+      "loss": 1.7508,
+      "step": 354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.976771637398582e-05,
+      "loss": 1.6365,
+      "step": 355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9765408446510766e-05,
+      "loss": 1.9101,
+      "step": 356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.976308924614285e-05,
+      "loss": 1.7317,
+      "step": 357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9760758775559275e-05,
+      "loss": 1.7233,
+      "step": 358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9758417037450268e-05,
+      "loss": 1.7028,
+      "step": 359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9756064034519066e-05,
+      "loss": 1.7722,
+      "step": 360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9753699769481898e-05,
+      "loss": 1.8861,
+      "step": 361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9751324245068008e-05,
+      "loss": 1.9386,
+      "step": 362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.974893746401962e-05,
+      "loss": 1.8221,
+      "step": 363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.974653942909197e-05,
+      "loss": 1.6888,
+      "step": 364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9744130143053275e-05,
+      "loss": 1.8236,
+      "step": 365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.974170960868474e-05,
+      "loss": 1.645,
+      "step": 366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9739277828780563e-05,
+      "loss": 1.6805,
+      "step": 367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9736834806147907e-05,
+      "loss": 1.7565,
+      "step": 368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9734380543606932e-05,
+      "loss": 1.845,
+      "step": 369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.973191504399076e-05,
+      "loss": 1.9799,
+      "step": 370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9729438310145493e-05,
+      "loss": 2.0065,
+      "step": 371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9726950344930193e-05,
+      "loss": 1.8359,
+      "step": 372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9724451151216896e-05,
+      "loss": 1.9645,
+      "step": 373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.97219407318906e-05,
+      "loss": 1.7875,
+      "step": 374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.971941908984925e-05,
+      "loss": 1.6119,
+      "step": 375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9716886228003752e-05,
+      "loss": 1.617,
+      "step": 376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9714342149277973e-05,
+      "loss": 1.768,
+      "step": 377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9711786856608714e-05,
+      "loss": 1.6316,
+      "step": 378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.970922035294573e-05,
+      "loss": 1.5247,
+      "step": 379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9706642641251717e-05,
+      "loss": 1.4706,
+      "step": 380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.970405372450231e-05,
+      "loss": 1.5362,
+      "step": 381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.970145360568607e-05,
+      "loss": 1.75,
+      "step": 382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.96988422878045e-05,
+      "loss": 1.7899,
+      "step": 383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9696219773872023e-05,
+      "loss": 1.6294,
+      "step": 384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.969358606691599e-05,
+      "loss": 1.8169,
+      "step": 385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.969094116997668e-05,
+      "loss": 1.8124,
+      "step": 386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9688285086107277e-05,
+      "loss": 1.8343,
+      "step": 387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9685617818373882e-05,
+      "loss": 1.7585,
+      "step": 388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.968293936985551e-05,
+      "loss": 1.6932,
+      "step": 389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.968024974364408e-05,
+      "loss": 2.2422,
+      "step": 390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9677548942844426e-05,
+      "loss": 1.6425,
+      "step": 391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9674836970574253e-05,
+      "loss": 1.8344,
+      "step": 392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9672113829964193e-05,
+      "loss": 1.5743,
+      "step": 393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9669379524157755e-05,
+      "loss": 1.6486,
+      "step": 394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9666634056311332e-05,
+      "loss": 1.9086,
+      "step": 395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.966387742959422e-05,
+      "loss": 1.5008,
+      "step": 396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.966110964718858e-05,
+      "loss": 1.6269,
+      "step": 397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9658330712289456e-05,
+      "loss": 1.5499,
+      "step": 398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9655540628104764e-05,
+      "loss": 1.8059,
+      "step": 399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.96527393978553e-05,
+      "loss": 1.5598,
+      "step": 400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.964992702477472e-05,
+      "loss": 1.7569,
+      "step": 401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9647103512109535e-05,
+      "loss": 1.6824,
+      "step": 402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.964426886311913e-05,
+      "loss": 1.8274,
+      "step": 403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.964142308107573e-05,
+      "loss": 1.6802,
+      "step": 404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.963856616926443e-05,
+      "loss": 1.9647,
+      "step": 405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9635698130983153e-05,
+      "loss": 1.8397,
+      "step": 406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9632818969542682e-05,
+      "loss": 2.0337,
+      "step": 407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9629928688266636e-05,
+      "loss": 1.6994,
+      "step": 408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.962702729049146e-05,
+      "loss": 1.894,
+      "step": 409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.962411477956645e-05,
+      "loss": 1.7808,
+      "step": 410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.962119115885371e-05,
+      "loss": 1.8946,
+      "step": 411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.961825643172819e-05,
+      "loss": 1.5389,
+      "step": 412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.961531060157765e-05,
+      "loss": 1.7442,
+      "step": 413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9612353671802658e-05,
+      "loss": 1.6907,
+      "step": 414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.960938564581661e-05,
+      "loss": 1.8326,
+      "step": 415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9606406527045708e-05,
+      "loss": 1.8827,
+      "step": 416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.960341631892895e-05,
+      "loss": 1.6993,
+      "step": 417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.960041502491815e-05,
+      "loss": 1.8629,
+      "step": 418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.959740264847791e-05,
+      "loss": 1.733,
+      "step": 419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9594379193085618e-05,
+      "loss": 1.7979,
+      "step": 420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.959134466223147e-05,
+      "loss": 1.6863,
+      "step": 421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9588299059418434e-05,
+      "loss": 1.536,
+      "step": 422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.958524238816226e-05,
+      "loss": 1.842,
+      "step": 423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9582174651991474e-05,
+      "loss": 1.7918,
+      "step": 424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9579095854447386e-05,
+      "loss": 1.7509,
+      "step": 425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.957600599908406e-05,
+      "loss": 1.7738,
+      "step": 426
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9572905089468337e-05,
+      "loss": 1.7023,
+      "step": 427
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.956979312917981e-05,
+      "loss": 1.7062,
+      "step": 428
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9566670121810834e-05,
+      "loss": 1.8695,
+      "step": 429
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9563536070966513e-05,
+      "loss": 1.6029,
+      "step": 430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9560390980264704e-05,
+      "loss": 1.7784,
+      "step": 431
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9557234853335997e-05,
+      "loss": 1.8444,
+      "step": 432
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9554067693823738e-05,
+      "loss": 1.8925,
+      "step": 433
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9550889505383996e-05,
+      "loss": 1.6609,
+      "step": 434
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.954770029168558e-05,
+      "loss": 1.8348,
+      "step": 435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9544500056410018e-05,
+      "loss": 1.9367,
+      "step": 436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9541288803251565e-05,
+      "loss": 1.8231,
+      "step": 437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9538066535917196e-05,
+      "loss": 1.8258,
+      "step": 438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9534833258126604e-05,
+      "loss": 1.7927,
+      "step": 439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9531588973612177e-05,
+      "loss": 1.9152,
+      "step": 440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9528333686119026e-05,
+      "loss": 1.6697,
+      "step": 441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.952506739940496e-05,
+      "loss": 1.7706,
+      "step": 442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9521790117240472e-05,
+      "loss": 1.6133,
+      "step": 443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9518501843408763e-05,
+      "loss": 1.7327,
+      "step": 444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9515202581705716e-05,
+      "loss": 1.8659,
+      "step": 445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9511892335939904e-05,
+      "loss": 1.6167,
+      "step": 446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.950857110993257e-05,
+      "loss": 1.7971,
+      "step": 447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9505238907517637e-05,
+      "loss": 1.7787,
+      "step": 448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.95018957325417e-05,
+      "loss": 1.9342,
+      "step": 449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9498541588864022e-05,
+      "loss": 1.9511,
+      "step": 450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9495176480356518e-05,
+      "loss": 1.8879,
+      "step": 451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9491800410903776e-05,
+      "loss": 1.7453,
+      "step": 452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9488413384403023e-05,
+      "loss": 1.719,
+      "step": 453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.948501540476414e-05,
+      "loss": 1.7164,
+      "step": 454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.948160647590966e-05,
+      "loss": 1.5412,
+      "step": 455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.947818660177474e-05,
+      "loss": 1.7268,
+      "step": 456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.947475578630718e-05,
+      "loss": 1.9529,
+      "step": 457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9471314033467413e-05,
+      "loss": 1.8673,
+      "step": 458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9467861347228492e-05,
+      "loss": 1.6668,
+      "step": 459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9464397731576093e-05,
+      "loss": 1.869,
+      "step": 460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9460923190508516e-05,
+      "loss": 1.7929,
+      "step": 461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.945743772803666e-05,
+      "loss": 1.6692,
+      "step": 462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9453941348184036e-05,
+      "loss": 1.8869,
+      "step": 463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9450434054986766e-05,
+      "loss": 1.5765,
+      "step": 464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9446915852493558e-05,
+      "loss": 1.7702,
+      "step": 465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9443386744765726e-05,
+      "loss": 1.8814,
+      "step": 466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9439846735877155e-05,
+      "loss": 1.7241,
+      "step": 467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9436295829914332e-05,
+      "loss": 1.6648,
+      "step": 468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9432734030976314e-05,
+      "loss": 1.7014,
+      "step": 469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.942916134317473e-05,
+      "loss": 1.8706,
+      "step": 470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9425577770633794e-05,
+      "loss": 1.7884,
+      "step": 471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.942198331749026e-05,
+      "loss": 1.4971,
+      "step": 472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.941837798789347e-05,
+      "loss": 1.6726,
+      "step": 473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9414761786005293e-05,
+      "loss": 1.7007,
+      "step": 474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9411134716000177e-05,
+      "loss": 1.7307,
+      "step": 475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9407496782065096e-05,
+      "loss": 1.525,
+      "step": 476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.940384798839957e-05,
+      "loss": 1.747,
+      "step": 477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9400188339215657e-05,
+      "loss": 1.7561,
+      "step": 478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.939651783873795e-05,
+      "loss": 1.7766,
+      "step": 479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9392836491203558e-05,
+      "loss": 1.8235,
+      "step": 480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.938914430086212e-05,
+      "loss": 1.8071,
+      "step": 481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9385441271975786e-05,
+      "loss": 1.6837,
+      "step": 482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9381727408819227e-05,
+      "loss": 1.8678,
+      "step": 483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9378002715679613e-05,
+      "loss": 2.0677,
+      "step": 484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9374267196856618e-05,
+      "loss": 1.8712,
+      "step": 485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9370520856662406e-05,
+      "loss": 1.7708,
+      "step": 486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9366763699421645e-05,
+      "loss": 1.7327,
+      "step": 487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.936299572947148e-05,
+      "loss": 1.646,
+      "step": 488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9359216951161546e-05,
+      "loss": 1.5801,
+      "step": 489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9355427368853946e-05,
+      "loss": 1.7493,
+      "step": 490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9351626986923265e-05,
+      "loss": 1.8237,
+      "step": 491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9347815809756544e-05,
+      "loss": 1.8113,
+      "step": 492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9343993841753293e-05,
+      "loss": 1.7218,
+      "step": 493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9340161087325483e-05,
+      "loss": 1.8655,
+      "step": 494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.933631755089752e-05,
+      "loss": 1.856,
+      "step": 495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.933246323690627e-05,
+      "loss": 1.9378,
+      "step": 496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9328598149801037e-05,
+      "loss": 1.6061,
+      "step": 497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.932472229404356e-05,
+      "loss": 1.8996,
+      "step": 498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9320835674108012e-05,
+      "loss": 1.7959,
+      "step": 499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9316938294480988e-05,
+      "loss": 1.3371,
+      "step": 500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.931303015966151e-05,
+      "loss": 1.7987,
+      "step": 501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9309111274161005e-05,
+      "loss": 1.8354,
+      "step": 502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9305181642503318e-05,
+      "loss": 1.875,
+      "step": 503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.93012412692247e-05,
+      "loss": 1.6204,
+      "step": 504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.929729015887379e-05,
+      "loss": 1.4167,
+      "step": 505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9293328316011645e-05,
+      "loss": 1.8976,
+      "step": 506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9289355745211685e-05,
+      "loss": 1.8918,
+      "step": 507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.928537245105973e-05,
+      "loss": 1.9931,
+      "step": 508
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9281378438153973e-05,
+      "loss": 1.482,
+      "step": 509
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.927737371110499e-05,
+      "loss": 1.722,
+      "step": 510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9273358274535703e-05,
+      "loss": 1.6606,
+      "step": 511
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9269332133081425e-05,
+      "loss": 1.9861,
+      "step": 512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9265295291389803e-05,
+      "loss": 1.8291,
+      "step": 513
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9261247754120846e-05,
+      "loss": 1.7223,
+      "step": 514
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.925718952594691e-05,
+      "loss": 1.6795,
+      "step": 515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9253120611552697e-05,
+      "loss": 2.0152,
+      "step": 516
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9249041015635232e-05,
+      "loss": 1.7562,
+      "step": 517
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.924495074290388e-05,
+      "loss": 1.7421,
+      "step": 518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9240849798080326e-05,
+      "loss": 1.7958,
+      "step": 519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9236738185898577e-05,
+      "loss": 1.8144,
+      "step": 520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.923261591110496e-05,
+      "loss": 1.6845,
+      "step": 521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.92284829784581e-05,
+      "loss": 1.4245,
+      "step": 522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9224339392728928e-05,
+      "loss": 1.953,
+      "step": 523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9220185158700677e-05,
+      "loss": 1.8366,
+      "step": 524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.921602028116887e-05,
+      "loss": 1.5909,
+      "step": 525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9211844764941318e-05,
+      "loss": 1.8508,
+      "step": 526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.92076586148381e-05,
+      "loss": 1.5519,
+      "step": 527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9203461835691596e-05,
+      "loss": 1.8339,
+      "step": 528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.919925443234643e-05,
+      "loss": 1.5713,
+      "step": 529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.919503640965951e-05,
+      "loss": 1.805,
+      "step": 530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.919080777249999e-05,
+      "loss": 1.6221,
+      "step": 531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.918656852574928e-05,
+      "loss": 1.8889,
+      "step": 532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.918231867430104e-05,
+      "loss": 1.8096,
+      "step": 533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.917805822306117e-05,
+      "loss": 1.5302,
+      "step": 534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9173787176947805e-05,
+      "loss": 1.6731,
+      "step": 535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9169505540891317e-05,
+      "loss": 1.5763,
+      "step": 536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9165213319834294e-05,
+      "loss": 1.6019,
+      "step": 537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.916091051873154e-05,
+      "loss": 1.9008,
+      "step": 538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.915659714255009e-05,
+      "loss": 1.6872,
+      "step": 539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9152273196269168e-05,
+      "loss": 1.6404,
+      "step": 540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9147938684880213e-05,
+      "loss": 1.9071,
+      "step": 541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9143593613386845e-05,
+      "loss": 1.7477,
+      "step": 542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9139237986804896e-05,
+      "loss": 1.7335,
+      "step": 543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9134871810162358e-05,
+      "loss": 1.7341,
+      "step": 544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9130495088499417e-05,
+      "loss": 1.6991,
+      "step": 545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9126107826868436e-05,
+      "loss": 1.8856,
+      "step": 546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9121710030333928e-05,
+      "loss": 1.7529,
+      "step": 547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9117301703972576e-05,
+      "loss": 1.8376,
+      "step": 548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9112882852873228e-05,
+      "loss": 1.6009,
+      "step": 549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9108453482136866e-05,
+      "loss": 1.4621,
+      "step": 550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9104013596876624e-05,
+      "loss": 1.7644,
+      "step": 551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.909956320221777e-05,
+      "loss": 1.5465,
+      "step": 552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.909510230329771e-05,
+      "loss": 1.3783,
+      "step": 553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9090630905265963e-05,
+      "loss": 1.7462,
+      "step": 554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.908614901328418e-05,
+      "loss": 1.7621,
+      "step": 555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9081656632526123e-05,
+      "loss": 1.7911,
+      "step": 556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9077153768177662e-05,
+      "loss": 1.9265,
+      "step": 557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9072640425436762e-05,
+      "loss": 1.7597,
+      "step": 558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.90681166095135e-05,
+      "loss": 1.8347,
+      "step": 559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9063582325630018e-05,
+      "loss": 1.6422,
+      "step": 560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9059037579020572e-05,
+      "loss": 1.7309,
+      "step": 561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.905448237493147e-05,
+      "loss": 1.6012,
+      "step": 562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.90499167186211e-05,
+      "loss": 1.4604,
+      "step": 563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9045340615359927e-05,
+      "loss": 1.7878,
+      "step": 564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9040754070430464e-05,
+      "loss": 1.6162,
+      "step": 565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9036157089127278e-05,
+      "loss": 1.6079,
+      "step": 566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9031549676756987e-05,
+      "loss": 1.889,
+      "step": 567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9026931838638245e-05,
+      "loss": 1.6915,
+      "step": 568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.902230358010175e-05,
+      "loss": 1.917,
+      "step": 569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.901766490649022e-05,
+      "loss": 1.7692,
+      "step": 570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9013015823158402e-05,
+      "loss": 1.6312,
+      "step": 571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.900835633547306e-05,
+      "loss": 1.6896,
+      "step": 572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9003686448812957e-05,
+      "loss": 1.8938,
+      "step": 573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8999006168568883e-05,
+      "loss": 1.7499,
+      "step": 574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8994315500143604e-05,
+      "loss": 1.6409,
+      "step": 575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8989614448951884e-05,
+      "loss": 1.7132,
+      "step": 576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8984903020420477e-05,
+      "loss": 1.3914,
+      "step": 577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8980181219988117e-05,
+      "loss": 1.7553,
+      "step": 578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8975449053105505e-05,
+      "loss": 1.7272,
+      "step": 579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8970706525235313e-05,
+      "loss": 1.7466,
+      "step": 580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8965953641852168e-05,
+      "loss": 1.7639,
+      "step": 581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8961190408442662e-05,
+      "loss": 1.5734,
+      "step": 582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8956416830505323e-05,
+      "loss": 1.6802,
+      "step": 583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8951632913550625e-05,
+      "loss": 1.9558,
+      "step": 584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8946838663100983e-05,
+      "loss": 1.7251,
+      "step": 585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8942034084690727e-05,
+      "loss": 1.897,
+      "step": 586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8937219183866124e-05,
+      "loss": 1.6324,
+      "step": 587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8932393966185346e-05,
+      "loss": 1.9967,
+      "step": 588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8927558437218482e-05,
+      "loss": 1.7535,
+      "step": 589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8922712602547516e-05,
+      "loss": 1.9121,
+      "step": 590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8917856467766342e-05,
+      "loss": 1.9566,
+      "step": 591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8912990038480726e-05,
+      "loss": 1.6893,
+      "step": 592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8908113320308333e-05,
+      "loss": 1.4574,
+      "step": 593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.89032263188787e-05,
+      "loss": 1.8363,
+      "step": 594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8898329039833226e-05,
+      "loss": 1.5241,
+      "step": 595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.889342148882519e-05,
+      "loss": 1.9233,
+      "step": 596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8888503671519716e-05,
+      "loss": 1.6432,
+      "step": 597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8883575593593793e-05,
+      "loss": 1.5075,
+      "step": 598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8878637260736233e-05,
+      "loss": 1.6269,
+      "step": 599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.887368867864771e-05,
+      "loss": 1.4122,
+      "step": 600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8868729853040705e-05,
+      "loss": 1.6215,
+      "step": 601
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8863760789639548e-05,
+      "loss": 1.6862,
+      "step": 602
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.885878149418037e-05,
+      "loss": 1.5872,
+      "step": 603
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8853791972411124e-05,
+      "loss": 1.895,
+      "step": 604
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.884879223009156e-05,
+      "loss": 1.8277,
+      "step": 605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8843782272993225e-05,
+      "loss": 1.5571,
+      "step": 606
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8838762106899474e-05,
+      "loss": 1.5747,
+      "step": 607
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8833731737605425e-05,
+      "loss": 1.5048,
+      "step": 608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.882869117091799e-05,
+      "loss": 1.8415,
+      "step": 609
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8823640412655844e-05,
+      "loss": 1.8252,
+      "step": 610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.881857946864943e-05,
+      "loss": 1.7839,
+      "step": 611
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8813508344740957e-05,
+      "loss": 1.7213,
+      "step": 612
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8808427046784365e-05,
+      "loss": 1.8811,
+      "step": 613
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.880333558064536e-05,
+      "loss": 1.7629,
+      "step": 614
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.879823395220137e-05,
+      "loss": 2.0449,
+      "step": 615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8793122167341565e-05,
+      "loss": 1.6233,
+      "step": 616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8788000231966837e-05,
+      "loss": 1.7363,
+      "step": 617
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.878286815198979e-05,
+      "loss": 1.6735,
+      "step": 618
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.877772593333475e-05,
+      "loss": 1.7614,
+      "step": 619
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8772573581937727e-05,
+      "loss": 1.7222,
+      "step": 620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8767411103746448e-05,
+      "loss": 1.8761,
+      "step": 621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.876223850472032e-05,
+      "loss": 1.7694,
+      "step": 622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8757055790830435e-05,
+      "loss": 1.9006,
+      "step": 623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.875186296805956e-05,
+      "loss": 1.7671,
+      "step": 624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.874666004240214e-05,
+      "loss": 1.7244,
+      "step": 625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8741447019864263e-05,
+      "loss": 1.6109,
+      "step": 626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8736223906463698e-05,
+      "loss": 1.6257,
+      "step": 627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8730990708229838e-05,
+      "loss": 1.6745,
+      "step": 628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.872574743120374e-05,
+      "loss": 1.4838,
+      "step": 629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.872049408143808e-05,
+      "loss": 1.8254,
+      "step": 630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8715230664997163e-05,
+      "loss": 1.9155,
+      "step": 631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.870995718795692e-05,
+      "loss": 1.642,
+      "step": 632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8704673656404896e-05,
+      "loss": 1.7147,
+      "step": 633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8699380076440242e-05,
+      "loss": 1.7104,
+      "step": 634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8694076454173702e-05,
+      "loss": 1.4929,
+      "step": 635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.868876279572762e-05,
+      "loss": 1.7098,
+      "step": 636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.868343910723592e-05,
+      "loss": 1.6934,
+      "step": 637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8678105394844114e-05,
+      "loss": 1.7491,
+      "step": 638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8672761664709277e-05,
+      "loss": 1.9946,
+      "step": 639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8667407923000042e-05,
+      "loss": 1.7294,
+      "step": 640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8662044175896616e-05,
+      "loss": 1.6223,
+      "step": 641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8656670429590745e-05,
+      "loss": 1.8601,
+      "step": 642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8651286690285713e-05,
+      "loss": 1.7429,
+      "step": 643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8645892964196356e-05,
+      "loss": 1.7666,
+      "step": 644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8640489257549023e-05,
+      "loss": 1.7421,
+      "step": 645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8635075576581587e-05,
+      "loss": 1.6193,
+      "step": 646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8629651927543447e-05,
+      "loss": 1.8405,
+      "step": 647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.862421831669549e-05,
+      "loss": 1.509,
+      "step": 648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.861877475031012e-05,
+      "loss": 1.5281,
+      "step": 649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.861332123467122e-05,
+      "loss": 1.8329,
+      "step": 650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.860785777607417e-05,
+      "loss": 1.5501,
+      "step": 651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8602384380825816e-05,
+      "loss": 1.7975,
+      "step": 652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8596901055244486e-05,
+      "loss": 1.7015,
+      "step": 653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.859140780565996e-05,
+      "loss": 1.7431,
+      "step": 654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8585904638413485e-05,
+      "loss": 1.9761,
+      "step": 655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8580391559857744e-05,
+      "loss": 1.5233,
+      "step": 656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8574868576356874e-05,
+      "loss": 1.643,
+      "step": 657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.856933569428644e-05,
+      "loss": 1.8038,
+      "step": 658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8563792920033433e-05,
+      "loss": 1.7902,
+      "step": 659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8558240259996262e-05,
+      "loss": 1.7472,
+      "step": 660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.855267772058476e-05,
+      "loss": 1.6681,
+      "step": 661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8547105308220142e-05,
+      "loss": 1.6426,
+      "step": 662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.854152302933504e-05,
+      "loss": 1.7434,
+      "step": 663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8535930890373467e-05,
+      "loss": 1.7315,
+      "step": 664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8530328897790818e-05,
+      "loss": 1.4859,
+      "step": 665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.852471705805387e-05,
+      "loss": 1.7589,
+      "step": 666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.851909537764075e-05,
+      "loss": 2.0904,
+      "step": 667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8513463863040962e-05,
+      "loss": 1.6423,
+      "step": 668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8507822520755364e-05,
+      "loss": 1.7641,
+      "step": 669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8502171357296144e-05,
+      "loss": 1.8769,
+      "step": 670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8496510379186834e-05,
+      "loss": 1.6561,
+      "step": 671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8490839592962297e-05,
+      "loss": 1.6555,
+      "step": 672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.848515900516872e-05,
+      "loss": 1.6858,
+      "step": 673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.84794686223636e-05,
+      "loss": 1.7091,
+      "step": 674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8473768451115743e-05,
+      "loss": 1.6634,
+      "step": 675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8468058498005258e-05,
+      "loss": 1.9912,
+      "step": 676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8462338769623536e-05,
+      "loss": 1.4505,
+      "step": 677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8456609272573268e-05,
+      "loss": 1.6924,
+      "step": 678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8450870013468404e-05,
+      "loss": 1.6259,
+      "step": 679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8445120998934176e-05,
+      "loss": 1.6977,
+      "step": 680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8439362235607074e-05,
+      "loss": 1.863,
+      "step": 681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8433593730134835e-05,
+      "loss": 1.6938,
+      "step": 682
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8427815489176453e-05,
+      "loss": 2.0575,
+      "step": 683
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.842202751940215e-05,
+      "loss": 1.7801,
+      "step": 684
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.841622982749339e-05,
+      "loss": 1.8823,
+      "step": 685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.841042242014285e-05,
+      "loss": 1.9318,
+      "step": 686
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8404605304054427e-05,
+      "loss": 1.7233,
+      "step": 687
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8398778485943226e-05,
+      "loss": 1.6986,
+      "step": 688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8392941972535542e-05,
+      "loss": 1.9375,
+      "step": 689
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.838709577056888e-05,
+      "loss": 1.7496,
+      "step": 690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8381239886791914e-05,
+      "loss": 1.6486,
+      "step": 691
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.83753743279645e-05,
+      "loss": 1.616,
+      "step": 692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.836949910085766e-05,
+      "loss": 1.5628,
+      "step": 693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8363614212253585e-05,
+      "loss": 1.8322,
+      "step": 694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.835771966894561e-05,
+      "loss": 1.7217,
+      "step": 695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8351815477738214e-05,
+      "loss": 1.6875,
+      "step": 696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8345901645447023e-05,
+      "loss": 1.6585,
+      "step": 697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.833997817889878e-05,
+      "loss": 1.7619,
+      "step": 698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8334045084931364e-05,
+      "loss": 1.754,
+      "step": 699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8328102370393756e-05,
+      "loss": 1.882,
+      "step": 700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.832215004214604e-05,
+      "loss": 1.685,
+      "step": 701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8316188107059418e-05,
+      "loss": 1.6703,
+      "step": 702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.831021657201616e-05,
+      "loss": 1.5117,
+      "step": 703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8304235443909618e-05,
+      "loss": 1.7566,
+      "step": 704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.829824472964424e-05,
+      "loss": 1.6342,
+      "step": 705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8292244436135517e-05,
+      "loss": 1.7662,
+      "step": 706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8286234570310012e-05,
+      "loss": 1.6164,
+      "step": 707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.828021513910533e-05,
+      "loss": 1.732,
+      "step": 708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8274186149470118e-05,
+      "loss": 1.7158,
+      "step": 709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8268147608364068e-05,
+      "loss": 1.8344,
+      "step": 710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.826209952275788e-05,
+      "loss": 1.9967,
+      "step": 711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8256041899633285e-05,
+      "loss": 1.6985,
+      "step": 712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8249974745983023e-05,
+      "loss": 1.7677,
+      "step": 713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8243898068810833e-05,
+      "loss": 1.4142,
+      "step": 714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8237811875131446e-05,
+      "loss": 1.6172,
+      "step": 715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.823171617197058e-05,
+      "loss": 2.0012,
+      "step": 716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8225610966364932e-05,
+      "loss": 1.5762,
+      "step": 717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8219496265362164e-05,
+      "loss": 1.9279,
+      "step": 718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8213372076020906e-05,
+      "loss": 1.5075,
+      "step": 719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8207238405410736e-05,
+      "loss": 1.9357,
+      "step": 720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.820109526061218e-05,
+      "loss": 1.6867,
+      "step": 721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.81949426487167e-05,
+      "loss": 1.416,
+      "step": 722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8188780576826675e-05,
+      "loss": 1.7476,
+      "step": 723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.818260905205543e-05,
+      "loss": 1.4246,
+      "step": 724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8176428081527176e-05,
+      "loss": 1.5943,
+      "step": 725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8170237672377046e-05,
+      "loss": 1.707,
+      "step": 726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8164037831751055e-05,
+      "loss": 1.7107,
+      "step": 727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.815782856680612e-05,
+      "loss": 1.8623,
+      "step": 728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.815160988471002e-05,
+      "loss": 1.7462,
+      "step": 729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.814538179264142e-05,
+      "loss": 1.7079,
+      "step": 730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.813914429778984e-05,
+      "loss": 1.6657,
+      "step": 731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8132897407355657e-05,
+      "loss": 1.515,
+      "step": 732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.812664112855009e-05,
+      "loss": 1.9187,
+      "step": 733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.81203754685952e-05,
+      "loss": 1.8612,
+      "step": 734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8114100434723875e-05,
+      "loss": 1.6637,
+      "step": 735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.810781603417983e-05,
+      "loss": 1.7603,
+      "step": 736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.810152227421758e-05,
+      "loss": 1.8188,
+      "step": 737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8095219162102453e-05,
+      "loss": 1.5943,
+      "step": 738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8088906705110577e-05,
+      "loss": 1.6853,
+      "step": 739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8082584910528856e-05,
+      "loss": 1.7072,
+      "step": 740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.807625378565498e-05,
+      "loss": 1.6609,
+      "step": 741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8069913337797414e-05,
+      "loss": 1.5954,
+      "step": 742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.806356357427537e-05,
+      "loss": 1.7572,
+      "step": 743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.805720450241883e-05,
+      "loss": 1.5204,
+      "step": 744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8050836129568518e-05,
+      "loss": 1.5685,
+      "step": 745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.804445846307588e-05,
+      "loss": 1.6966,
+      "step": 746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8038071510303114e-05,
+      "loss": 1.5989,
+      "step": 747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8031675278623116e-05,
+      "loss": 1.5943,
+      "step": 748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.802526977541951e-05,
+      "loss": 1.8119,
+      "step": 749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.801885500808661e-05,
+      "loss": 1.785,
+      "step": 750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8012430984029436e-05,
+      "loss": 1.5177,
+      "step": 751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8005997710663684e-05,
+      "loss": 1.5845,
+      "step": 752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.799955519541573e-05,
+      "loss": 1.708,
+      "step": 753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7993103445722615e-05,
+      "loss": 1.7466,
+      "step": 754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7986642469032053e-05,
+      "loss": 1.6492,
+      "step": 755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7980172272802398e-05,
+      "loss": 1.8176,
+      "step": 756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7973692864502645e-05,
+      "loss": 1.5066,
+      "step": 757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7967204251612432e-05,
+      "loss": 1.5908,
+      "step": 758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.796070644162202e-05,
+      "loss": 1.6564,
+      "step": 759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7954199442032277e-05,
+      "loss": 1.7745,
+      "step": 760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.794768326035469e-05,
+      "loss": 1.6973,
+      "step": 761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7941157904111346e-05,
+      "loss": 1.7706,
+      "step": 762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7934623380834916e-05,
+      "loss": 1.7292,
+      "step": 763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.792807969806866e-05,
+      "loss": 1.5996,
+      "step": 764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7921526863366404e-05,
+      "loss": 1.9638,
+      "step": 765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7914964884292543e-05,
+      "loss": 1.6959,
+      "step": 766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.790839376842203e-05,
+      "loss": 1.7973,
+      "step": 767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.790181352334036e-05,
+      "loss": 1.3868,
+      "step": 768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7895224156643575e-05,
+      "loss": 1.9182,
+      "step": 769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7888625675938237e-05,
+      "loss": 1.5193,
+      "step": 770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.788201808884143e-05,
+      "loss": 1.7098,
+      "step": 771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7875401402980757e-05,
+      "loss": 1.9965,
+      "step": 772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7868775625994322e-05,
+      "loss": 1.6491,
+      "step": 773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7862140765530718e-05,
+      "loss": 2.0452,
+      "step": 774
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.785549682924903e-05,
+      "loss": 1.47,
+      "step": 775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.784884382481882e-05,
+      "loss": 1.7482,
+      "step": 776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.784218175992011e-05,
+      "loss": 1.6209,
+      "step": 777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.783551064224339e-05,
+      "loss": 1.8654,
+      "step": 778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7828830479489596e-05,
+      "loss": 1.5741,
+      "step": 779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.782214127937011e-05,
+      "loss": 1.6093,
+      "step": 780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.781544304960674e-05,
+      "loss": 2.1079,
+      "step": 781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7808735797931715e-05,
+      "loss": 1.6667,
+      "step": 782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7802019532087692e-05,
+      "loss": 1.7764,
+      "step": 783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7795294259827725e-05,
+      "loss": 1.5845,
+      "step": 784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.778855998891526e-05,
+      "loss": 1.6344,
+      "step": 785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7781816727124138e-05,
+      "loss": 1.639,
+      "step": 786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7775064482238585e-05,
+      "loss": 1.7192,
+      "step": 787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7768303262053175e-05,
+      "loss": 1.457,
+      "step": 788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7761533074372867e-05,
+      "loss": 1.7024,
+      "step": 789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7754753927012955e-05,
+      "loss": 1.4046,
+      "step": 790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7747965827799087e-05,
+      "loss": 1.8535,
+      "step": 791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7741168784567237e-05,
+      "loss": 1.8363,
+      "step": 792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.773436280516371e-05,
+      "loss": 1.7443,
+      "step": 793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7727547897445117e-05,
+      "loss": 1.7183,
+      "step": 794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.772072406927839e-05,
+      "loss": 1.9646,
+      "step": 795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7713891328540745e-05,
+      "loss": 1.8104,
+      "step": 796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7707049683119694e-05,
+      "loss": 1.9423,
+      "step": 797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.770019914091302e-05,
+      "loss": 1.6001,
+      "step": 798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.769333970982879e-05,
+      "loss": 1.541,
+      "step": 799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7686471397785322e-05,
+      "loss": 1.4299,
+      "step": 800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.767959421271119e-05,
+      "loss": 1.7618,
+      "step": 801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.76727081625452e-05,
+      "loss": 1.6685,
+      "step": 802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7665813255236413e-05,
+      "loss": 1.8127,
+      "step": 803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.765890949874409e-05,
+      "loss": 1.4162,
+      "step": 804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7651996901037734e-05,
+      "loss": 1.813,
+      "step": 805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7645075470097024e-05,
+      "loss": 1.73,
+      "step": 806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7638145213911858e-05,
+      "loss": 1.6279,
+      "step": 807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7631206140482316e-05,
+      "loss": 1.6889,
+      "step": 808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7624258257818657e-05,
+      "loss": 1.662,
+      "step": 809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7617301573941296e-05,
+      "loss": 1.741,
+      "step": 810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7610336096880828e-05,
+      "loss": 1.8032,
+      "step": 811
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.760336183467799e-05,
+      "loss": 1.7217,
+      "step": 812
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7596378795383656e-05,
+      "loss": 1.7925,
+      "step": 813
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.758938698705884e-05,
+      "loss": 1.4323,
+      "step": 814
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7582386417774672e-05,
+      "loss": 1.5944,
+      "step": 815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7575377095612396e-05,
+      "loss": 1.9034,
+      "step": 816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7568359028663365e-05,
+      "loss": 1.6656,
+      "step": 817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7561332225029022e-05,
+      "loss": 1.6387,
+      "step": 818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7554296692820897e-05,
+      "loss": 1.8055,
+      "step": 819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.75472524401606e-05,
+      "loss": 1.8159,
+      "step": 820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.75401994751798e-05,
+      "loss": 1.7307,
+      "step": 821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7533137806020226e-05,
+      "loss": 1.682,
+      "step": 822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.752606744083366e-05,
+      "loss": 1.7563,
+      "step": 823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7518988387781915e-05,
+      "loss": 1.8459,
+      "step": 824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7511900655036837e-05,
+      "loss": 1.6718,
+      "step": 825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7504804250780292e-05,
+      "loss": 1.8076,
+      "step": 826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.749769918320415e-05,
+      "loss": 1.7787,
+      "step": 827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7490585460510296e-05,
+      "loss": 1.5822,
+      "step": 828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7483463090910588e-05,
+      "loss": 1.6396,
+      "step": 829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.747633208262688e-05,
+      "loss": 1.6496,
+      "step": 830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7469192443890987e-05,
+      "loss": 1.799,
+      "step": 831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.74620441829447e-05,
+      "loss": 1.6521,
+      "step": 832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.745488730803975e-05,
+      "loss": 1.6198,
+      "step": 833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.744772182743782e-05,
+      "loss": 1.9551,
+      "step": 834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7440547749410525e-05,
+      "loss": 1.7749,
+      "step": 835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.74333650822394e-05,
+      "loss": 1.5487,
+      "step": 836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7426173834215908e-05,
+      "loss": 1.6781,
+      "step": 837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.74189740136414e-05,
+      "loss": 1.8144,
+      "step": 838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7411765628827135e-05,
+      "loss": 1.94,
+      "step": 839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7404548688094256e-05,
+      "loss": 1.7038,
+      "step": 840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.739732319977378e-05,
+      "loss": 1.8419,
+      "step": 841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7390089172206594e-05,
+      "loss": 1.6915,
+      "step": 842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.738284661374344e-05,
+      "loss": 1.8341,
+      "step": 843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.737559553274491e-05,
+      "loss": 1.775,
+      "step": 844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.736833593758143e-05,
+      "loss": 1.7329,
+      "step": 845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.736106783663326e-05,
+      "loss": 1.9322,
+      "step": 846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7353791238290482e-05,
+      "loss": 1.6226,
+      "step": 847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7346506150952967e-05,
+      "loss": 1.9106,
+      "step": 848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7339212583030417e-05,
+      "loss": 1.7041,
+      "step": 849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7331910542942298e-05,
+      "loss": 1.5396,
+      "step": 850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7324600039117862e-05,
+      "loss": 1.6297,
+      "step": 851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7317281079996142e-05,
+      "loss": 1.5199,
+      "step": 852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7309953674025916e-05,
+      "loss": 1.7743,
+      "step": 853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7302617829665725e-05,
+      "loss": 1.7561,
+      "step": 854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7295273555383843e-05,
+      "loss": 1.6274,
+      "step": 855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7287920859658284e-05,
+      "loss": 1.8578,
+      "step": 856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7280559750976775e-05,
+      "loss": 1.5625,
+      "step": 857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7273190237836757e-05,
+      "loss": 1.8671,
+      "step": 858
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7265812328745372e-05,
+      "loss": 1.5314,
+      "step": 859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.725842603221946e-05,
+      "loss": 1.6224,
+      "step": 860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7251031356785534e-05,
+      "loss": 1.9108,
+      "step": 861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7243628310979793e-05,
+      "loss": 1.8679,
+      "step": 862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7236216903348078e-05,
+      "loss": 1.4604,
+      "step": 863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7228797142445902e-05,
+      "loss": 1.7644,
+      "step": 864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.722136903683841e-05,
+      "loss": 1.578,
+      "step": 865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7213932595100384e-05,
+      "loss": 1.7849,
+      "step": 866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.720648782581623e-05,
+      "loss": 1.7362,
+      "step": 867
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7199034737579962e-05,
+      "loss": 1.609,
+      "step": 868
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.71915733389952e-05,
+      "loss": 1.5514,
+      "step": 869
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7184103638675157e-05,
+      "loss": 1.6677,
+      "step": 870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7176625645242637e-05,
+      "loss": 1.6335,
+      "step": 871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.716913936733e-05,
+      "loss": 1.8014,
+      "step": 872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7161644813579187e-05,
+      "loss": 1.7109,
+      "step": 873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.715414199264168e-05,
+      "loss": 1.5313,
+      "step": 874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.714663091317851e-05,
+      "loss": 1.6772,
+      "step": 875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.713911158386024e-05,
+      "loss": 1.9867,
+      "step": 876
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.713158401336696e-05,
+      "loss": 1.6982,
+      "step": 877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7124048210388268e-05,
+      "loss": 1.8234,
+      "step": 878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7116504183623267e-05,
+      "loss": 1.8332,
+      "step": 879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7108951941780553e-05,
+      "loss": 1.6124,
+      "step": 880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.710139149357821e-05,
+      "loss": 1.5396,
+      "step": 881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.709382284774379e-05,
+      "loss": 1.7418,
+      "step": 882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7086246013014304e-05,
+      "loss": 1.6085,
+      "step": 883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7078660998136232e-05,
+      "loss": 1.5135,
+      "step": 884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 1.7902,
+      "step": 885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.706346646296739e-05,
+      "loss": 1.6506,
+      "step": 886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.705585696021673e-05,
+      "loss": 1.8779,
+      "step": 887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.704823931239769e-05,
+      "loss": 1.7934,
+      "step": 888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.704061352830384e-05,
+      "loss": 1.7081,
+      "step": 889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7032979616738167e-05,
+      "loss": 1.81,
+      "step": 890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.702533758651302e-05,
+      "loss": 1.6619,
+      "step": 891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7017687446450134e-05,
+      "loss": 1.5481,
+      "step": 892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.701002920538059e-05,
+      "loss": 1.5529,
+      "step": 893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7002362872144843e-05,
+      "loss": 1.5711,
+      "step": 894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.699468845559267e-05,
+      "loss": 1.8462,
+      "step": 895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6987005964583186e-05,
+      "loss": 1.6297,
+      "step": 896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6979315407984826e-05,
+      "loss": 1.9974,
+      "step": 897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.697161679467534e-05,
+      "loss": 1.7201,
+      "step": 898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6963910133541774e-05,
+      "loss": 1.6625,
+      "step": 899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.695619543348046e-05,
+      "loss": 1.8369,
+      "step": 900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6948472703397022e-05,
+      "loss": 1.7757,
+      "step": 901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6940741952206342e-05,
+      "loss": 1.7673,
+      "step": 902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6933003188832565e-05,
+      "loss": 1.7776,
+      "step": 903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6925256422209094e-05,
+      "loss": 1.7101,
+      "step": 904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6917501661278555e-05,
+      "loss": 1.9591,
+      "step": 905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6909738914992812e-05,
+      "loss": 1.9096,
+      "step": 906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6901968192312943e-05,
+      "loss": 1.5387,
+      "step": 907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.689418950220924e-05,
+      "loss": 1.8123,
+      "step": 908
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6886402853661187e-05,
+      "loss": 1.5839,
+      "step": 909
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6878608255657457e-05,
+      "loss": 1.7317,
+      "step": 910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6870805717195896e-05,
+      "loss": 1.6013,
+      "step": 911
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.686299524728352e-05,
+      "loss": 1.8742,
+      "step": 912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.68551768549365e-05,
+      "loss": 1.572,
+      "step": 913
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6847350549180148e-05,
+      "loss": 1.6005,
+      "step": 914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6839516339048922e-05,
+      "loss": 1.5211,
+      "step": 915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.68316742335864e-05,
+      "loss": 1.7892,
+      "step": 916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6823824241845263e-05,
+      "loss": 1.7745,
+      "step": 917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6815966372887305e-05,
+      "loss": 1.9154,
+      "step": 918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.680810063578342e-05,
+      "loss": 1.6781,
+      "step": 919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.680022703961358e-05,
+      "loss": 1.5625,
+      "step": 920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6792345593466812e-05,
+      "loss": 1.6617,
+      "step": 921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6784456306441234e-05,
+      "loss": 1.6671,
+      "step": 922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6776559187643994e-05,
+      "loss": 1.4649,
+      "step": 923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.676865424619129e-05,
+      "loss": 1.7191,
+      "step": 924
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6760741491208344e-05,
+      "loss": 1.8322,
+      "step": 925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.675282093182941e-05,
+      "loss": 1.8016,
+      "step": 926
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6744892577197735e-05,
+      "loss": 1.8184,
+      "step": 927
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6736956436465573e-05,
+      "loss": 1.8148,
+      "step": 928
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.672901251879417e-05,
+      "loss": 1.9649,
+      "step": 929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.672106083335374e-05,
+      "loss": 1.6884,
+      "step": 930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6713101389323475e-05,
+      "loss": 1.8313,
+      "step": 931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6705134195891507e-05,
+      "loss": 1.7527,
+      "step": 932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6697159262254934e-05,
+      "loss": 1.8313,
+      "step": 933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6689176597619773e-05,
+      "loss": 1.8027,
+      "step": 934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6681186211200974e-05,
+      "loss": 1.9283,
+      "step": 935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6673188112222394e-05,
+      "loss": 1.7859,
+      "step": 936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6665182309916804e-05,
+      "loss": 1.5358,
+      "step": 937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6657168813525855e-05,
+      "loss": 1.6628,
+      "step": 938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6649147632300085e-05,
+      "loss": 1.9546,
+      "step": 939
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6641118775498906e-05,
+      "loss": 1.7629,
+      "step": 940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.663308225239059e-05,
+      "loss": 1.5453,
+      "step": 941
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.662503807225225e-05,
+      "loss": 1.7409,
+      "step": 942
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.661698624436985e-05,
+      "loss": 1.8509,
+      "step": 943
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.660892677803818e-05,
+      "loss": 1.8324,
+      "step": 944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.660085968256084e-05,
+      "loss": 1.7873,
+      "step": 945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.659278496725024e-05,
+      "loss": 1.683,
+      "step": 946
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.658470264142759e-05,
+      "loss": 1.6149,
+      "step": 947
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6576612714422888e-05,
+      "loss": 1.6758,
+      "step": 948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6568515195574893e-05,
+      "loss": 1.4733,
+      "step": 949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6560410094231144e-05,
+      "loss": 1.6312,
+      "step": 950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.655229741974792e-05,
+      "loss": 1.8033,
+      "step": 951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6544177181490253e-05,
+      "loss": 1.7936,
+      "step": 952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6536049388831897e-05,
+      "loss": 1.7148,
+      "step": 953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6527914051155328e-05,
+      "loss": 1.5131,
+      "step": 954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6519771177851736e-05,
+      "loss": 1.8227,
+      "step": 955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6511620778321016e-05,
+      "loss": 1.8475,
+      "step": 956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6503462861971733e-05,
+      "loss": 1.6473,
+      "step": 957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6495297438221145e-05,
+      "loss": 2.0009,
+      "step": 958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.648712451649517e-05,
+      "loss": 1.7045,
+      "step": 959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6478944106228383e-05,
+      "loss": 1.7114,
+      "step": 960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6470756216864e-05,
+      "loss": 1.5554,
+      "step": 961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6462560857853876e-05,
+      "loss": 1.5655,
+      "step": 962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6454358038658483e-05,
+      "loss": 1.676,
+      "step": 963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6446147768746913e-05,
+      "loss": 1.6189,
+      "step": 964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.643793005759685e-05,
+      "loss": 1.6477,
+      "step": 965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6429704914694573e-05,
+      "loss": 1.5552,
+      "step": 966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6421472349534938e-05,
+      "loss": 1.9597,
+      "step": 967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6413232371621372e-05,
+      "loss": 1.7342,
+      "step": 968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6404984990465857e-05,
+      "loss": 1.7305,
+      "step": 969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6396730215588913e-05,
+      "loss": 1.9098,
+      "step": 970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.638846805651961e-05,
+      "loss": 1.7734,
+      "step": 971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6380198522795542e-05,
+      "loss": 1.6998,
+      "step": 972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6371921623962798e-05,
+      "loss": 1.7381,
+      "step": 973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6363637369575984e-05,
+      "loss": 1.7009,
+      "step": 974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6355345769198194e-05,
+      "loss": 1.6588,
+      "step": 975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6347046832401e-05,
+      "loss": 1.7129,
+      "step": 976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6338740568764443e-05,
+      "loss": 1.7798,
+      "step": 977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.633042698787703e-05,
+      "loss": 1.6635,
+      "step": 978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6322106099335698e-05,
+      "loss": 1.7802,
+      "step": 979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6313777912745838e-05,
+      "loss": 1.7116,
+      "step": 980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6305442437721253e-05,
+      "loss": 1.4266,
+      "step": 981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6297099683884163e-05,
+      "loss": 1.5949,
+      "step": 982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6288749660865195e-05,
+      "loss": 1.768,
+      "step": 983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.628039237830336e-05,
+      "loss": 1.6451,
+      "step": 984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6272027845846057e-05,
+      "loss": 1.5685,
+      "step": 985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.626365607314905e-05,
+      "loss": 1.5497,
+      "step": 986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6255277069876454e-05,
+      "loss": 1.5408,
+      "step": 987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6246890845700745e-05,
+      "loss": 1.6302,
+      "step": 988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.623849741030272e-05,
+      "loss": 1.845,
+      "step": 989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6230096773371514e-05,
+      "loss": 1.6124,
+      "step": 990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6221688944604568e-05,
+      "loss": 1.8816,
+      "step": 991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6213273933707622e-05,
+      "loss": 1.8108,
+      "step": 992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6204851750394715e-05,
+      "loss": 1.6138,
+      "step": 993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.619642240438816e-05,
+      "loss": 2.0318,
+      "step": 994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6187985905418536e-05,
+      "loss": 1.8554,
+      "step": 995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.617954226322469e-05,
+      "loss": 1.597,
+      "step": 996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.61710914875537e-05,
+      "loss": 1.7496,
+      "step": 997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.616263358816089e-05,
+      "loss": 1.8927,
+      "step": 998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6154168574809803e-05,
+      "loss": 1.5725,
+      "step": 999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6145696457272203e-05,
+      "loss": 1.5245,
+      "step": 1000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.613721724532803e-05,
+      "loss": 1.9525,
+      "step": 1001
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.612873094876545e-05,
+      "loss": 1.6676,
+      "step": 1002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.612023757738077e-05,
+      "loss": 1.51,
+      "step": 1003
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6111737140978495e-05,
+      "loss": 1.7033,
+      "step": 1004
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6103229649371264e-05,
+      "loss": 1.6155,
+      "step": 1005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6094715112379874e-05,
+      "loss": 1.8862,
+      "step": 1006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6086193539833246e-05,
+      "loss": 1.6979,
+      "step": 1007
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6077664941568426e-05,
+      "loss": 1.8855,
+      "step": 1008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6069129327430575e-05,
+      "loss": 1.715,
+      "step": 1009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6060586707272943e-05,
+      "loss": 1.7139,
+      "step": 1010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6052037090956877e-05,
+      "loss": 1.7741,
+      "step": 1011
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6043480488351792e-05,
+      "loss": 1.4925,
+      "step": 1012
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6034916909335177e-05,
+      "loss": 1.4427,
+      "step": 1013
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6026346363792565e-05,
+      "loss": 1.7371,
+      "step": 1014
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6017768861617537e-05,
+      "loss": 1.6062,
+      "step": 1015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6009184412711703e-05,
+      "loss": 1.6185,
+      "step": 1016
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6000593026984692e-05,
+      "loss": 1.7214,
+      "step": 1017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.599199471435414e-05,
+      "loss": 1.6419,
+      "step": 1018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5983389484745677e-05,
+      "loss": 1.6851,
+      "step": 1019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5974777348092925e-05,
+      "loss": 1.8249,
+      "step": 1020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5966158314337472e-05,
+      "loss": 1.641,
+      "step": 1021
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5957532393428872e-05,
+      "loss": 1.7478,
+      "step": 1022
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5948899595324625e-05,
+      "loss": 2.0402,
+      "step": 1023
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5940259929990177e-05,
+      "loss": 1.6596,
+      "step": 1024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.593161340739889e-05,
+      "loss": 1.4834,
+      "step": 1025
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5922960037532057e-05,
+      "loss": 1.5842,
+      "step": 1026
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5914299830378865e-05,
+      "loss": 1.8075,
+      "step": 1027
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.590563279593639e-05,
+      "loss": 1.3571,
+      "step": 1028
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.58969589442096e-05,
+      "loss": 1.4079,
+      "step": 1029
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.588827828521133e-05,
+      "loss": 1.3868,
+      "step": 1030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5879590828962267e-05,
+      "loss": 1.4351,
+      "step": 1031
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5870896585490952e-05,
+      "loss": 1.7236,
+      "step": 1032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5862195564833748e-05,
+      "loss": 1.371,
+      "step": 1033
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.585348777703486e-05,
+      "loss": 1.8216,
+      "step": 1034
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5844773232146297e-05,
+      "loss": 1.327,
+      "step": 1035
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5836051940227857e-05,
+      "loss": 1.7791,
+      "step": 1036
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5827323911347146e-05,
+      "loss": 1.3035,
+      "step": 1037
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.581858915557953e-05,
+      "loss": 1.5431,
+      "step": 1038
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5809847683008152e-05,
+      "loss": 1.6112,
+      "step": 1039
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.58010995037239e-05,
+      "loss": 1.4338,
+      "step": 1040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5792344627825416e-05,
+      "loss": 1.8351,
+      "step": 1041
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5783583065419054e-05,
+      "loss": 1.6719,
+      "step": 1042
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5774814826618906e-05,
+      "loss": 1.5407,
+      "step": 1043
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5766039921546762e-05,
+      "loss": 1.538,
+      "step": 1044
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5757258360332103e-05,
+      "loss": 1.4798,
+      "step": 1045
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5748470153112093e-05,
+      "loss": 1.6694,
+      "step": 1046
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5739675310031588e-05,
+      "loss": 1.4276,
+      "step": 1047
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5730873841243082e-05,
+      "loss": 1.466,
+      "step": 1048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.572206575690672e-05,
+      "loss": 1.4486,
+      "step": 1049
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.57132510671903e-05,
+      "loss": 1.2553,
+      "step": 1050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5704429782269222e-05,
+      "loss": 1.61,
+      "step": 1051
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5695601912326518e-05,
+      "loss": 1.5392,
+      "step": 1052
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5686767467552812e-05,
+      "loss": 1.629,
+      "step": 1053
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5677926458146327e-05,
+      "loss": 1.499,
+      "step": 1054
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5669078894312848e-05,
+      "loss": 1.5372,
+      "step": 1055
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.566022478626574e-05,
+      "loss": 1.4266,
+      "step": 1056
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.565136414422592e-05,
+      "loss": 1.6381,
+      "step": 1057
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5642496978421842e-05,
+      "loss": 1.523,
+      "step": 1058
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.56336232990895e-05,
+      "loss": 1.5242,
+      "step": 1059
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5624743116472398e-05,
+      "loss": 1.4173,
+      "step": 1060
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5615856440821547e-05,
+      "loss": 1.4201,
+      "step": 1061
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.560696328239547e-05,
+      "loss": 1.3505,
+      "step": 1062
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5598063651460146e-05,
+      "loss": 1.4402,
+      "step": 1063
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5589157558289047e-05,
+      "loss": 1.6149,
+      "step": 1064
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.55802450131631e-05,
+      "loss": 1.4408,
+      "step": 1065
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5571326026370676e-05,
+      "loss": 1.4283,
+      "step": 1066
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.556240060820758e-05,
+      "loss": 1.5737,
+      "step": 1067
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5553468768977057e-05,
+      "loss": 1.325,
+      "step": 1068
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5544530518989744e-05,
+      "loss": 1.457,
+      "step": 1069
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5535585868563688e-05,
+      "loss": 1.8669,
+      "step": 1070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5526634828024323e-05,
+      "loss": 1.4187,
+      "step": 1071
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.551767740770446e-05,
+      "loss": 1.3371,
+      "step": 1072
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.550871361794428e-05,
+      "loss": 1.3672,
+      "step": 1073
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5499743469091303e-05,
+      "loss": 1.4158,
+      "step": 1074
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5490766971500408e-05,
+      "loss": 1.478,
+      "step": 1075
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5481784135533784e-05,
+      "loss": 1.6453,
+      "step": 1076
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5472794971560952e-05,
+      "loss": 1.4174,
+      "step": 1077
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5463799489958727e-05,
+      "loss": 1.458,
+      "step": 1078
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5454797701111225e-05,
+      "loss": 1.7133,
+      "step": 1079
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.544578961540984e-05,
+      "loss": 1.7226,
+      "step": 1080
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5436775243253232e-05,
+      "loss": 1.5548,
+      "step": 1081
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.542775459504732e-05,
+      "loss": 1.5957,
+      "step": 1082
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.541872768120527e-05,
+      "loss": 1.6579,
+      "step": 1083
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5409694512147477e-05,
+      "loss": 1.4221,
+      "step": 1084
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5400655098301558e-05,
+      "loss": 1.5109,
+      "step": 1085
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5391609450102346e-05,
+      "loss": 1.1408,
+      "step": 1086
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5382557577991854e-05,
+      "loss": 1.2196,
+      "step": 1087
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.53734994924193e-05,
+      "loss": 1.5867,
+      "step": 1088
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5364435203841058e-05,
+      "loss": 1.6142,
+      "step": 1089
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5355364722720674e-05,
+      "loss": 1.5948,
+      "step": 1090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5346288059528835e-05,
+      "loss": 1.7858,
+      "step": 1091
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5337205224743367e-05,
+      "loss": 1.5192,
+      "step": 1092
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.532811622884922e-05,
+      "loss": 1.3382,
+      "step": 1093
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5319021082338458e-05,
+      "loss": 1.5438,
+      "step": 1094
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5309919795710248e-05,
+      "loss": 1.3998,
+      "step": 1095
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5300812379470832e-05,
+      "loss": 1.4802,
+      "step": 1096
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.529169884413354e-05,
+      "loss": 1.7461,
+      "step": 1097
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5282579200218762e-05,
+      "loss": 1.4801,
+      "step": 1098
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5273453458253942e-05,
+      "loss": 1.5252,
+      "step": 1099
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.526432162877356e-05,
+      "loss": 1.4398,
+      "step": 1100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5255183722319124e-05,
+      "loss": 1.4525,
+      "step": 1101
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5246039749439159e-05,
+      "loss": 1.7295,
+      "step": 1102
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5236889720689187e-05,
+      "loss": 1.6127,
+      "step": 1103
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.522773364663173e-05,
+      "loss": 1.4742,
+      "step": 1104
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.521857153783628e-05,
+      "loss": 1.5412,
+      "step": 1105
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5209403404879305e-05,
+      "loss": 1.619,
+      "step": 1106
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5200229258344212e-05,
+      "loss": 1.9911,
+      "step": 1107
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.519104910882137e-05,
+      "loss": 1.5251,
+      "step": 1108
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5181862966908054e-05,
+      "loss": 1.5937,
+      "step": 1109
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5172670843208477e-05,
+      "loss": 1.2891,
+      "step": 1110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.516347274833375e-05,
+      "loss": 1.5201,
+      "step": 1111
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5154268692901876e-05,
+      "loss": 1.7639,
+      "step": 1112
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5145058687537733e-05,
+      "loss": 1.66,
+      "step": 1113
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5135842742873077e-05,
+      "loss": 1.3905,
+      "step": 1114
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5126620869546517e-05,
+      "loss": 1.5876,
+      "step": 1115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5117393078203506e-05,
+      "loss": 1.5667,
+      "step": 1116
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5108159379496325e-05,
+      "loss": 1.6344,
+      "step": 1117
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5098919784084083e-05,
+      "loss": 1.0704,
+      "step": 1118
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5089674302632683e-05,
+      "loss": 1.661,
+      "step": 1119
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5080422945814835e-05,
+      "loss": 1.8619,
+      "step": 1120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5071165724310021e-05,
+      "loss": 1.5473,
+      "step": 1121
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5061902648804503e-05,
+      "loss": 1.6422,
+      "step": 1122
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5052633729991296e-05,
+      "loss": 1.6047,
+      "step": 1123
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5043358978570158e-05,
+      "loss": 1.4343,
+      "step": 1124
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.503407840524758e-05,
+      "loss": 1.4234,
+      "step": 1125
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.502479202073678e-05,
+      "loss": 1.5382,
+      "step": 1126
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.501549983575768e-05,
+      "loss": 1.5557,
+      "step": 1127
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5006201861036895e-05,
+      "loss": 1.6375,
+      "step": 1128
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4996898107307729e-05,
+      "loss": 1.5783,
+      "step": 1129
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4987588585310154e-05,
+      "loss": 1.4857,
+      "step": 1130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4978273305790798e-05,
+      "loss": 1.4683,
+      "step": 1131
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4968952279502945e-05,
+      "loss": 1.572,
+      "step": 1132
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4959625517206502e-05,
+      "loss": 1.5051,
+      "step": 1133
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4950293029668004e-05,
+      "loss": 1.3847,
+      "step": 1134
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.494095482766059e-05,
+      "loss": 1.7328,
+      "step": 1135
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4931610921964004e-05,
+      "loss": 1.4809,
+      "step": 1136
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4922261323364561e-05,
+      "loss": 1.3702,
+      "step": 1137
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4912906042655164e-05,
+      "loss": 1.7492,
+      "step": 1138
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4903545090635262e-05,
+      "loss": 1.5597,
+      "step": 1139
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4894178478110856e-05,
+      "loss": 1.435,
+      "step": 1140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4884806215894483e-05,
+      "loss": 1.6026,
+      "step": 1141
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4875428314805195e-05,
+      "loss": 1.3812,
+      "step": 1142
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4866044785668563e-05,
+      "loss": 1.2505,
+      "step": 1143
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.485665563931665e-05,
+      "loss": 1.4917,
+      "step": 1144
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4847260886588e-05,
+      "loss": 1.488,
+      "step": 1145
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.483786053832763e-05,
+      "loss": 1.4528,
+      "step": 1146
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4828454605387025e-05,
+      "loss": 1.6644,
+      "step": 1147
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4819043098624106e-05,
+      "loss": 1.7426,
+      "step": 1148
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4809626028903237e-05,
+      "loss": 1.5483,
+      "step": 1149
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4800203407095194e-05,
+      "loss": 1.4537,
+      "step": 1150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4790775244077168e-05,
+      "loss": 1.5489,
+      "step": 1151
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4781341550732746e-05,
+      "loss": 1.38,
+      "step": 1152
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4771902337951903e-05,
+      "loss": 1.5639,
+      "step": 1153
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4762457616630972e-05,
+      "loss": 1.4766,
+      "step": 1154
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4753007397672667e-05,
+      "loss": 1.4076,
+      "step": 1155
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.474355169198603e-05,
+      "loss": 1.4983,
+      "step": 1156
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4734090510486435e-05,
+      "loss": 1.4774,
+      "step": 1157
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4724623864095595e-05,
+      "loss": 1.5475,
+      "step": 1158
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4715151763741518e-05,
+      "loss": 1.6004,
+      "step": 1159
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4705674220358507e-05,
+      "loss": 1.4697,
+      "step": 1160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4696191244887156e-05,
+      "loss": 1.5456,
+      "step": 1161
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4686702848274328e-05,
+      "loss": 1.5106,
+      "step": 1162
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4677209041473132e-05,
+      "loss": 1.4553,
+      "step": 1163
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4667709835442944e-05,
+      "loss": 1.6499,
+      "step": 1164
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4658205241149355e-05,
+      "loss": 1.5725,
+      "step": 1165
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4648695269564182e-05,
+      "loss": 1.5387,
+      "step": 1166
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4639179931665448e-05,
+      "loss": 1.5978,
+      "step": 1167
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4629659238437377e-05,
+      "loss": 1.6021,
+      "step": 1168
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4620133200870361e-05,
+      "loss": 1.4084,
+      "step": 1169
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.461060182996098e-05,
+      "loss": 1.3895,
+      "step": 1170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4601065136711955e-05,
+      "loss": 1.4654,
+      "step": 1171
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4591523132132161e-05,
+      "loss": 1.4048,
+      "step": 1172
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4581975827236598e-05,
+      "loss": 1.4343,
+      "step": 1173
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4572423233046386e-05,
+      "loss": 1.6183,
+      "step": 1174
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4562865360588751e-05,
+      "loss": 1.5888,
+      "step": 1175
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4553302220897018e-05,
+      "loss": 1.6468,
+      "step": 1176
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.454373382501058e-05,
+      "loss": 1.5995,
+      "step": 1177
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4534160183974908e-05,
+      "loss": 1.6392,
+      "step": 1178
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4524581308841524e-05,
+      "loss": 1.5639,
+      "step": 1179
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4514997210667991e-05,
+      "loss": 1.251,
+      "step": 1180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4505407900517903e-05,
+      "loss": 1.4649,
+      "step": 1181
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4495813389460875e-05,
+      "loss": 1.416,
+      "step": 1182
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4486213688572513e-05,
+      "loss": 1.5222,
+      "step": 1183
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.447660880893443e-05,
+      "loss": 1.28,
+      "step": 1184
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4466998761634203e-05,
+      "loss": 1.2936,
+      "step": 1185
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4457383557765385e-05,
+      "loss": 1.4656,
+      "step": 1186
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4447763208427474e-05,
+      "loss": 1.4636,
+      "step": 1187
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4438137724725914e-05,
+      "loss": 1.4226,
+      "step": 1188
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.442850711777207e-05,
+      "loss": 1.542,
+      "step": 1189
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4418871398683227e-05,
+      "loss": 1.4573,
+      "step": 1190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4409230578582566e-05,
+      "loss": 1.4758,
+      "step": 1191
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.439958466859916e-05,
+      "loss": 1.645,
+      "step": 1192
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4389933679867957e-05,
+      "loss": 1.4046,
+      "step": 1193
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4380277623529766e-05,
+      "loss": 1.559,
+      "step": 1194
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4370616510731246e-05,
+      "loss": 1.8745,
+      "step": 1195
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4360950352624898e-05,
+      "loss": 1.7127,
+      "step": 1196
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4351279160369041e-05,
+      "loss": 1.6269,
+      "step": 1197
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4341602945127806e-05,
+      "loss": 1.7087,
+      "step": 1198
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4331921718071127e-05,
+      "loss": 1.3341,
+      "step": 1199
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4322235490374718e-05,
+      "loss": 1.4046,
+      "step": 1200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4312544273220068e-05,
+      "loss": 1.3492,
+      "step": 1201
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4302848077794427e-05,
+      "loss": 1.3417,
+      "step": 1202
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4293146915290792e-05,
+      "loss": 1.7759,
+      "step": 1203
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4283440796907888e-05,
+      "loss": 1.8054,
+      "step": 1204
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4273729733850168e-05,
+      "loss": 1.4621,
+      "step": 1205
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.426401373732779e-05,
+      "loss": 1.3796,
+      "step": 1206
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4254292818556606e-05,
+      "loss": 1.4166,
+      "step": 1207
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4244566988758152e-05,
+      "loss": 1.5732,
+      "step": 1208
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4234836259159635e-05,
+      "loss": 1.5566,
+      "step": 1209
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.422510064099391e-05,
+      "loss": 1.4763,
+      "step": 1210
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4215360145499486e-05,
+      "loss": 1.5692,
+      "step": 1211
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4205614783920497e-05,
+      "loss": 1.4426,
+      "step": 1212
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.419586456750669e-05,
+      "loss": 1.6941,
+      "step": 1213
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4186109507513425e-05,
+      "loss": 1.3912,
+      "step": 1214
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4176349615201648e-05,
+      "loss": 1.5505,
+      "step": 1215
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.416658490183788e-05,
+      "loss": 1.7898,
+      "step": 1216
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4156815378694218e-05,
+      "loss": 1.6724,
+      "step": 1217
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4147041057048303e-05,
+      "loss": 1.4052,
+      "step": 1218
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4137261948183314e-05,
+      "loss": 1.5454,
+      "step": 1219
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4127478063387961e-05,
+      "loss": 1.6163,
+      "step": 1220
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4117689413956464e-05,
+      "loss": 1.7238,
+      "step": 1221
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4107896011188546e-05,
+      "loss": 1.4835,
+      "step": 1222
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4098097866389417e-05,
+      "loss": 1.5385,
+      "step": 1223
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4088294990869757e-05,
+      "loss": 1.3845,
+      "step": 1224
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4078487395945712e-05,
+      "loss": 1.3795,
+      "step": 1225
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4068675092938872e-05,
+      "loss": 1.4493,
+      "step": 1226
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4058858093176269e-05,
+      "loss": 1.6271,
+      "step": 1227
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.404903640799034e-05,
+      "loss": 1.6677,
+      "step": 1228
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.403921004871895e-05,
+      "loss": 1.5685,
+      "step": 1229
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4029379026705352e-05,
+      "loss": 1.8045,
+      "step": 1230
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4019543353298179e-05,
+      "loss": 1.4206,
+      "step": 1231
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4009703039851437e-05,
+      "loss": 1.5676,
+      "step": 1232
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3999858097724486e-05,
+      "loss": 1.2841,
+      "step": 1233
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3990008538282027e-05,
+      "loss": 1.3163,
+      "step": 1234
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3980154372894101e-05,
+      "loss": 1.3957,
+      "step": 1235
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3970295612936053e-05,
+      "loss": 1.6965,
+      "step": 1236
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3960432269788542e-05,
+      "loss": 1.4793,
+      "step": 1237
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3950564354837512e-05,
+      "loss": 1.45,
+      "step": 1238
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3940691879474186e-05,
+      "loss": 1.3742,
+      "step": 1239
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3930814855095056e-05,
+      "loss": 1.5464,
+      "step": 1240
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3920933293101854e-05,
+      "loss": 1.4473,
+      "step": 1241
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.391104720490156e-05,
+      "loss": 1.4857,
+      "step": 1242
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3901156601906377e-05,
+      "loss": 1.6068,
+      "step": 1243
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.389126149553372e-05,
+      "loss": 1.6329,
+      "step": 1244
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3881361897206196e-05,
+      "loss": 1.6012,
+      "step": 1245
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.387145781835161e-05,
+      "loss": 1.6549,
+      "step": 1246
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3861549270402926e-05,
+      "loss": 1.6136,
+      "step": 1247
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3851636264798271e-05,
+      "loss": 1.5298,
+      "step": 1248
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3841718812980926e-05,
+      "loss": 1.4631,
+      "step": 1249
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3831796926399295e-05,
+      "loss": 1.3395,
+      "step": 1250
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3821870616506905e-05,
+      "loss": 1.7195,
+      "step": 1251
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3811939894762387e-05,
+      "loss": 1.5456,
+      "step": 1252
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3802004772629468e-05,
+      "loss": 1.2659,
+      "step": 1253
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3792065261576953e-05,
+      "loss": 1.6262,
+      "step": 1254
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3782121373078715e-05,
+      "loss": 1.3683,
+      "step": 1255
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3772173118613679e-05,
+      "loss": 1.671,
+      "step": 1256
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3762220509665803e-05,
+      "loss": 1.6182,
+      "step": 1257
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3752263557724088e-05,
+      "loss": 1.7478,
+      "step": 1258
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3742302274282532e-05,
+      "loss": 1.643,
+      "step": 1259
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3732336670840146e-05,
+      "loss": 1.6071,
+      "step": 1260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3722366758900915e-05,
+      "loss": 1.328,
+      "step": 1261
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3712392549973814e-05,
+      "loss": 1.6267,
+      "step": 1262
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3702414055572759e-05,
+      "loss": 1.636,
+      "step": 1263
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3692431287216627e-05,
+      "loss": 1.0392,
+      "step": 1264
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3682444256429226e-05,
+      "loss": 1.2935,
+      "step": 1265
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3672452974739278e-05,
+      "loss": 1.3875,
+      "step": 1266
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3662457453680427e-05,
+      "loss": 1.629,
+      "step": 1267
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3652457704791191e-05,
+      "loss": 1.5141,
+      "step": 1268
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3642453739614982e-05,
+      "loss": 1.7621,
+      "step": 1269
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3632445569700078e-05,
+      "loss": 1.5815,
+      "step": 1270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3622433206599603e-05,
+      "loss": 1.6716,
+      "step": 1271
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3612416661871532e-05,
+      "loss": 1.6589,
+      "step": 1272
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.360239594707866e-05,
+      "loss": 1.1986,
+      "step": 1273
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3592371073788595e-05,
+      "loss": 1.5435,
+      "step": 1274
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3582342053573752e-05,
+      "loss": 1.4024,
+      "step": 1275
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3572308898011328e-05,
+      "loss": 1.421,
+      "step": 1276
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3562271618683298e-05,
+      "loss": 1.3642,
+      "step": 1277
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.355223022717639e-05,
+      "loss": 1.4523,
+      "step": 1278
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3542184735082087e-05,
+      "loss": 1.6044,
+      "step": 1279
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3532135153996595e-05,
+      "loss": 1.5002,
+      "step": 1280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3522081495520856e-05,
+      "loss": 1.3517,
+      "step": 1281
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3512023771260507e-05,
+      "loss": 1.6981,
+      "step": 1282
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.350196199282588e-05,
+      "loss": 1.4958,
+      "step": 1283
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3491896171831991e-05,
+      "loss": 1.5878,
+      "step": 1284
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3481826319898516e-05,
+      "loss": 1.5268,
+      "step": 1285
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.347175244864979e-05,
+      "loss": 1.4466,
+      "step": 1286
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3461674569714786e-05,
+      "loss": 1.4006,
+      "step": 1287
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.345159269472711e-05,
+      "loss": 1.556,
+      "step": 1288
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.344150683532496e-05,
+      "loss": 1.4465,
+      "step": 1289
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3431417003151162e-05,
+      "loss": 1.3849,
+      "step": 1290
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3421323209853105e-05,
+      "loss": 1.6262,
+      "step": 1291
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.341122546708276e-05,
+      "loss": 1.3189,
+      "step": 1292
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3401123786496664e-05,
+      "loss": 1.5887,
+      "step": 1293
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3391018179755886e-05,
+      "loss": 1.6179,
+      "step": 1294
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3380908658526034e-05,
+      "loss": 1.6837,
+      "step": 1295
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3370795234477231e-05,
+      "loss": 1.5105,
+      "step": 1296
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3360677919284117e-05,
+      "loss": 1.5152,
+      "step": 1297
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3350556724625809e-05,
+      "loss": 1.5633,
+      "step": 1298
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3340431662185913e-05,
+      "loss": 1.5434,
+      "step": 1299
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3330302743652492e-05,
+      "loss": 1.2813,
+      "step": 1300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3320169980718065e-05,
+      "loss": 1.7152,
+      "step": 1301
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3310033385079589e-05,
+      "loss": 1.4406,
+      "step": 1302
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3299892968438441e-05,
+      "loss": 1.3858,
+      "step": 1303
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3289748742500415e-05,
+      "loss": 1.6339,
+      "step": 1304
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.32796007189757e-05,
+      "loss": 1.3747,
+      "step": 1305
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3269448909578866e-05,
+      "loss": 1.484,
+      "step": 1306
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3259293326028852e-05,
+      "loss": 1.6567,
+      "step": 1307
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3249133980048967e-05,
+      "loss": 1.4637,
+      "step": 1308
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3238970883366847e-05,
+      "loss": 1.4648,
+      "step": 1309
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3228804047714462e-05,
+      "loss": 1.5431,
+      "step": 1310
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3218633484828108e-05,
+      "loss": 1.6303,
+      "step": 1311
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3208459206448367e-05,
+      "loss": 1.4695,
+      "step": 1312
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3198281224320123e-05,
+      "loss": 1.6769,
+      "step": 1313
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3188099550192537e-05,
+      "loss": 1.5322,
+      "step": 1314
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3177914195819018e-05,
+      "loss": 1.2896,
+      "step": 1315
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3167725172957236e-05,
+      "loss": 1.25,
+      "step": 1316
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.315753249336909e-05,
+      "loss": 1.5092,
+      "step": 1317
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.31473361688207e-05,
+      "loss": 1.4815,
+      "step": 1318
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.31371362110824e-05,
+      "loss": 1.5949,
+      "step": 1319
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.312693263192871e-05,
+      "loss": 1.3182,
+      "step": 1320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3116725443138337e-05,
+      "loss": 1.6017,
+      "step": 1321
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3106514656494147e-05,
+      "loss": 1.385,
+      "step": 1322
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.309630028378317e-05,
+      "loss": 1.5237,
+      "step": 1323
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3086082336796567e-05,
+      "loss": 1.4196,
+      "step": 1324
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3075860827329623e-05,
+      "loss": 1.4194,
+      "step": 1325
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3065635767181748e-05,
+      "loss": 1.601,
+      "step": 1326
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3055407168156438e-05,
+      "loss": 1.4975,
+      "step": 1327
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.304517504206128e-05,
+      "loss": 1.5767,
+      "step": 1328
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3034939400707932e-05,
+      "loss": 1.3225,
+      "step": 1329
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.302470025591211e-05,
+      "loss": 1.4718,
+      "step": 1330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3014457619493576e-05,
+      "loss": 1.639,
+      "step": 1331
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3004211503276116e-05,
+      "loss": 1.7433,
+      "step": 1332
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2993961919087536e-05,
+      "loss": 1.56,
+      "step": 1333
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2983708878759655e-05,
+      "loss": 1.3519,
+      "step": 1334
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2973452394128265e-05,
+      "loss": 1.4561,
+      "step": 1335
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2963192477033148e-05,
+      "loss": 1.4435,
+      "step": 1336
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.295292913931804e-05,
+      "loss": 1.2965,
+      "step": 1337
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2942662392830632e-05,
+      "loss": 1.6036,
+      "step": 1338
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2932392249422536e-05,
+      "loss": 1.4111,
+      "step": 1339
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2922118720949308e-05,
+      "loss": 1.5651,
+      "step": 1340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2911841819270395e-05,
+      "loss": 1.3633,
+      "step": 1341
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.290156155624914e-05,
+      "loss": 1.6841,
+      "step": 1342
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.289127794375277e-05,
+      "loss": 1.5879,
+      "step": 1343
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2880990993652379e-05,
+      "loss": 1.3773,
+      "step": 1344
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2870700717822903e-05,
+      "loss": 1.6074,
+      "step": 1345
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.286040712814314e-05,
+      "loss": 1.6436,
+      "step": 1346
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2850110236495692e-05,
+      "loss": 1.3313,
+      "step": 1347
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2839810054766976e-05,
+      "loss": 1.6385,
+      "step": 1348
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2829506594847218e-05,
+      "loss": 1.3108,
+      "step": 1349
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2819199868630419e-05,
+      "loss": 1.3765,
+      "step": 1350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2808889888014348e-05,
+      "loss": 1.5129,
+      "step": 1351
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2798576664900546e-05,
+      "loss": 1.5171,
+      "step": 1352
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2788260211194277e-05,
+      "loss": 1.574,
+      "step": 1353
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2777940538804545e-05,
+      "loss": 1.3371,
+      "step": 1354
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2767617659644071e-05,
+      "loss": 1.6494,
+      "step": 1355
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2757291585629279e-05,
+      "loss": 1.4826,
+      "step": 1356
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2746962328680267e-05,
+      "loss": 1.5753,
+      "step": 1357
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2736629900720832e-05,
+      "loss": 1.5668,
+      "step": 1358
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2726294313678405e-05,
+      "loss": 1.4189,
+      "step": 1359
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2715955579484076e-05,
+      "loss": 1.6712,
+      "step": 1360
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2705613710072575e-05,
+      "loss": 1.4726,
+      "step": 1361
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2695268717382242e-05,
+      "loss": 1.3699,
+      "step": 1362
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.268492061335502e-05,
+      "loss": 1.3417,
+      "step": 1363
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2674569409936453e-05,
+      "loss": 1.4807,
+      "step": 1364
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2664215119075654e-05,
+      "loss": 1.2893,
+      "step": 1365
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2653857752725305e-05,
+      "loss": 1.4053,
+      "step": 1366
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2643497322841643e-05,
+      "loss": 1.4237,
+      "step": 1367
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2633133841384426e-05,
+      "loss": 1.4772,
+      "step": 1368
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2622767320316953e-05,
+      "loss": 1.5675,
+      "step": 1369
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2612397771606015e-05,
+      "loss": 1.7899,
+      "step": 1370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2602025207221914e-05,
+      "loss": 1.277,
+      "step": 1371
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.259164963913842e-05,
+      "loss": 1.5904,
+      "step": 1372
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2581271079332777e-05,
+      "loss": 1.4314,
+      "step": 1373
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2570889539785683e-05,
+      "loss": 1.2438,
+      "step": 1374
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2560505032481274e-05,
+      "loss": 1.4464,
+      "step": 1375
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2550117569407108e-05,
+      "loss": 1.6312,
+      "step": 1376
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.253972716255416e-05,
+      "loss": 1.724,
+      "step": 1377
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2529333823916807e-05,
+      "loss": 1.5566,
+      "step": 1378
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2518937565492807e-05,
+      "loss": 1.3994,
+      "step": 1379
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2508538399283274e-05,
+      "loss": 1.6813,
+      "step": 1380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2498136337292706e-05,
+      "loss": 1.4131,
+      "step": 1381
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2487731391528919e-05,
+      "loss": 1.1103,
+      "step": 1382
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2477323574003076e-05,
+      "loss": 1.4672,
+      "step": 1383
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2466912896729646e-05,
+      "loss": 1.3076,
+      "step": 1384
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2456499371726398e-05,
+      "loss": 1.5203,
+      "step": 1385
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2446083011014389e-05,
+      "loss": 1.6774,
+      "step": 1386
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2435663826617958e-05,
+      "loss": 1.7135,
+      "step": 1387
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2425241830564695e-05,
+      "loss": 1.4335,
+      "step": 1388
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2414817034885433e-05,
+      "loss": 1.2838,
+      "step": 1389
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2404389451614253e-05,
+      "loss": 1.4817,
+      "step": 1390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.239395909278843e-05,
+      "loss": 1.4776,
+      "step": 1391
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.238352597044846e-05,
+      "loss": 1.3462,
+      "step": 1392
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2373090096638026e-05,
+      "loss": 1.6948,
+      "step": 1393
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2362651483403985e-05,
+      "loss": 1.5238,
+      "step": 1394
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2352210142796359e-05,
+      "loss": 1.5889,
+      "step": 1395
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2341766086868312e-05,
+      "loss": 1.4156,
+      "step": 1396
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2331319327676153e-05,
+      "loss": 1.5625,
+      "step": 1397
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2320869877279297e-05,
+      "loss": 1.4565,
+      "step": 1398
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2310417747740284e-05,
+      "loss": 1.4683,
+      "step": 1399
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2299962951124731e-05,
+      "loss": 1.5503,
+      "step": 1400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2289505499501341e-05,
+      "loss": 1.6381,
+      "step": 1401
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2279045404941883e-05,
+      "loss": 1.3612,
+      "step": 1402
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2268582679521173e-05,
+      "loss": 1.5754,
+      "step": 1403
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.225811733531706e-05,
+      "loss": 1.4143,
+      "step": 1404
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2247649384410431e-05,
+      "loss": 1.4092,
+      "step": 1405
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2237178838885168e-05,
+      "loss": 1.6457,
+      "step": 1406
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2226705710828155e-05,
+      "loss": 1.5011,
+      "step": 1407
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2216230012329251e-05,
+      "loss": 1.4825,
+      "step": 1408
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.220575175548129e-05,
+      "loss": 1.4843,
+      "step": 1409
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2195270952380052e-05,
+      "loss": 1.782,
+      "step": 1410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2184787615124264e-05,
+      "loss": 1.7921,
+      "step": 1411
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2174301755815572e-05,
+      "loss": 1.6019,
+      "step": 1412
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2163813386558535e-05,
+      "loss": 1.6066,
+      "step": 1413
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.215332251946061e-05,
+      "loss": 1.6066,
+      "step": 1414
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2142829166632139e-05,
+      "loss": 1.4195,
+      "step": 1415
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2132333340186334e-05,
+      "loss": 1.5712,
+      "step": 1416
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2121835052239255e-05,
+      "loss": 1.5557,
+      "step": 1417
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2111334314909811e-05,
+      "loss": 1.8155,
+      "step": 1418
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2100831140319735e-05,
+      "loss": 1.3335,
+      "step": 1419
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2090325540593581e-05,
+      "loss": 1.8632,
+      "step": 1420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.207981752785869e-05,
+      "loss": 1.6933,
+      "step": 1421
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2069307114245197e-05,
+      "loss": 1.4222,
+      "step": 1422
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2058794311886006e-05,
+      "loss": 1.6138,
+      "step": 1423
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2048279132916778e-05,
+      "loss": 1.4912,
+      "step": 1424
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.203776158947592e-05,
+      "loss": 1.5629,
+      "step": 1425
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2027241693704567e-05,
+      "loss": 1.318,
+      "step": 1426
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.201671945774657e-05,
+      "loss": 1.5856,
+      "step": 1427
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2006194893748475e-05,
+      "loss": 1.4944,
+      "step": 1428
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.199566801385953e-05,
+      "loss": 1.7742,
+      "step": 1429
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1985138830231638e-05,
+      "loss": 1.5245,
+      "step": 1430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1974607355019379e-05,
+      "loss": 1.638,
+      "step": 1431
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1964073600379966e-05,
+      "loss": 1.448,
+      "step": 1432
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1953537578473248e-05,
+      "loss": 1.4826,
+      "step": 1433
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1942999301461694e-05,
+      "loss": 1.4911,
+      "step": 1434
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1932458781510368e-05,
+      "loss": 1.4674,
+      "step": 1435
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.192191603078693e-05,
+      "loss": 1.6542,
+      "step": 1436
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1911371061461612e-05,
+      "loss": 1.6934,
+      "step": 1437
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1900823885707216e-05,
+      "loss": 1.5369,
+      "step": 1438
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1890274515699072e-05,
+      "loss": 1.645,
+      "step": 1439
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1879722963615063e-05,
+      "loss": 1.4193,
+      "step": 1440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.186916924163558e-05,
+      "loss": 1.3539,
+      "step": 1441
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1858613361943518e-05,
+      "loss": 1.2849,
+      "step": 1442
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1848055336724273e-05,
+      "loss": 1.584,
+      "step": 1443
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1837495178165706e-05,
+      "loss": 1.2871,
+      "step": 1444
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1826932898458143e-05,
+      "loss": 1.5704,
+      "step": 1445
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1816368509794365e-05,
+      "loss": 1.7231,
+      "step": 1446
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1805802024369584e-05,
+      "loss": 1.5432,
+      "step": 1447
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.179523345438143e-05,
+      "loss": 1.36,
+      "step": 1448
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1784662812029941e-05,
+      "loss": 1.5682,
+      "step": 1449
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.177409010951755e-05,
+      "loss": 1.3113,
+      "step": 1450
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.176351535904906e-05,
+      "loss": 1.3686,
+      "step": 1451
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1752938572831653e-05,
+      "loss": 1.5892,
+      "step": 1452
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1742359763074844e-05,
+      "loss": 1.6072,
+      "step": 1453
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1731778941990497e-05,
+      "loss": 1.8373,
+      "step": 1454
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1721196121792788e-05,
+      "loss": 1.5935,
+      "step": 1455
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.171061131469821e-05,
+      "loss": 1.6446,
+      "step": 1456
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.170002453292554e-05,
+      "loss": 1.2029,
+      "step": 1457
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1689435788695844e-05,
+      "loss": 1.3958,
+      "step": 1458
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1678845094232446e-05,
+      "loss": 1.3595,
+      "step": 1459
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1668252461760928e-05,
+      "loss": 1.4901,
+      "step": 1460
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.16576579035091e-05,
+      "loss": 1.4942,
+      "step": 1461
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1647061431707e-05,
+      "loss": 1.5878,
+      "step": 1462
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1636463058586882e-05,
+      "loss": 1.589,
+      "step": 1463
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1625862796383187e-05,
+      "loss": 1.504,
+      "step": 1464
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.161526065733253e-05,
+      "loss": 1.6544,
+      "step": 1465
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1604656653673707e-05,
+      "loss": 1.3777,
+      "step": 1466
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1594050797647658e-05,
+      "loss": 1.3827,
+      "step": 1467
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1583443101497465e-05,
+      "loss": 1.3612,
+      "step": 1468
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.157283357746833e-05,
+      "loss": 1.4462,
+      "step": 1469
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.156222223780757e-05,
+      "loss": 1.6413,
+      "step": 1470
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.155160909476459e-05,
+      "loss": 1.696,
+      "step": 1471
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1540994160590891e-05,
+      "loss": 1.4645,
+      "step": 1472
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1530377447540026e-05,
+      "loss": 1.6276,
+      "step": 1473
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1519758967867608e-05,
+      "loss": 1.4645,
+      "step": 1474
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1509138733831295e-05,
+      "loss": 1.3701,
+      "step": 1475
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1498516757690759e-05,
+      "loss": 1.5479,
+      "step": 1476
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.148789305170769e-05,
+      "loss": 1.3279,
+      "step": 1477
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1477267628145777e-05,
+      "loss": 1.6175,
+      "step": 1478
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1466640499270683e-05,
+      "loss": 1.3472,
+      "step": 1479
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1456011677350052e-05,
+      "loss": 1.2008,
+      "step": 1480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1445381174653468e-05,
+      "loss": 1.5397,
+      "step": 1481
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1434749003452467e-05,
+      "loss": 1.5215,
+      "step": 1482
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1424115176020503e-05,
+      "loss": 1.6105,
+      "step": 1483
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.141347970463295e-05,
+      "loss": 1.4648,
+      "step": 1484
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1402842601567075e-05,
+      "loss": 1.4675,
+      "step": 1485
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1392203879102027e-05,
+      "loss": 1.4437,
+      "step": 1486
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1381563549518823e-05,
+      "loss": 1.668,
+      "step": 1487
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1370921625100344e-05,
+      "loss": 1.6595,
+      "step": 1488
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1360278118131303e-05,
+      "loss": 1.3414,
+      "step": 1489
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1349633040898246e-05,
+      "loss": 1.7191,
+      "step": 1490
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1338986405689528e-05,
+      "loss": 1.2963,
+      "step": 1491
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1328338224795304e-05,
+      "loss": 1.4483,
+      "step": 1492
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.131768851050751e-05,
+      "loss": 1.6848,
+      "step": 1493
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1307037275119854e-05,
+      "loss": 1.5495,
+      "step": 1494
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1296384530927807e-05,
+      "loss": 1.7276,
+      "step": 1495
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1285730290228572e-05,
+      "loss": 1.7049,
+      "step": 1496
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.127507456532108e-05,
+      "loss": 1.5022,
+      "step": 1497
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1264417368505981e-05,
+      "loss": 1.6119,
+      "step": 1498
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1253758712085626e-05,
+      "loss": 1.3764,
+      "step": 1499
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1243098608364039e-05,
+      "loss": 1.6344,
+      "step": 1500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1232437069646924e-05,
+      "loss": 1.3564,
+      "step": 1501
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1221774108241646e-05,
+      "loss": 1.3322,
+      "step": 1502
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1211109736457194e-05,
+      "loss": 1.6564,
+      "step": 1503
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1200443966604207e-05,
+      "loss": 1.4543,
+      "step": 1504
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1189776810994923e-05,
+      "loss": 1.4062,
+      "step": 1505
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.117910828194319e-05,
+      "loss": 1.3152,
+      "step": 1506
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1168438391764429e-05,
+      "loss": 1.5472,
+      "step": 1507
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1157767152775643e-05,
+      "loss": 1.4108,
+      "step": 1508
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1147094577295385e-05,
+      "loss": 1.4983,
+      "step": 1509
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1136420677643763e-05,
+      "loss": 1.4504,
+      "step": 1510
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1125745466142394e-05,
+      "loss": 1.2936,
+      "step": 1511
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1115068955114428e-05,
+      "loss": 1.5347,
+      "step": 1512
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1104391156884502e-05,
+      "loss": 1.6413,
+      "step": 1513
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1093712083778748e-05,
+      "loss": 1.6272,
+      "step": 1514
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.108303174812476e-05,
+      "loss": 1.6904,
+      "step": 1515
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.10723501622516e-05,
+      "loss": 1.4042,
+      "step": 1516
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.106166733848977e-05,
+      "loss": 1.335,
+      "step": 1517
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1050983289171195e-05,
+      "loss": 1.6112,
+      "step": 1518
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1040298026629221e-05,
+      "loss": 1.5474,
+      "step": 1519
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1029611563198595e-05,
+      "loss": 1.621,
+      "step": 1520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1018923911215443e-05,
+      "loss": 1.528,
+      "step": 1521
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1008235083017272e-05,
+      "loss": 1.5818,
+      "step": 1522
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0997545090942943e-05,
+      "loss": 1.7977,
+      "step": 1523
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0986853947332656e-05,
+      "loss": 1.4196,
+      "step": 1524
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0976161664527949e-05,
+      "loss": 1.3376,
+      "step": 1525
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.096546825487167e-05,
+      "loss": 1.5287,
+      "step": 1526
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0954773730707964e-05,
+      "loss": 1.5497,
+      "step": 1527
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0944078104382276e-05,
+      "loss": 1.4437,
+      "step": 1528
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0933381388241306e-05,
+      "loss": 1.4014,
+      "step": 1529
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.092268359463302e-05,
+      "loss": 1.5057,
+      "step": 1530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0911984735906635e-05,
+      "loss": 1.5877,
+      "step": 1531
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0901284824412588e-05,
+      "loss": 1.2757,
+      "step": 1532
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0890583872502529e-05,
+      "loss": 1.6175,
+      "step": 1533
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0879881892529325e-05,
+      "loss": 1.3167,
+      "step": 1534
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0869178896847008e-05,
+      "loss": 1.4795,
+      "step": 1535
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0858474897810795e-05,
+      "loss": 1.4583,
+      "step": 1536
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0847769907777066e-05,
+      "loss": 1.7205,
+      "step": 1537
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0837063939103332e-05,
+      "loss": 1.6122,
+      "step": 1538
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0826357004148243e-05,
+      "loss": 1.4572,
+      "step": 1539
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0815649115271561e-05,
+      "loss": 1.4342,
+      "step": 1540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0804940284834151e-05,
+      "loss": 1.3951,
+      "step": 1541
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0794230525197959e-05,
+      "loss": 1.8243,
+      "step": 1542
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0783519848726017e-05,
+      "loss": 1.4073,
+      "step": 1543
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0772808267782397e-05,
+      "loss": 1.5324,
+      "step": 1544
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0762095794732235e-05,
+      "loss": 1.4756,
+      "step": 1545
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0751382441941677e-05,
+      "loss": 1.7171,
+      "step": 1546
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.07406682217779e-05,
+      "loss": 1.495,
+      "step": 1547
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0729953146609076e-05,
+      "loss": 1.6261,
+      "step": 1548
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0719237228804365e-05,
+      "loss": 1.4229,
+      "step": 1549
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0708520480733895e-05,
+      "loss": 1.4619,
+      "step": 1550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.069780291476876e-05,
+      "loss": 1.4653,
+      "step": 1551
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0687084543280997e-05,
+      "loss": 1.3248,
+      "step": 1552
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0676365378643566e-05,
+      "loss": 1.445,
+      "step": 1553
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0665645433230345e-05,
+      "loss": 1.294,
+      "step": 1554
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0654924719416127e-05,
+      "loss": 1.5404,
+      "step": 1555
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0644203249576566e-05,
+      "loss": 1.641,
+      "step": 1556
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0633481036088215e-05,
+      "loss": 1.464,
+      "step": 1557
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0622758091328469e-05,
+      "loss": 1.524,
+      "step": 1558
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0612034427675569e-05,
+      "loss": 1.4601,
+      "step": 1559
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0601310057508596e-05,
+      "loss": 1.6066,
+      "step": 1560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0590584993207435e-05,
+      "loss": 1.6947,
+      "step": 1561
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0579859247152774e-05,
+      "loss": 1.4894,
+      "step": 1562
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0569132831726094e-05,
+      "loss": 1.2765,
+      "step": 1563
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0558405759309647e-05,
+      "loss": 1.6583,
+      "step": 1564
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0547678042286435e-05,
+      "loss": 1.5172,
+      "step": 1565
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0536949693040224e-05,
+      "loss": 1.3617,
+      "step": 1566
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0526220723955483e-05,
+      "loss": 1.4889,
+      "step": 1567
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0515491147417414e-05,
+      "loss": 1.4693,
+      "step": 1568
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0504760975811919e-05,
+      "loss": 1.5143,
+      "step": 1569
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0494030221525582e-05,
+      "loss": 1.3994,
+      "step": 1570
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0483298896945666e-05,
+      "loss": 1.2161,
+      "step": 1571
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0472567014460084e-05,
+      "loss": 1.3966,
+      "step": 1572
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0461834586457398e-05,
+      "loss": 1.4431,
+      "step": 1573
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0451101625326798e-05,
+      "loss": 1.4509,
+      "step": 1574
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0440368143458094e-05,
+      "loss": 1.5882,
+      "step": 1575
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0429634153241696e-05,
+      "loss": 1.3833,
+      "step": 1576
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0418899667068589e-05,
+      "loss": 1.6161,
+      "step": 1577
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0408164697330348e-05,
+      "loss": 1.5876,
+      "step": 1578
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0397429256419097e-05,
+      "loss": 1.5011,
+      "step": 1579
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0386693356727498e-05,
+      "loss": 1.5359,
+      "step": 1580
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0375957010648759e-05,
+      "loss": 1.6699,
+      "step": 1581
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0365220230576592e-05,
+      "loss": 1.4582,
+      "step": 1582
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0354483028905205e-05,
+      "loss": 1.3734,
+      "step": 1583
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.034374541802931e-05,
+      "loss": 1.4869,
+      "step": 1584
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0333007410344073e-05,
+      "loss": 1.5385,
+      "step": 1585
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0322269018245128e-05,
+      "loss": 1.7099,
+      "step": 1586
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0311530254128554e-05,
+      "loss": 1.244,
+      "step": 1587
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0300791130390853e-05,
+      "loss": 1.2162,
+      "step": 1588
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0290051659428949e-05,
+      "loss": 1.3823,
+      "step": 1589
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0279311853640157e-05,
+      "loss": 1.6968,
+      "step": 1590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0268571725422192e-05,
+      "loss": 1.5301,
+      "step": 1591
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0257831287173128e-05,
+      "loss": 1.6138,
+      "step": 1592
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0247090551291405e-05,
+      "loss": 1.5803,
+      "step": 1593
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0236349530175807e-05,
+      "loss": 1.5447,
+      "step": 1594
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.022560823622544e-05,
+      "loss": 1.5393,
+      "step": 1595
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0214866681839735e-05,
+      "loss": 1.4851,
+      "step": 1596
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0204124879418414e-05,
+      "loss": 1.3898,
+      "step": 1597
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.019338284136149e-05,
+      "loss": 1.5791,
+      "step": 1598
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0182640580069249e-05,
+      "loss": 1.4625,
+      "step": 1599
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0171898107942233e-05,
+      "loss": 1.6567,
+      "step": 1600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.016115543738123e-05,
+      "loss": 1.5293,
+      "step": 1601
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.015041258078725e-05,
+      "loss": 1.218,
+      "step": 1602
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0139669550561526e-05,
+      "loss": 1.4975,
+      "step": 1603
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0128926359105486e-05,
+      "loss": 1.7081,
+      "step": 1604
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0118183018820748e-05,
+      "loss": 1.5369,
+      "step": 1605
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0107439542109097e-05,
+      "loss": 1.4781,
+      "step": 1606
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0096695941372481e-05,
+      "loss": 1.3985,
+      "step": 1607
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0085952229012993e-05,
+      "loss": 1.4682,
+      "step": 1608
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0075208417432837e-05,
+      "loss": 1.3249,
+      "step": 1609
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0064464519034358e-05,
+      "loss": 1.4237,
+      "step": 1610
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0053720546219984e-05,
+      "loss": 1.5491,
+      "step": 1611
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0042976511392226e-05,
+      "loss": 1.4904,
+      "step": 1612
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0032232426953685e-05,
+      "loss": 1.5166,
+      "step": 1613
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0021488305307003e-05,
+      "loss": 1.4938,
+      "step": 1614
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0010744158854869e-05,
+      "loss": 1.5615,
+      "step": 1615
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4769,
+      "step": 1616
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.989255841145134e-06,
+      "loss": 1.6829,
+      "step": 1617
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.978511694692999e-06,
+      "loss": 1.6776,
+      "step": 1618
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.967767573046318e-06,
+      "loss": 1.4886,
+      "step": 1619
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.957023488607775e-06,
+      "loss": 1.4845,
+      "step": 1620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.946279453780021e-06,
+      "loss": 1.7475,
+      "step": 1621
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.935535480965647e-06,
+      "loss": 1.6902,
+      "step": 1622
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.924791582567164e-06,
+      "loss": 1.3427,
+      "step": 1623
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.914047770987012e-06,
+      "loss": 1.5427,
+      "step": 1624
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.903304058627519e-06,
+      "loss": 1.5205,
+      "step": 1625
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.892560457890907e-06,
+      "loss": 1.4769,
+      "step": 1626
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.881816981179256e-06,
+      "loss": 1.6765,
+      "step": 1627
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.871073640894515e-06,
+      "loss": 1.6277,
+      "step": 1628
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.860330449438479e-06,
+      "loss": 1.5609,
+      "step": 1629
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.849587419212751e-06,
+      "loss": 1.5834,
+      "step": 1630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.838844562618774e-06,
+      "loss": 1.4719,
+      "step": 1631
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.828101892057767e-06,
+      "loss": 1.6199,
+      "step": 1632
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.817359419930753e-06,
+      "loss": 1.6725,
+      "step": 1633
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.806617158638515e-06,
+      "loss": 1.3306,
+      "step": 1634
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.79587512058159e-06,
+      "loss": 1.5501,
+      "step": 1635
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.785133318160268e-06,
+      "loss": 1.4852,
+      "step": 1636
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.77439176377456e-06,
+      "loss": 1.8348,
+      "step": 1637
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.763650469824198e-06,
+      "loss": 1.4346,
+      "step": 1638
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.752909448708598e-06,
+      "loss": 1.555,
+      "step": 1639
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.742168712826876e-06,
+      "loss": 1.4869,
+      "step": 1640
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.731428274577812e-06,
+      "loss": 1.8488,
+      "step": 1641
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.720688146359843e-06,
+      "loss": 1.5448,
+      "step": 1642
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.709948340571056e-06,
+      "loss": 1.5486,
+      "step": 1643
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.69920886960915e-06,
+      "loss": 1.6329,
+      "step": 1644
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.688469745871447e-06,
+      "loss": 1.4295,
+      "step": 1645
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.677730981754875e-06,
+      "loss": 1.3833,
+      "step": 1646
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.66699258965593e-06,
+      "loss": 1.418,
+      "step": 1647
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.656254581970693e-06,
+      "loss": 1.4126,
+      "step": 1648
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.6455169710948e-06,
+      "loss": 1.3916,
+      "step": 1649
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.634779769423412e-06,
+      "loss": 1.7731,
+      "step": 1650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.624042989351245e-06,
+      "loss": 1.5519,
+      "step": 1651
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.613306643272504e-06,
+      "loss": 1.3099,
+      "step": 1652
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.602570743580908e-06,
+      "loss": 1.4315,
+      "step": 1653
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.591835302669657e-06,
+      "loss": 1.2628,
+      "step": 1654
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.581100332931413e-06,
+      "loss": 1.4136,
+      "step": 1655
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.570365846758309e-06,
+      "loss": 1.3965,
+      "step": 1656
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.559631856541906e-06,
+      "loss": 1.5249,
+      "step": 1657
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.548898374673205e-06,
+      "loss": 1.5252,
+      "step": 1658
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.538165413542607e-06,
+      "loss": 1.2183,
+      "step": 1659
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.52743298553992e-06,
+      "loss": 1.2063,
+      "step": 1660
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.51670110305434e-06,
+      "loss": 1.4271,
+      "step": 1661
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.505969778474418e-06,
+      "loss": 1.372,
+      "step": 1662
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.495239024188085e-06,
+      "loss": 1.4457,
+      "step": 1663
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.484508852582587e-06,
+      "loss": 1.4709,
+      "step": 1664
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.47377927604452e-06,
+      "loss": 1.4046,
+      "step": 1665
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.463050306959782e-06,
+      "loss": 1.5086,
+      "step": 1666
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.452321957713563e-06,
+      "loss": 1.6328,
+      "step": 1667
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.441594240690356e-06,
+      "loss": 1.3144,
+      "step": 1668
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.430867168273906e-06,
+      "loss": 1.7755,
+      "step": 1669
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.42014075284723e-06,
+      "loss": 1.5649,
+      "step": 1670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.40941500679257e-06,
+      "loss": 1.6632,
+      "step": 1671
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.398689942491406e-06,
+      "loss": 1.401,
+      "step": 1672
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.387965572324433e-06,
+      "loss": 1.5523,
+      "step": 1673
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.377241908671533e-06,
+      "loss": 1.4811,
+      "step": 1674
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.366518963911788e-06,
+      "loss": 1.564,
+      "step": 1675
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.355796750423436e-06,
+      "loss": 1.423,
+      "step": 1676
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.345075280583877e-06,
+      "loss": 1.5388,
+      "step": 1677
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.334354566769658e-06,
+      "loss": 1.4689,
+      "step": 1678
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.32363462135644e-06,
+      "loss": 1.4237,
+      "step": 1679
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.312915456719008e-06,
+      "loss": 1.5297,
+      "step": 1680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.302197085231244e-06,
+      "loss": 1.508,
+      "step": 1681
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.291479519266108e-06,
+      "loss": 1.3045,
+      "step": 1682
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.28076277119564e-06,
+      "loss": 1.5766,
+      "step": 1683
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.270046853390924e-06,
+      "loss": 1.457,
+      "step": 1684
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.2593317782221e-06,
+      "loss": 1.5868,
+      "step": 1685
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.248617558058328e-06,
+      "loss": 1.3938,
+      "step": 1686
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.23790420526777e-06,
+      "loss": 1.5023,
+      "step": 1687
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.227191732217605e-06,
+      "loss": 1.1972,
+      "step": 1688
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.216480151273985e-06,
+      "loss": 1.4058,
+      "step": 1689
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.205769474802045e-06,
+      "loss": 1.3072,
+      "step": 1690
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.195059715165856e-06,
+      "loss": 1.3134,
+      "step": 1691
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.18435088472844e-06,
+      "loss": 1.5893,
+      "step": 1692
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.173642995851762e-06,
+      "loss": 1.3873,
+      "step": 1693
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.162936060896672e-06,
+      "loss": 1.507,
+      "step": 1694
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.152230092222937e-06,
+      "loss": 1.5432,
+      "step": 1695
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.141525102189205e-06,
+      "loss": 1.4368,
+      "step": 1696
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.130821103152995e-06,
+      "loss": 1.484,
+      "step": 1697
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.12011810747068e-06,
+      "loss": 1.4962,
+      "step": 1698
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.10941612749747e-06,
+      "loss": 1.7323,
+      "step": 1699
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.098715175587415e-06,
+      "loss": 1.5804,
+      "step": 1700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.088015264093365e-06,
+      "loss": 1.5823,
+      "step": 1701
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.07731640536698e-06,
+      "loss": 1.5142,
+      "step": 1702
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.066618611758699e-06,
+      "loss": 1.6676,
+      "step": 1703
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.055921895617727e-06,
+      "loss": 1.6122,
+      "step": 1704
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.045226269292039e-06,
+      "loss": 1.7893,
+      "step": 1705
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.034531745128334e-06,
+      "loss": 1.4237,
+      "step": 1706
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.023838335472055e-06,
+      "loss": 1.479,
+      "step": 1707
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.013146052667346e-06,
+      "loss": 1.5775,
+      "step": 1708
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.002454909057058e-06,
+      "loss": 1.5629,
+      "step": 1709
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.991764916982731e-06,
+      "loss": 1.7483,
+      "step": 1710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.981076088784557e-06,
+      "loss": 1.3477,
+      "step": 1711
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.970388436801408e-06,
+      "loss": 1.7882,
+      "step": 1712
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.959701973370784e-06,
+      "loss": 1.4399,
+      "step": 1713
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.949016710828808e-06,
+      "loss": 1.5013,
+      "step": 1714
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.938332661510234e-06,
+      "loss": 1.4854,
+      "step": 1715
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.9276498377484e-06,
+      "loss": 1.4276,
+      "step": 1716
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.916968251875244e-06,
+      "loss": 1.3301,
+      "step": 1717
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.906287916221259e-06,
+      "loss": 1.6421,
+      "step": 1718
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.895608843115501e-06,
+      "loss": 1.51,
+      "step": 1719
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.884931044885577e-06,
+      "loss": 1.8015,
+      "step": 1720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.874254533857607e-06,
+      "loss": 1.4523,
+      "step": 1721
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.863579322356242e-06,
+      "loss": 1.5051,
+      "step": 1722
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.852905422704616e-06,
+      "loss": 1.5617,
+      "step": 1723
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.84223284722436e-06,
+      "loss": 1.5491,
+      "step": 1724
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.831561608235578e-06,
+      "loss": 1.5579,
+      "step": 1725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.820891718056815e-06,
+      "loss": 1.4524,
+      "step": 1726
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.810223189005079e-06,
+      "loss": 1.7026,
+      "step": 1727
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.799556033395794e-06,
+      "loss": 1.346,
+      "step": 1728
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.78889026354281e-06,
+      "loss": 1.6576,
+      "step": 1729
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.77822589175836e-06,
+      "loss": 1.7151,
+      "step": 1730
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.767562930353078e-06,
+      "loss": 1.4132,
+      "step": 1731
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.756901391635963e-06,
+      "loss": 1.2924,
+      "step": 1732
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.746241287914376e-06,
+      "loss": 1.5416,
+      "step": 1733
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.73558263149402e-06,
+      "loss": 1.5723,
+      "step": 1734
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.724925434678923e-06,
+      "loss": 1.585,
+      "step": 1735
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.714269709771432e-06,
+      "loss": 1.287,
+      "step": 1736
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.703615469072198e-06,
+      "loss": 1.332,
+      "step": 1737
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.692962724880148e-06,
+      "loss": 1.2451,
+      "step": 1738
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.682311489492495e-06,
+      "loss": 1.5102,
+      "step": 1739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.671661775204703e-06,
+      "loss": 1.596,
+      "step": 1740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.661013594310474e-06,
+      "loss": 1.4446,
+      "step": 1741
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.650366959101757e-06,
+      "loss": 1.3595,
+      "step": 1742
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.639721881868697e-06,
+      "loss": 1.6099,
+      "step": 1743
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.629078374899657e-06,
+      "loss": 1.5575,
+      "step": 1744
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.618436450481182e-06,
+      "loss": 1.4255,
+      "step": 1745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.607796120897978e-06,
+      "loss": 1.5131,
+      "step": 1746
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.597157398432928e-06,
+      "loss": 1.4646,
+      "step": 1747
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.58652029536705e-06,
+      "loss": 1.4404,
+      "step": 1748
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.5758848239795e-06,
+      "loss": 1.2982,
+      "step": 1749
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.565250996547538e-06,
+      "loss": 1.5175,
+      "step": 1750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.554618825346534e-06,
+      "loss": 1.1866,
+      "step": 1751
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.543988322649954e-06,
+      "loss": 1.517,
+      "step": 1752
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.533359500729317e-06,
+      "loss": 1.8491,
+      "step": 1753
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.522732371854228e-06,
+      "loss": 1.438,
+      "step": 1754
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.512106948292313e-06,
+      "loss": 1.4073,
+      "step": 1755
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.501483242309243e-06,
+      "loss": 1.3171,
+      "step": 1756
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.490861266168711e-06,
+      "loss": 1.6529,
+      "step": 1757
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.480241032132394e-06,
+      "loss": 1.5813,
+      "step": 1758
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.469622552459977e-06,
+      "loss": 1.4463,
+      "step": 1759
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.459005839409109e-06,
+      "loss": 1.353,
+      "step": 1760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.448390905235411e-06,
+      "loss": 1.3517,
+      "step": 1761
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.437777762192434e-06,
+      "loss": 1.6284,
+      "step": 1762
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.427166422531672e-06,
+      "loss": 1.329,
+      "step": 1763
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.41655689850254e-06,
+      "loss": 1.2541,
+      "step": 1764
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.405949202352342e-06,
+      "loss": 1.4916,
+      "step": 1765
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.395343346326295e-06,
+      "loss": 1.6485,
+      "step": 1766
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.384739342667473e-06,
+      "loss": 1.5962,
+      "step": 1767
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.374137203616817e-06,
+      "loss": 1.7891,
+      "step": 1768
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.363536941413121e-06,
+      "loss": 1.169,
+      "step": 1769
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.352938568293e-06,
+      "loss": 1.5405,
+      "step": 1770
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.342342096490903e-06,
+      "loss": 1.2313,
+      "step": 1771
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.331747538239077e-06,
+      "loss": 1.5019,
+      "step": 1772
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.321154905767555e-06,
+      "loss": 1.328,
+      "step": 1773
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.310564211304159e-06,
+      "loss": 1.3569,
+      "step": 1774
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.29997546707446e-06,
+      "loss": 1.6528,
+      "step": 1775
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.289388685301792e-06,
+      "loss": 1.8018,
+      "step": 1776
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.278803878207215e-06,
+      "loss": 1.4207,
+      "step": 1777
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.268221058009506e-06,
+      "loss": 1.7549,
+      "step": 1778
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.257640236925158e-06,
+      "loss": 1.6103,
+      "step": 1779
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.247061427168349e-06,
+      "loss": 1.6564,
+      "step": 1780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.236484640950942e-06,
+      "loss": 1.4276,
+      "step": 1781
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.225909890482456e-06,
+      "loss": 1.395,
+      "step": 1782
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.21533718797006e-06,
+      "loss": 1.5064,
+      "step": 1783
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.204766545618574e-06,
+      "loss": 1.556,
+      "step": 1784
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.194197975630418e-06,
+      "loss": 1.6437,
+      "step": 1785
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.183631490205636e-06,
+      "loss": 1.3784,
+      "step": 1786
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.173067101541862e-06,
+      "loss": 1.4209,
+      "step": 1787
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.162504821834296e-06,
+      "loss": 1.4551,
+      "step": 1788
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.15194466327573e-06,
+      "loss": 1.2532,
+      "step": 1789
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.141386638056482e-06,
+      "loss": 1.5615,
+      "step": 1790
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.130830758364422e-06,
+      "loss": 1.3597,
+      "step": 1791
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.120277036384937e-06,
+      "loss": 1.8915,
+      "step": 1792
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.10972548430093e-06,
+      "loss": 1.2352,
+      "step": 1793
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.09917611429279e-06,
+      "loss": 1.4329,
+      "step": 1794
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.088628938538388e-06,
+      "loss": 1.4839,
+      "step": 1795
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.078083969213076e-06,
+      "loss": 1.5242,
+      "step": 1796
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.067541218489636e-06,
+      "loss": 1.5399,
+      "step": 1797
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.057000698538311e-06,
+      "loss": 1.4487,
+      "step": 1798
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.046462421526757e-06,
+      "loss": 1.2873,
+      "step": 1799
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.035926399620036e-06,
+      "loss": 1.5895,
+      "step": 1800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.025392644980625e-06,
+      "loss": 1.663,
+      "step": 1801
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.014861169768362e-06,
+      "loss": 1.4611,
+      "step": 1802
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.004331986140474e-06,
+      "loss": 1.4775,
+      "step": 1803
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.993805106251528e-06,
+      "loss": 1.7779,
+      "step": 1804
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.983280542253434e-06,
+      "loss": 1.5543,
+      "step": 1805
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.972758306295436e-06,
+      "loss": 1.4543,
+      "step": 1806
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.96223841052408e-06,
+      "loss": 1.4499,
+      "step": 1807
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.951720867083225e-06,
+      "loss": 1.4907,
+      "step": 1808
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.941205688114e-06,
+      "loss": 1.4289,
+      "step": 1809
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.930692885754806e-06,
+      "loss": 1.6214,
+      "step": 1810
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.920182472141313e-06,
+      "loss": 1.3517,
+      "step": 1811
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.90967445940642e-06,
+      "loss": 1.4579,
+      "step": 1812
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.899168859680267e-06,
+      "loss": 1.3465,
+      "step": 1813
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.888665685090194e-06,
+      "loss": 1.4627,
+      "step": 1814
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.878164947760748e-06,
+      "loss": 1.4665,
+      "step": 1815
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.867666659813671e-06,
+      "loss": 1.614,
+      "step": 1816
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.857170833367862e-06,
+      "loss": 1.8233,
+      "step": 1817
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.846677480539392e-06,
+      "loss": 1.6569,
+      "step": 1818
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.83618661344147e-06,
+      "loss": 1.5918,
+      "step": 1819
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.825698244184432e-06,
+      "loss": 1.8777,
+      "step": 1820
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.81521238487574e-06,
+      "loss": 1.5248,
+      "step": 1821
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.80472904761995e-06,
+      "loss": 1.7925,
+      "step": 1822
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.794248244518714e-06,
+      "loss": 1.7008,
+      "step": 1823
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.783769987670749e-06,
+      "loss": 1.4547,
+      "step": 1824
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.773294289171848e-06,
+      "loss": 1.3195,
+      "step": 1825
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.762821161114834e-06,
+      "loss": 1.6725,
+      "step": 1826
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.752350615589569e-06,
+      "loss": 1.6621,
+      "step": 1827
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.74188266468294e-06,
+      "loss": 1.3337,
+      "step": 1828
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.73141732047883e-06,
+      "loss": 1.6276,
+      "step": 1829
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.720954595058118e-06,
+      "loss": 1.3571,
+      "step": 1830
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.710494500498662e-06,
+      "loss": 1.6328,
+      "step": 1831
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.700037048875272e-06,
+      "loss": 1.4859,
+      "step": 1832
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.68958225225972e-06,
+      "loss": 1.6407,
+      "step": 1833
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.679130122720704e-06,
+      "loss": 1.4501,
+      "step": 1834
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.668680672323852e-06,
+      "loss": 1.5637,
+      "step": 1835
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.658233913131691e-06,
+      "loss": 1.5341,
+      "step": 1836
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.647789857203644e-06,
+      "loss": 1.5118,
+      "step": 1837
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.637348516596016e-06,
+      "loss": 1.4645,
+      "step": 1838
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.626909903361974e-06,
+      "loss": 1.5651,
+      "step": 1839
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.616474029551544e-06,
+      "loss": 1.4499,
+      "step": 1840
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.606040907211575e-06,
+      "loss": 1.7073,
+      "step": 1841
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.59561054838575e-06,
+      "loss": 1.7412,
+      "step": 1842
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.585182965114568e-06,
+      "loss": 1.4417,
+      "step": 1843
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5747581694353056e-06,
+      "loss": 1.5282,
+      "step": 1844
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5643361733820444e-06,
+      "loss": 1.1414,
+      "step": 1845
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5539169889856135e-06,
+      "loss": 1.5679,
+      "step": 1846
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.543500628273605e-06,
+      "loss": 1.5666,
+      "step": 1847
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5330871032703605e-06,
+      "loss": 1.5347,
+      "step": 1848
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.522676425996927e-06,
+      "loss": 1.7683,
+      "step": 1849
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.512268608471083e-06,
+      "loss": 1.4895,
+      "step": 1850
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5018636627073e-06,
+      "loss": 1.4519,
+      "step": 1851
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.491461600716729e-06,
+      "loss": 1.6018,
+      "step": 1852
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.481062434507199e-06,
+      "loss": 1.5031,
+      "step": 1853
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.470666176083193e-06,
+      "loss": 1.5826,
+      "step": 1854
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.460272837445841e-06,
+      "loss": 1.7268,
+      "step": 1855
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.449882430592893e-06,
+      "loss": 1.6162,
+      "step": 1856
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.43949496751873e-06,
+      "loss": 1.2963,
+      "step": 1857
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.42911046021432e-06,
+      "loss": 1.5238,
+      "step": 1858
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.4187289206672245e-06,
+      "loss": 1.6771,
+      "step": 1859
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.4083503608615846e-06,
+      "loss": 1.4157,
+      "step": 1860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.397974792778089e-06,
+      "loss": 1.3462,
+      "step": 1861
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.387602228393987e-06,
+      "loss": 1.5207,
+      "step": 1862
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.377232679683053e-06,
+      "loss": 1.504,
+      "step": 1863
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.366866158615575e-06,
+      "loss": 1.4711,
+      "step": 1864
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.356502677158361e-06,
+      "loss": 1.4349,
+      "step": 1865
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.346142247274695e-06,
+      "loss": 1.4745,
+      "step": 1866
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.3357848809243505e-06,
+      "loss": 1.4827,
+      "step": 1867
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.3254305900635525e-06,
+      "loss": 1.5509,
+      "step": 1868
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.3150793866449834e-06,
+      "loss": 1.3821,
+      "step": 1869
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.304731282617762e-06,
+      "loss": 1.624,
+      "step": 1870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.294386289927425e-06,
+      "loss": 1.5397,
+      "step": 1871
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.284044420515927e-06,
+      "loss": 1.5431,
+      "step": 1872
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.2737056863216015e-06,
+      "loss": 1.5223,
+      "step": 1873
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.263370099279173e-06,
+      "loss": 1.4568,
+      "step": 1874
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.253037671319734e-06,
+      "loss": 1.5948,
+      "step": 1875
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.242708414370725e-06,
+      "loss": 1.5103,
+      "step": 1876
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.23238234035593e-06,
+      "loss": 1.5594,
+      "step": 1877
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.2220594611954606e-06,
+      "loss": 1.5512,
+      "step": 1878
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.211739788805727e-06,
+      "loss": 1.3181,
+      "step": 1879
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.20142333509946e-06,
+      "loss": 1.4319,
+      "step": 1880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.191110111985653e-06,
+      "loss": 1.5361,
+      "step": 1881
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.1808001313695855e-06,
+      "loss": 1.4435,
+      "step": 1882
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.170493405152788e-06,
+      "loss": 1.5366,
+      "step": 1883
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.160189945233027e-06,
+      "loss": 1.4455,
+      "step": 1884
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.149889763504312e-06,
+      "loss": 1.466,
+      "step": 1885
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.1395928718568605e-06,
+      "loss": 1.8937,
+      "step": 1886
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.129299282177098e-06,
+      "loss": 1.3026,
+      "step": 1887
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.119009006347625e-06,
+      "loss": 1.4058,
+      "step": 1888
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.108722056247233e-06,
+      "loss": 1.4823,
+      "step": 1889
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.098438443750865e-06,
+      "loss": 1.6478,
+      "step": 1890
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.088158180729606e-06,
+      "loss": 1.2193,
+      "step": 1891
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.077881279050693e-06,
+      "loss": 1.5232,
+      "step": 1892
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.067607750577463e-06,
+      "loss": 1.4289,
+      "step": 1893
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.057337607169373e-06,
+      "loss": 1.3342,
+      "step": 1894
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.047070860681964e-06,
+      "loss": 1.3467,
+      "step": 1895
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.0368075229668535e-06,
+      "loss": 1.6121,
+      "step": 1896
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.026547605871736e-06,
+      "loss": 1.5548,
+      "step": 1897
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.016291121240346e-06,
+      "loss": 1.3587,
+      "step": 1898
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.006038080912465e-06,
+      "loss": 1.5816,
+      "step": 1899
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.9957884967238885e-06,
+      "loss": 1.3931,
+      "step": 1900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.985542380506427e-06,
+      "loss": 1.6788,
+      "step": 1901
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.975299744087891e-06,
+      "loss": 1.5523,
+      "step": 1902
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.9650605992920675e-06,
+      "loss": 1.3829,
+      "step": 1903
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.954824957938723e-06,
+      "loss": 1.5935,
+      "step": 1904
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.944592831843566e-06,
+      "loss": 1.3489,
+      "step": 1905
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.934364232818254e-06,
+      "loss": 1.5336,
+      "step": 1906
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.924139172670381e-06,
+      "loss": 1.4969,
+      "step": 1907
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.913917663203438e-06,
+      "loss": 1.4919,
+      "step": 1908
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.903699716216833e-06,
+      "loss": 1.4839,
+      "step": 1909
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.8934853435058566e-06,
+      "loss": 1.5603,
+      "step": 1910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.883274556861667e-06,
+      "loss": 1.253,
+      "step": 1911
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.873067368071294e-06,
+      "loss": 1.6143,
+      "step": 1912
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.862863788917602e-06,
+      "loss": 1.5947,
+      "step": 1913
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.852663831179303e-06,
+      "loss": 1.3282,
+      "step": 1914
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.8424675066309165e-06,
+      "loss": 1.7235,
+      "step": 1915
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.8322748270427684e-06,
+      "loss": 1.2116,
+      "step": 1916
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.822085804180985e-06,
+      "loss": 1.7103,
+      "step": 1917
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.811900449807465e-06,
+      "loss": 1.5345,
+      "step": 1918
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.801718775679877e-06,
+      "loss": 1.2341,
+      "step": 1919
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.791540793551635e-06,
+      "loss": 1.777,
+      "step": 1920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.781366515171894e-06,
+      "loss": 1.3128,
+      "step": 1921
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.771195952285541e-06,
+      "loss": 1.201,
+      "step": 1922
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.761029116633156e-06,
+      "loss": 1.343,
+      "step": 1923
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.7508660199510356e-06,
+      "loss": 1.3073,
+      "step": 1924
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.740706673971147e-06,
+      "loss": 1.2528,
+      "step": 1925
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.730551090421137e-06,
+      "loss": 1.8275,
+      "step": 1926
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.720399281024306e-06,
+      "loss": 1.7886,
+      "step": 1927
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.710251257499587e-06,
+      "loss": 1.699,
+      "step": 1928
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.700107031561563e-06,
+      "loss": 1.4264,
+      "step": 1929
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.689966614920414e-06,
+      "loss": 1.4701,
+      "step": 1930
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.679830019281939e-06,
+      "loss": 1.6261,
+      "step": 1931
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.6696972563475114e-06,
+      "loss": 1.9164,
+      "step": 1932
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.659568337814089e-06,
+      "loss": 1.5896,
+      "step": 1933
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.6494432753741935e-06,
+      "loss": 1.1571,
+      "step": 1934
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.639322080715884e-06,
+      "loss": 1.5337,
+      "step": 1935
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.6292047655227695e-06,
+      "loss": 1.4426,
+      "step": 1936
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.619091341473971e-06,
+      "loss": 1.5778,
+      "step": 1937
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.608981820244116e-06,
+      "loss": 1.4254,
+      "step": 1938
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.5988762135033405e-06,
+      "loss": 1.3787,
+      "step": 1939
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.588774532917239e-06,
+      "loss": 1.3265,
+      "step": 1940
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.5786767901468985e-06,
+      "loss": 1.5091,
+      "step": 1941
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.568582996848844e-06,
+      "loss": 1.3768,
+      "step": 1942
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.558493164675041e-06,
+      "loss": 1.7122,
+      "step": 1943
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.548407305272896e-06,
+      "loss": 1.2422,
+      "step": 1944
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.538325430285213e-06,
+      "loss": 1.4983,
+      "step": 1945
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.528247551350213e-06,
+      "loss": 1.4401,
+      "step": 1946
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.51817368010149e-06,
+      "loss": 1.29,
+      "step": 1947
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.508103828168013e-06,
+      "loss": 1.3499,
+      "step": 1948
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.498038007174123e-06,
+      "loss": 1.6459,
+      "step": 1949
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.487976228739493e-06,
+      "loss": 1.4029,
+      "step": 1950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.477918504479146e-06,
+      "loss": 1.4964,
+      "step": 1951
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.467864846003404e-06,
+      "loss": 1.4241,
+      "step": 1952
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.4578152649179175e-06,
+      "loss": 1.5626,
+      "step": 1953
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.4477697728236146e-06,
+      "loss": 1.4762,
+      "step": 1954
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.437728381316705e-06,
+      "loss": 1.6353,
+      "step": 1955
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.427691101988673e-06,
+      "loss": 1.5046,
+      "step": 1956
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.417657946426249e-06,
+      "loss": 1.4596,
+      "step": 1957
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.407628926211409e-06,
+      "loss": 1.5093,
+      "step": 1958
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.397604052921346e-06,
+      "loss": 1.4112,
+      "step": 1959
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.387583338128471e-06,
+      "loss": 1.42,
+      "step": 1960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.377566793400399e-06,
+      "loss": 1.3564,
+      "step": 1961
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.367554430299924e-06,
+      "loss": 1.4984,
+      "step": 1962
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.35754626038502e-06,
+      "loss": 1.3189,
+      "step": 1963
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.347542295208813e-06,
+      "loss": 1.248,
+      "step": 1964
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.337542546319575e-06,
+      "loss": 1.6866,
+      "step": 1965
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.327547025260723e-06,
+      "loss": 1.4282,
+      "step": 1966
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.317555743570776e-06,
+      "loss": 1.3998,
+      "step": 1967
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.3075687127833754e-06,
+      "loss": 1.2947,
+      "step": 1968
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.297585944427247e-06,
+      "loss": 1.583,
+      "step": 1969
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.287607450026189e-06,
+      "loss": 1.536,
+      "step": 1970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.2776332410990855e-06,
+      "loss": 1.2205,
+      "step": 1971
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.267663329159857e-06,
+      "loss": 1.2003,
+      "step": 1972
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.257697725717469e-06,
+      "loss": 1.6093,
+      "step": 1973
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.247736442275918e-06,
+      "loss": 1.5194,
+      "step": 1974
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.237779490334199e-06,
+      "loss": 1.3725,
+      "step": 1975
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.227826881386328e-06,
+      "loss": 1.7043,
+      "step": 1976
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.217878626921288e-06,
+      "loss": 1.3394,
+      "step": 1977
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.2079347384230505e-06,
+      "loss": 1.5908,
+      "step": 1978
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.197995227370537e-06,
+      "loss": 1.3906,
+      "step": 1979
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.188060105237617e-06,
+      "loss": 1.3093,
+      "step": 1980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.178129383493101e-06,
+      "loss": 1.649,
+      "step": 1981
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.168203073600706e-06,
+      "loss": 1.3755,
+      "step": 1982
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.158281187019077e-06,
+      "loss": 1.5392,
+      "step": 1983
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.14836373520173e-06,
+      "loss": 1.435,
+      "step": 1984
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.138450729597078e-06,
+      "loss": 1.3713,
+      "step": 1985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.128542181648395e-06,
+      "loss": 1.6633,
+      "step": 1986
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.118638102793805e-06,
+      "loss": 1.4763,
+      "step": 1987
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.108738504466283e-06,
+      "loss": 1.3324,
+      "step": 1988
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.098843398093621e-06,
+      "loss": 1.2667,
+      "step": 1989
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.088952795098442e-06,
+      "loss": 1.4732,
+      "step": 1990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.079066706898149e-06,
+      "loss": 1.3757,
+      "step": 1991
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.069185144904945e-06,
+      "loss": 1.3396,
+      "step": 1992
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.059308120525816e-06,
+      "loss": 1.632,
+      "step": 1993
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.049435645162487e-06,
+      "loss": 1.3935,
+      "step": 1994
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.0395677302114594e-06,
+      "loss": 1.4107,
+      "step": 1995
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.029704387063949e-06,
+      "loss": 1.7625,
+      "step": 1996
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.0198456271059e-06,
+      "loss": 1.5096,
+      "step": 1997
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.009991461717977e-06,
+      "loss": 1.4553,
+      "step": 1998
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.000141902275518e-06,
+      "loss": 1.5631,
+      "step": 1999
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.990296960148566e-06,
+      "loss": 1.3888,
+      "step": 2000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.9804566467018255e-06,
+      "loss": 1.4644,
+      "step": 2001
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.9706209732946495e-06,
+      "loss": 1.4841,
+      "step": 2002
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.960789951281052e-06,
+      "loss": 1.416,
+      "step": 2003
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.950963592009662e-06,
+      "loss": 1.3853,
+      "step": 2004
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.941141906823735e-06,
+      "loss": 1.6921,
+      "step": 2005
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.931324907061131e-06,
+      "loss": 1.4963,
+      "step": 2006
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.921512604054289e-06,
+      "loss": 1.5602,
+      "step": 2007
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.911705009130245e-06,
+      "loss": 1.2594,
+      "step": 2008
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.901902133610584e-06,
+      "loss": 1.4997,
+      "step": 2009
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.892103988811457e-06,
+      "loss": 1.5552,
+      "step": 2010
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.8823105860435405e-06,
+      "loss": 1.3855,
+      "step": 2011
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.872521936612043e-06,
+      "loss": 1.4072,
+      "step": 2012
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.862738051816692e-06,
+      "loss": 1.4106,
+      "step": 2013
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.852958942951701e-06,
+      "loss": 1.6015,
+      "step": 2014
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.843184621305784e-06,
+      "loss": 1.6864,
+      "step": 2015
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.833415098162121e-06,
+      "loss": 1.2497,
+      "step": 2016
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.823650384798354e-06,
+      "loss": 1.4745,
+      "step": 2017
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.8138904924865766e-06,
+      "loss": 1.6512,
+      "step": 2018
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.804135432493312e-06,
+      "loss": 1.3009,
+      "step": 2019
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.794385216079506e-06,
+      "loss": 1.3523,
+      "step": 2020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.784639854500516e-06,
+      "loss": 1.4951,
+      "step": 2021
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.774899359006092e-06,
+      "loss": 1.6757,
+      "step": 2022
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.765163740840369e-06,
+      "loss": 1.5397,
+      "step": 2023
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.755433011241851e-06,
+      "loss": 1.6306,
+      "step": 2024
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.7457071814433985e-06,
+      "loss": 1.6603,
+      "step": 2025
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.735986262672211e-06,
+      "loss": 1.7083,
+      "step": 2026
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.7262702661498385e-06,
+      "loss": 1.3724,
+      "step": 2027
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.716559203092118e-06,
+      "loss": 1.6614,
+      "step": 2028
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.7068530847092105e-06,
+      "loss": 1.6806,
+      "step": 2029
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.697151922205575e-06,
+      "loss": 1.4728,
+      "step": 2030
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.687455726779933e-06,
+      "loss": 1.4632,
+      "step": 2031
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.677764509625286e-06,
+      "loss": 1.3558,
+      "step": 2032
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.668078281928876e-06,
+      "loss": 1.4241,
+      "step": 2033
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.658397054872197e-06,
+      "loss": 1.4815,
+      "step": 2034
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.648720839630963e-06,
+      "loss": 1.5368,
+      "step": 2035
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.639049647375106e-06,
+      "loss": 1.5919,
+      "step": 2036
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.629383489268756e-06,
+      "loss": 1.3492,
+      "step": 2037
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.619722376470238e-06,
+      "loss": 1.55,
+      "step": 2038
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.610066320132044e-06,
+      "loss": 1.5479,
+      "step": 2039
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.600415331400844e-06,
+      "loss": 1.3407,
+      "step": 2040
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.590769421417435e-06,
+      "loss": 1.5264,
+      "step": 2041
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.581128601316774e-06,
+      "loss": 1.438,
+      "step": 2042
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.571492882227936e-06,
+      "loss": 1.5509,
+      "step": 2043
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.561862275274088e-06,
+      "loss": 1.583,
+      "step": 2044
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.552236791572528e-06,
+      "loss": 1.6311,
+      "step": 2045
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.542616442234618e-06,
+      "loss": 1.3332,
+      "step": 2046
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.5330012383658e-06,
+      "loss": 1.342,
+      "step": 2047
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.523391191065575e-06,
+      "loss": 1.5525,
+      "step": 2048
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.5137863114274895e-06,
+      "loss": 1.4105,
+      "step": 2049
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.504186610539131e-06,
+      "loss": 1.4668,
+      "step": 2050
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.4945920994820965e-06,
+      "loss": 1.4489,
+      "step": 2051
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.485002789332015e-06,
+      "loss": 1.4309,
+      "step": 2052
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.4754186911584784e-06,
+      "loss": 1.5879,
+      "step": 2053
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.465839816025093e-06,
+      "loss": 1.4515,
+      "step": 2054
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.456266174989425e-06,
+      "loss": 1.2766,
+      "step": 2055
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.446697779102984e-06,
+      "loss": 1.4414,
+      "step": 2056
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.437134639411248e-06,
+      "loss": 1.2883,
+      "step": 2057
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.427576766953615e-06,
+      "loss": 1.2834,
+      "step": 2058
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.418024172763403e-06,
+      "loss": 1.5109,
+      "step": 2059
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.40847686786784e-06,
+      "loss": 1.3691,
+      "step": 2060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.3989348632880454e-06,
+      "loss": 1.1664,
+      "step": 2061
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.3893981700390215e-06,
+      "loss": 1.3533,
+      "step": 2062
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.379866799129637e-06,
+      "loss": 1.1776,
+      "step": 2063
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.370340761562629e-06,
+      "loss": 1.2499,
+      "step": 2064
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.360820068334556e-06,
+      "loss": 1.1365,
+      "step": 2065
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.35130473043582e-06,
+      "loss": 1.4688,
+      "step": 2066
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.34179475885065e-06,
+      "loss": 1.3507,
+      "step": 2067
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.332290164557057e-06,
+      "loss": 1.3723,
+      "step": 2068
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.322790958526867e-06,
+      "loss": 1.3547,
+      "step": 2069
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.313297151725679e-06,
+      "loss": 1.3899,
+      "step": 2070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.303808755112845e-06,
+      "loss": 1.2218,
+      "step": 2071
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.294325779641495e-06,
+      "loss": 1.3613,
+      "step": 2072
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.284848236258486e-06,
+      "loss": 1.3062,
+      "step": 2073
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.275376135904408e-06,
+      "loss": 1.517,
+      "step": 2074
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.265909489513568e-06,
+      "loss": 1.6136,
+      "step": 2075
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.256448308013977e-06,
+      "loss": 1.2242,
+      "step": 2076
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.246992602327337e-06,
+      "loss": 1.3882,
+      "step": 2077
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.2375423833690255e-06,
+      "loss": 1.193,
+      "step": 2078
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.228097662048104e-06,
+      "loss": 1.2132,
+      "step": 2079
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.218658449267254e-06,
+      "loss": 1.1066,
+      "step": 2080
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.209224755922834e-06,
+      "loss": 1.1639,
+      "step": 2081
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1997965929048125e-06,
+      "loss": 1.3483,
+      "step": 2082
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.190373971096765e-06,
+      "loss": 1.1136,
+      "step": 2083
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.180956901375893e-06,
+      "loss": 1.2645,
+      "step": 2084
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.171545394612975e-06,
+      "loss": 1.3639,
+      "step": 2085
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1621394616723705e-06,
+      "loss": 1.3123,
+      "step": 2086
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.152739113412004e-06,
+      "loss": 1.0803,
+      "step": 2087
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1433443606833535e-06,
+      "loss": 1.187,
+      "step": 2088
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.133955214331439e-06,
+      "loss": 1.2414,
+      "step": 2089
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.124571685194804e-06,
+      "loss": 1.3383,
+      "step": 2090
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.115193784105521e-06,
+      "loss": 1.3694,
+      "step": 2091
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.105821521889147e-06,
+      "loss": 1.3097,
+      "step": 2092
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.096454909364738e-06,
+      "loss": 1.3137,
+      "step": 2093
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.087093957344841e-06,
+      "loss": 1.2966,
+      "step": 2094
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.07773867663544e-06,
+      "loss": 1.3454,
+      "step": 2095
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.068389078035999e-06,
+      "loss": 1.1664,
+      "step": 2096
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.059045172339412e-06,
+      "loss": 1.3639,
+      "step": 2097
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.049706970332e-06,
+      "loss": 1.2479,
+      "step": 2098
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.040374482793501e-06,
+      "loss": 1.2679,
+      "step": 2099
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.031047720497058e-06,
+      "loss": 1.3481,
+      "step": 2100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.021726694209205e-06,
+      "loss": 1.4994,
+      "step": 2101
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.0124114146898505e-06,
+      "loss": 1.3932,
+      "step": 2102
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.003101892692276e-06,
+      "loss": 1.0975,
+      "step": 2103
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.99379813896311e-06,
+      "loss": 1.2963,
+      "step": 2104
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.984500164242322e-06,
+      "loss": 1.5983,
+      "step": 2105
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.9752079792632244e-06,
+      "loss": 1.3584,
+      "step": 2106
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.965921594752426e-06,
+      "loss": 1.5457,
+      "step": 2107
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.956641021429845e-06,
+      "loss": 1.2898,
+      "step": 2108
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.947366270008708e-06,
+      "loss": 1.3527,
+      "step": 2109
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.938097351195499e-06,
+      "loss": 1.4179,
+      "step": 2110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.928834275689982e-06,
+      "loss": 1.5586,
+      "step": 2111
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.919577054185169e-06,
+      "loss": 1.0501,
+      "step": 2112
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.910325697367321e-06,
+      "loss": 1.4704,
+      "step": 2113
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.9010802159159224e-06,
+      "loss": 1.1302,
+      "step": 2114
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.891840620503677e-06,
+      "loss": 1.1881,
+      "step": 2115
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.882606921796498e-06,
+      "loss": 1.2761,
+      "step": 2116
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.873379130453483e-06,
+      "loss": 1.2665,
+      "step": 2117
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.864157257126928e-06,
+      "loss": 1.3159,
+      "step": 2118
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.854941312462272e-06,
+      "loss": 1.421,
+      "step": 2119
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.845731307098127e-06,
+      "loss": 1.1092,
+      "step": 2120
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.836527251666251e-06,
+      "loss": 1.3528,
+      "step": 2121
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.8273291567915225e-06,
+      "loss": 1.2405,
+      "step": 2122
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.818137033091948e-06,
+      "loss": 1.2538,
+      "step": 2123
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.808950891178634e-06,
+      "loss": 1.3561,
+      "step": 2124
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.79977074165579e-06,
+      "loss": 1.2636,
+      "step": 2125
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.790596595120699e-06,
+      "loss": 1.106,
+      "step": 2126
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.781428462163722e-06,
+      "loss": 1.1109,
+      "step": 2127
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.772266353368273e-06,
+      "loss": 1.3674,
+      "step": 2128
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.763110279310816e-06,
+      "loss": 1.4383,
+      "step": 2129
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.753960250560843e-06,
+      "loss": 1.3045,
+      "step": 2130
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.74481627768088e-06,
+      "loss": 1.2212,
+      "step": 2131
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.7356783712264405e-06,
+      "loss": 1.1761,
+      "step": 2132
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.7265465417460596e-06,
+      "loss": 1.4969,
+      "step": 2133
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.7174207997812436e-06,
+      "loss": 1.3492,
+      "step": 2134
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.708301155866463e-06,
+      "loss": 1.6779,
+      "step": 2135
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.699187620529173e-06,
+      "loss": 1.4172,
+      "step": 2136
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.690080204289757e-06,
+      "loss": 1.2919,
+      "step": 2137
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.680978917661544e-06,
+      "loss": 1.4052,
+      "step": 2138
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.671883771150783e-06,
+      "loss": 1.3718,
+      "step": 2139
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.662794775256637e-06,
+      "loss": 1.3918,
+      "step": 2140
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.653711940471169e-06,
+      "loss": 1.2978,
+      "step": 2141
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6446352772793256e-06,
+      "loss": 1.1112,
+      "step": 2142
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.635564796158946e-06,
+      "loss": 1.359,
+      "step": 2143
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.626500507580701e-06,
+      "loss": 1.3256,
+      "step": 2144
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.617442422008146e-06,
+      "loss": 1.27,
+      "step": 2145
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.608390549897661e-06,
+      "loss": 1.457,
+      "step": 2146
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.5993449016984425e-06,
+      "loss": 1.2275,
+      "step": 2147
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.590305487852525e-06,
+      "loss": 1.3401,
+      "step": 2148
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.581272318794733e-06,
+      "loss": 1.5835,
+      "step": 2149
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.5722454049526825e-06,
+      "loss": 1.1953,
+      "step": 2150
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.5632247567467716e-06,
+      "loss": 1.3855,
+      "step": 2151
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.554210384590162e-06,
+      "loss": 1.1849,
+      "step": 2152
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.545202298888777e-06,
+      "loss": 1.387,
+      "step": 2153
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.536200510041271e-06,
+      "loss": 1.332,
+      "step": 2154
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.5272050284390515e-06,
+      "loss": 1.4126,
+      "step": 2155
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.518215864466219e-06,
+      "loss": 1.686,
+      "step": 2156
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.5092330284995935e-06,
+      "loss": 1.3139,
+      "step": 2157
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.5002565309087e-06,
+      "loss": 1.1855,
+      "step": 2158
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.491286382055722e-06,
+      "loss": 1.4582,
+      "step": 2159
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.482322592295541e-06,
+      "loss": 1.0434,
+      "step": 2160
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.473365171975684e-06,
+      "loss": 1.4501,
+      "step": 2161
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.4644141314363165e-06,
+      "loss": 1.3294,
+      "step": 2162
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.45546948101026e-06,
+      "loss": 1.5021,
+      "step": 2163
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.446531231022946e-06,
+      "loss": 1.4195,
+      "step": 2164
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.43759939179242e-06,
+      "loss": 1.491,
+      "step": 2165
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.4286739736293285e-06,
+      "loss": 1.3028,
+      "step": 2166
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.419754986836904e-06,
+      "loss": 1.4098,
+      "step": 2167
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.4108424417109566e-06,
+      "loss": 1.367,
+      "step": 2168
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.401936348539855e-06,
+      "loss": 1.4177,
+      "step": 2169
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.393036717604536e-06,
+      "loss": 1.2032,
+      "step": 2170
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.384143559178454e-06,
+      "loss": 1.0496,
+      "step": 2171
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.375256883527603e-06,
+      "loss": 1.2313,
+      "step": 2172
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.366376700910504e-06,
+      "loss": 1.2404,
+      "step": 2173
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.357503021578158e-06,
+      "loss": 1.1631,
+      "step": 2174
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.348635855774082e-06,
+      "loss": 1.5188,
+      "step": 2175
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.339775213734262e-06,
+      "loss": 1.1959,
+      "step": 2176
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.330921105687155e-06,
+      "loss": 1.666,
+      "step": 2177
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.322073541853677e-06,
+      "loss": 1.2572,
+      "step": 2178
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.313232532447189e-06,
+      "loss": 1.3524,
+      "step": 2179
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.304398087673486e-06,
+      "loss": 1.3266,
+      "step": 2180
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.295570217730779e-06,
+      "loss": 1.237,
+      "step": 2181
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.286748932809707e-06,
+      "loss": 1.4659,
+      "step": 2182
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.277934243093284e-06,
+      "loss": 1.2491,
+      "step": 2183
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.26912615875692e-06,
+      "loss": 1.3558,
+      "step": 2184
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.260324689968416e-06,
+      "loss": 1.3576,
+      "step": 2185
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2515298468879064e-06,
+      "loss": 1.4268,
+      "step": 2186
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2427416396679024e-06,
+      "loss": 1.2021,
+      "step": 2187
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.233960078453242e-06,
+      "loss": 1.3033,
+      "step": 2188
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.225185173381095e-06,
+      "loss": 1.2348,
+      "step": 2189
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.216416934580947e-06,
+      "loss": 1.2703,
+      "step": 2190
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.207655372174587e-06,
+      "loss": 1.1938,
+      "step": 2191
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.198900496276101e-06,
+      "loss": 1.2686,
+      "step": 2192
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.190152316991851e-06,
+      "loss": 1.2872,
+      "step": 2193
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.181410844420473e-06,
+      "loss": 1.1363,
+      "step": 2194
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1726760886528595e-06,
+      "loss": 1.5274,
+      "step": 2195
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1639480597721425e-06,
+      "loss": 1.3338,
+      "step": 2196
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.155226767853709e-06,
+      "loss": 1.2767,
+      "step": 2197
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.146512222965144e-06,
+      "loss": 1.2614,
+      "step": 2198
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.137804435166253e-06,
+      "loss": 1.5594,
+      "step": 2199
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.129103414509052e-06,
+      "loss": 1.3395,
+      "step": 2200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.120409171037734e-06,
+      "loss": 1.3943,
+      "step": 2201
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.111721714788671e-06,
+      "loss": 1.4386,
+      "step": 2202
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.1030410557904e-06,
+      "loss": 1.725,
+      "step": 2203
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0943672040636115e-06,
+      "loss": 1.1222,
+      "step": 2204
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.085700169621139e-06,
+      "loss": 1.2422,
+      "step": 2205
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.077039962467946e-06,
+      "loss": 1.3387,
+      "step": 2206
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0683865926011125e-06,
+      "loss": 1.2143,
+      "step": 2207
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.059740070009826e-06,
+      "loss": 1.3916,
+      "step": 2208
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.05110040467538e-06,
+      "loss": 1.4048,
+      "step": 2209
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.042467606571134e-06,
+      "loss": 1.3357,
+      "step": 2210
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.033841685662529e-06,
+      "loss": 1.1663,
+      "step": 2211
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.025222651907077e-06,
+      "loss": 1.416,
+      "step": 2212
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.016610515254324e-06,
+      "loss": 1.2482,
+      "step": 2213
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.008005285645863e-06,
+      "loss": 1.7097,
+      "step": 2214
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.999406973015311e-06,
+      "loss": 1.3487,
+      "step": 2215
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.990815587288299e-06,
+      "loss": 1.3602,
+      "step": 2216
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.982231138382465e-06,
+      "loss": 1.2552,
+      "step": 2217
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.973653636207437e-06,
+      "loss": 1.2376,
+      "step": 2218
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.965083090664827e-06,
+      "loss": 1.4621,
+      "step": 2219
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.95651951164821e-06,
+      "loss": 1.5281,
+      "step": 2220
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9479629090431285e-06,
+      "loss": 1.3224,
+      "step": 2221
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.939413292727061e-06,
+      "loss": 1.1903,
+      "step": 2222
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.930870672569427e-06,
+      "loss": 1.4847,
+      "step": 2223
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.922335058431575e-06,
+      "loss": 1.1559,
+      "step": 2224
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9138064601667594e-06,
+      "loss": 1.3158,
+      "step": 2225
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9052848876201285e-06,
+      "loss": 1.2609,
+      "step": 2226
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.896770350628737e-06,
+      "loss": 1.4745,
+      "step": 2227
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.888262859021508e-06,
+      "loss": 1.4648,
+      "step": 2228
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.879762422619231e-06,
+      "loss": 1.1463,
+      "step": 2229
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8712690512345555e-06,
+      "loss": 1.1907,
+      "step": 2230
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.862782754671971e-06,
+      "loss": 1.0396,
+      "step": 2231
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.854303542727803e-06,
+      "loss": 1.3874,
+      "step": 2232
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.845831425190195e-06,
+      "loss": 1.1515,
+      "step": 2233
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.837366411839114e-06,
+      "loss": 1.242,
+      "step": 2234
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.828908512446305e-06,
+      "loss": 0.9936,
+      "step": 2235
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.820457736775312e-06,
+      "loss": 1.3179,
+      "step": 2236
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.8120140945814676e-06,
+      "loss": 1.1927,
+      "step": 2237
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.8035775956118416e-06,
+      "loss": 1.1989,
+      "step": 2238
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7951482496052862e-06,
+      "loss": 1.1265,
+      "step": 2239
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7867260662923776e-06,
+      "loss": 1.284,
+      "step": 2240
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7783110553954325e-06,
+      "loss": 1.3772,
+      "step": 2241
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7699032266284863e-06,
+      "loss": 1.489,
+      "step": 2242
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.76150258969728e-06,
+      "loss": 1.2711,
+      "step": 2243
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.753109154299258e-06,
+      "loss": 1.2114,
+      "step": 2244
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7447229301235443e-06,
+      "loss": 1.5145,
+      "step": 2245
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.736343926850954e-06,
+      "loss": 1.273,
+      "step": 2246
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7279721541539437e-06,
+      "loss": 1.323,
+      "step": 2247
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.719607621696637e-06,
+      "loss": 1.5705,
+      "step": 2248
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7112503391348077e-06,
+      "loss": 1.2517,
+      "step": 2249
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.702900316115836e-06,
+      "loss": 1.3973,
+      "step": 2250
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.694557562278749e-06,
+      "loss": 1.3289,
+      "step": 2251
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6862220872541664e-06,
+      "loss": 1.6706,
+      "step": 2252
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6778939006643024e-06,
+      "loss": 1.4896,
+      "step": 2253
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6695730121229734e-06,
+      "loss": 1.4187,
+      "step": 2254
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6612594312355576e-06,
+      "loss": 1.3973,
+      "step": 2255
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6529531675990027e-06,
+      "loss": 1.1565,
+      "step": 2256
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6446542308018097e-06,
+      "loss": 1.4017,
+      "step": 2257
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6363626304240185e-06,
+      "loss": 1.2934,
+      "step": 2258
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.628078376037205e-06,
+      "loss": 1.3579,
+      "step": 2259
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6198014772044575e-06,
+      "loss": 1.1242,
+      "step": 2260
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6115319434803897e-06,
+      "loss": 1.3785,
+      "step": 2261
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6032697844110896e-06,
+      "loss": 1.5389,
+      "step": 2262
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.595015009534146e-06,
+      "loss": 1.3565,
+      "step": 2263
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5867676283786335e-06,
+      "loss": 1.3843,
+      "step": 2264
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5785276504650633e-06,
+      "loss": 1.2854,
+      "step": 2265
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5702950853054284e-06,
+      "loss": 1.2567,
+      "step": 2266
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5620699424031557e-06,
+      "loss": 1.184,
+      "step": 2267
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.55385223125309e-06,
+      "loss": 1.1989,
+      "step": 2268
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5456419613415182e-06,
+      "loss": 1.3715,
+      "step": 2269
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5374391421461273e-06,
+      "loss": 1.0921,
+      "step": 2270
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.529243783136004e-06,
+      "loss": 1.2102,
+      "step": 2271
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.521055893771618e-06,
+      "loss": 1.2587,
+      "step": 2272
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.512875483504833e-06,
+      "loss": 1.3027,
+      "step": 2273
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5047025617788578e-06,
+      "loss": 1.3328,
+      "step": 2274
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4965371380282678e-06,
+      "loss": 1.5154,
+      "step": 2275
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.488379221678989e-06,
+      "loss": 1.341,
+      "step": 2276
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4802288221482637e-06,
+      "loss": 1.2705,
+      "step": 2277
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4720859488446744e-06,
+      "loss": 1.355,
+      "step": 2278
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.463950611168111e-06,
+      "loss": 1.0602,
+      "step": 2279
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.45582281850975e-06,
+      "loss": 1.5044,
+      "step": 2280
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4477025802520805e-06,
+      "loss": 1.3803,
+      "step": 2281
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4395899057688575e-06,
+      "loss": 1.2769,
+      "step": 2282
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.431484804425107e-06,
+      "loss": 0.897,
+      "step": 2283
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4233872855771145e-06,
+      "loss": 1.1887,
+      "step": 2284
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.41529735857241e-06,
+      "loss": 1.2734,
+      "step": 2285
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.407215032749763e-06,
+      "loss": 1.3257,
+      "step": 2286
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3991403174391614e-06,
+      "loss": 1.2252,
+      "step": 2287
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3910732219618247e-06,
+      "loss": 1.3858,
+      "step": 2288
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3830137556301524e-06,
+      "loss": 1.3296,
+      "step": 2289
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.374961927747751e-06,
+      "loss": 1.3807,
+      "step": 2290
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.366917747609414e-06,
+      "loss": 1.3836,
+      "step": 2291
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.358881224501096e-06,
+      "loss": 1.3385,
+      "step": 2292
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3508523676999174e-06,
+      "loss": 1.2616,
+      "step": 2293
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.342831186474149e-06,
+      "loss": 1.1951,
+      "step": 2294
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3348176900831996e-06,
+      "loss": 1.1339,
+      "step": 2295
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.326811887777607e-06,
+      "loss": 1.1549,
+      "step": 2296
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.318813788799029e-06,
+      "loss": 1.3469,
+      "step": 2297
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.31082340238023e-06,
+      "loss": 1.3222,
+      "step": 2298
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.302840737745069e-06,
+      "loss": 1.4261,
+      "step": 2299
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2948658041084956e-06,
+      "loss": 1.4198,
+      "step": 2300
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2868986106765312e-06,
+      "loss": 1.0948,
+      "step": 2301
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.27893916664626e-06,
+      "loss": 1.3823,
+      "step": 2302
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.270987481205832e-06,
+      "loss": 1.5382,
+      "step": 2303
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2630435635344283e-06,
+      "loss": 1.3864,
+      "step": 2304
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2551074228022685e-06,
+      "loss": 1.2965,
+      "step": 2305
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2471790681705928e-06,
+      "loss": 1.2093,
+      "step": 2306
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.239258508791656e-06,
+      "loss": 1.3189,
+      "step": 2307
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2313457538087133e-06,
+      "loss": 1.3511,
+      "step": 2308
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.223440812356009e-06,
+      "loss": 1.3273,
+      "step": 2309
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.215543693558769e-06,
+      "loss": 1.2002,
+      "step": 2310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2076544065331892e-06,
+      "loss": 1.1196,
+      "step": 2311
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1997729603864258e-06,
+      "loss": 1.4433,
+      "step": 2312
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.191899364216581e-06,
+      "loss": 1.3474,
+      "step": 2313
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1840336271126935e-06,
+      "loss": 1.176,
+      "step": 2314
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1761757581547403e-06,
+      "loss": 1.1506,
+      "step": 2315
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.168325766413607e-06,
+      "loss": 1.2698,
+      "step": 2316
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.160483660951078e-06,
+      "loss": 1.2386,
+      "step": 2317
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.152649450819852e-06,
+      "loss": 1.297,
+      "step": 2318
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1448231450635045e-06,
+      "loss": 1.1578,
+      "step": 2319
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1370047527164836e-06,
+      "loss": 1.1944,
+      "step": 2320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1291942828041077e-06,
+      "loss": 1.3597,
+      "step": 2321
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1213917443425456e-06,
+      "loss": 1.5561,
+      "step": 2322
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1135971463388146e-06,
+      "loss": 1.4728,
+      "step": 2323
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1058104977907576e-06,
+      "loss": 1.4585,
+      "step": 2324
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0980318076870586e-06,
+      "loss": 1.2941,
+      "step": 2325
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0902610850071922e-06,
+      "loss": 1.0645,
+      "step": 2326
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.082498338721447e-06,
+      "loss": 1.5819,
+      "step": 2327
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0747435777909106e-06,
+      "loss": 0.9561,
+      "step": 2328
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0669968111674343e-06,
+      "loss": 1.3767,
+      "step": 2329
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0592580477936606e-06,
+      "loss": 1.3289,
+      "step": 2330
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0515272966029852e-06,
+      "loss": 1.2773,
+      "step": 2331
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0438045665195436e-06,
+      "loss": 1.7359,
+      "step": 2332
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0360898664582316e-06,
+      "loss": 1.3402,
+      "step": 2333
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0283832053246644e-06,
+      "loss": 1.5816,
+      "step": 2334
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0206845920151773e-06,
+      "loss": 1.2385,
+      "step": 2335
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.012994035416815e-06,
+      "loss": 1.3047,
+      "step": 2336
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.0053115444073345e-06,
+      "loss": 1.6096,
+      "step": 2337
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.99763712785516e-06,
+      "loss": 1.3115,
+      "step": 2338
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.989970794619408e-06,
+      "loss": 1.2087,
+      "step": 2339
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.982312553549872e-06,
+      "loss": 1.5077,
+      "step": 2340
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9746624134869796e-06,
+      "loss": 1.2983,
+      "step": 2341
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.967020383261834e-06,
+      "loss": 1.3754,
+      "step": 2342
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.959386471696163e-06,
+      "loss": 1.5062,
+      "step": 2343
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.951760687602313e-06,
+      "loss": 1.132,
+      "step": 2344
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9441430397832703e-06,
+      "loss": 1.3687,
+      "step": 2345
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9365335370326143e-06,
+      "loss": 1.2675,
+      "step": 2346
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 1.2936,
+      "step": 2347
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.921339001863771e-06,
+      "loss": 1.0697,
+      "step": 2348
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9137539869856966e-06,
+      "loss": 1.3286,
+      "step": 2349
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9061771522562143e-06,
+      "loss": 1.1394,
+      "step": 2350
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8986085064217896e-06,
+      "loss": 1.4079,
+      "step": 2351
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.89104805821945e-06,
+      "loss": 1.4645,
+      "step": 2352
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.883495816376738e-06,
+      "loss": 1.3394,
+      "step": 2353
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.875951789611734e-06,
+      "loss": 1.5819,
+      "step": 2354
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.868415986633045e-06,
+      "loss": 1.3452,
+      "step": 2355
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.860888416139761e-06,
+      "loss": 1.2708,
+      "step": 2356
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8533690868214923e-06,
+      "loss": 1.4011,
+      "step": 2357
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8458580073583262e-06,
+      "loss": 1.5379,
+      "step": 2358
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8383551864208158e-06,
+      "loss": 1.2203,
+      "step": 2359
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8308606326700016e-06,
+      "loss": 1.2897,
+      "step": 2360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8233743547573668e-06,
+      "loss": 1.4008,
+      "step": 2361
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8158963613248437e-06,
+      "loss": 1.3286,
+      "step": 2362
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8084266610048007e-06,
+      "loss": 1.3081,
+      "step": 2363
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8009652624200436e-06,
+      "loss": 1.5473,
+      "step": 2364
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7935121741837756e-06,
+      "loss": 1.0324,
+      "step": 2365
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7860674048996174e-06,
+      "loss": 1.4509,
+      "step": 2366
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.778630963161595e-06,
+      "loss": 1.1272,
+      "step": 2367
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7712028575541005e-06,
+      "loss": 1.1759,
+      "step": 2368
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.763783096651924e-06,
+      "loss": 1.3931,
+      "step": 2369
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.756371689020214e-06,
+      "loss": 1.1027,
+      "step": 2370
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7489686432144668e-06,
+      "loss": 1.1906,
+      "step": 2371
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7415739677805432e-06,
+      "loss": 1.4453,
+      "step": 2372
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.73418767125463e-06,
+      "loss": 1.4182,
+      "step": 2373
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7268097621632473e-06,
+      "loss": 1.4502,
+      "step": 2374
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7194402490232286e-06,
+      "loss": 1.1489,
+      "step": 2375
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.712079140341718e-06,
+      "loss": 1.3734,
+      "step": 2376
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.704726444616158e-06,
+      "loss": 1.3697,
+      "step": 2377
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.697382170334275e-06,
+      "loss": 1.1551,
+      "step": 2378
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6900463259740873e-06,
+      "loss": 1.2474,
+      "step": 2379
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6827189200038627e-06,
+      "loss": 1.2226,
+      "step": 2380
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.675399960882138e-06,
+      "loss": 1.0981,
+      "step": 2381
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6680894570577042e-06,
+      "loss": 1.3181,
+      "step": 2382
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6607874169695847e-06,
+      "loss": 1.3118,
+      "step": 2383
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.653493849047032e-06,
+      "loss": 1.1207,
+      "step": 2384
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.646208761709523e-06,
+      "loss": 1.2679,
+      "step": 2385
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.638932163366742e-06,
+      "loss": 1.2581,
+      "step": 2386
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.631664062418574e-06,
+      "loss": 1.38,
+      "step": 2387
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.624404467255095e-06,
+      "loss": 1.2731,
+      "step": 2388
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6171533862565644e-06,
+      "loss": 1.5697,
+      "step": 2389
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6099108277934105e-06,
+      "loss": 1.1026,
+      "step": 2390
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6026768002262236e-06,
+      "loss": 1.5086,
+      "step": 2391
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5954513119057476e-06,
+      "loss": 1.28,
+      "step": 2392
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.588234371172866e-06,
+      "loss": 1.0798,
+      "step": 2393
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.581025986358602e-06,
+      "loss": 1.352,
+      "step": 2394
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5738261657840948e-06,
+      "loss": 1.1349,
+      "step": 2395
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5666349177605996e-06,
+      "loss": 1.3503,
+      "step": 2396
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.559452250589478e-06,
+      "loss": 1.396,
+      "step": 2397
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5522781725621814e-06,
+      "loss": 1.1827,
+      "step": 2398
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.545112691960252e-06,
+      "loss": 1.3093,
+      "step": 2399
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.537955817055303e-06,
+      "loss": 1.3883,
+      "step": 2400
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.530807556109015e-06,
+      "loss": 1.3333,
+      "step": 2401
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.523667917373125e-06,
+      "loss": 1.5305,
+      "step": 2402
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5165369090894153e-06,
+      "loss": 1.5228,
+      "step": 2403
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5094145394897087e-06,
+      "loss": 1.4256,
+      "step": 2404
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5023008167958486e-06,
+      "loss": 1.413,
+      "step": 2405
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4951957492197097e-06,
+      "loss": 1.1259,
+      "step": 2406
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4880993449631664e-06,
+      "loss": 1.2488,
+      "step": 2407
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4810116122180862e-06,
+      "loss": 1.4505,
+      "step": 2408
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.473932559166341e-06,
+      "loss": 1.393,
+      "step": 2409
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4668621939797745e-06,
+      "loss": 1.3179,
+      "step": 2410
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.459800524820202e-06,
+      "loss": 1.2817,
+      "step": 2411
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4527475598394013e-06,
+      "loss": 1.2026,
+      "step": 2412
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.445703307179104e-06,
+      "loss": 1.5467,
+      "step": 2413
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.438667774970981e-06,
+      "loss": 1.1992,
+      "step": 2414
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4316409713366353e-06,
+      "loss": 1.308,
+      "step": 2415
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.424622904387608e-06,
+      "loss": 1.3679,
+      "step": 2416
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4176135822253322e-06,
+      "loss": 1.4844,
+      "step": 2417
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.4106130129411608e-06,
+      "loss": 1.183,
+      "step": 2418
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.4036212046163465e-06,
+      "loss": 1.4464,
+      "step": 2419
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3966381653220105e-06,
+      "loss": 1.5431,
+      "step": 2420
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3896639031191714e-06,
+      "loss": 1.2408,
+      "step": 2421
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3826984260587084e-06,
+      "loss": 1.3841,
+      "step": 2422
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3757417421813476e-06,
+      "loss": 1.3587,
+      "step": 2423
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3687938595176853e-06,
+      "loss": 1.3939,
+      "step": 2424
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3618547860881426e-06,
+      "loss": 1.3774,
+      "step": 2425
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.354924529902978e-06,
+      "loss": 1.5344,
+      "step": 2426
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.348003098962267e-06,
+      "loss": 1.0509,
+      "step": 2427
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3410905012559095e-06,
+      "loss": 1.4069,
+      "step": 2428
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.334186744763591e-06,
+      "loss": 1.4163,
+      "step": 2429
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.327291837454799e-06,
+      "loss": 1.1845,
+      "step": 2430
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3204057872888164e-06,
+      "loss": 1.5327,
+      "step": 2431
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3135286022146785e-06,
+      "loss": 1.2367,
+      "step": 2432
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.306660290171211e-06,
+      "loss": 1.1969,
+      "step": 2433
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2998008590869838e-06,
+      "loss": 1.417,
+      "step": 2434
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2929503168803104e-06,
+      "loss": 1.6836,
+      "step": 2435
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2861086714592573e-06,
+      "loss": 1.2917,
+      "step": 2436
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2792759307216116e-06,
+      "loss": 1.3946,
+      "step": 2437
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2724521025548828e-06,
+      "loss": 1.4176,
+      "step": 2438
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.265637194836293e-06,
+      "loss": 1.3988,
+      "step": 2439
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2588312154327654e-06,
+      "loss": 1.1902,
+      "step": 2440
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2520341722009154e-06,
+      "loss": 1.588,
+      "step": 2441
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.245246072987045e-06,
+      "loss": 1.3222,
+      "step": 2442
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2384669256271373e-06,
+      "loss": 1.3044,
+      "step": 2443
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.231696737946829e-06,
+      "loss": 1.5043,
+      "step": 2444
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2249355177614183e-06,
+      "loss": 1.2065,
+      "step": 2445
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2181832728758635e-06,
+      "loss": 1.5124,
+      "step": 2446
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.211440011084741e-06,
+      "loss": 1.4303,
+      "step": 2447
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.204705740172277e-06,
+      "loss": 1.4405,
+      "step": 2448
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1979804679123108e-06,
+      "loss": 1.4272,
+      "step": 2449
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.191264202068286e-06,
+      "loss": 1.4603,
+      "step": 2450
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.184556950393264e-06,
+      "loss": 1.4484,
+      "step": 2451
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.177858720629892e-06,
+      "loss": 1.2461,
+      "step": 2452
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1711695205104043e-06,
+      "loss": 1.422,
+      "step": 2453
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1644893577566118e-06,
+      "loss": 1.3058,
+      "step": 2454
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.157818240079893e-06,
+      "loss": 1.4497,
+      "step": 2455
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.151156175181185e-06,
+      "loss": 1.119,
+      "step": 2456
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.14450317075097e-06,
+      "loss": 1.054,
+      "step": 2457
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.137859234469286e-06,
+      "loss": 1.503,
+      "step": 2458
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1312243740056803e-06,
+      "loss": 1.3501,
+      "step": 2459
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.124598597019244e-06,
+      "loss": 1.0866,
+      "step": 2460
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.117981911158575e-06,
+      "loss": 1.3162,
+      "step": 2461
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1113743240617668e-06,
+      "loss": 1.2359,
+      "step": 2462
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.104775843356428e-06,
+      "loss": 1.3287,
+      "step": 2463
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0981864766596406e-06,
+      "loss": 1.3052,
+      "step": 2464
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0916062315779728e-06,
+      "loss": 1.3839,
+      "step": 2465
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.08503511570746e-06,
+      "loss": 1.5091,
+      "step": 2466
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.078473136633601e-06,
+      "loss": 1.2078,
+      "step": 2467
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0719203019313448e-06,
+      "loss": 1.4813,
+      "step": 2468
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0653766191650848e-06,
+      "loss": 1.4596,
+      "step": 2469
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.058842095888658e-06,
+      "loss": 1.3702,
+      "step": 2470
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.052316739645314e-06,
+      "loss": 1.3194,
+      "step": 2471
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0458005579677264e-06,
+      "loss": 1.3156,
+      "step": 2472
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.039293558377985e-06,
+      "loss": 1.3358,
+      "step": 2473
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0327957483875693e-06,
+      "loss": 1.4639,
+      "step": 2474
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0263071354973563e-06,
+      "loss": 1.3007,
+      "step": 2475
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.019827727197605e-06,
+      "loss": 1.1504,
+      "step": 2476
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.01335753096795e-06,
+      "loss": 1.5432,
+      "step": 2477
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.006896554277388e-06,
+      "loss": 1.3973,
+      "step": 2478
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.000444804584276e-06,
+      "loss": 1.2579,
+      "step": 2479
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.994002289336322e-06,
+      "loss": 1.3901,
+      "step": 2480
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9875690159705687e-06,
+      "loss": 1.216,
+      "step": 2481
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.981144991913392e-06,
+      "loss": 1.3229,
+      "step": 2482
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9747302245804944e-06,
+      "loss": 1.5232,
+      "step": 2483
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.968324721376884e-06,
+      "loss": 1.5483,
+      "step": 2484
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.961928489696888e-06,
+      "loss": 1.4365,
+      "step": 2485
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9555415369241228e-06,
+      "loss": 1.5034,
+      "step": 2486
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9491638704314853e-06,
+      "loss": 1.3267,
+      "step": 2487
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9427954975811704e-06,
+      "loss": 1.1495,
+      "step": 2488
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9364364257246317e-06,
+      "loss": 1.0451,
+      "step": 2489
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.930086662202589e-06,
+      "loss": 1.4353,
+      "step": 2490
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9237462143450215e-06,
+      "loss": 1.3248,
+      "step": 2491
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9174150894711465e-06,
+      "loss": 1.3341,
+      "step": 2492
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9110932948894268e-06,
+      "loss": 1.4738,
+      "step": 2493
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9047808378975485e-06,
+      "loss": 1.1825,
+      "step": 2494
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.898477725782424e-06,
+      "loss": 1.4088,
+      "step": 2495
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8921839658201723e-06,
+      "loss": 1.4177,
+      "step": 2496
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8858995652761247e-06,
+      "loss": 1.2335,
+      "step": 2497
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8796245314048046e-06,
+      "loss": 1.3409,
+      "step": 2498
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.873358871449913e-06,
+      "loss": 1.2533,
+      "step": 2499
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8671025926443464e-06,
+      "loss": 1.4159,
+      "step": 2500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8608557022101625e-06,
+      "loss": 1.1466,
+      "step": 2501
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8546182073585828e-06,
+      "loss": 1.8138,
+      "step": 2502
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.848390115289983e-06,
+      "loss": 1.4311,
+      "step": 2503
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8421714331938845e-06,
+      "loss": 1.2748,
+      "step": 2504
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.835962168248947e-06,
+      "loss": 1.329,
+      "step": 2505
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.829762327622958e-06,
+      "loss": 1.3823,
+      "step": 2506
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8235719184728263e-06,
+      "loss": 1.2133,
+      "step": 2507
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.817390947944574e-06,
+      "loss": 1.279,
+      "step": 2508
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.811219423173325e-06,
+      "loss": 1.2243,
+      "step": 2509
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.805057351283307e-06,
+      "loss": 1.4938,
+      "step": 2510
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7989047393878212e-06,
+      "loss": 1.5492,
+      "step": 2511
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7927615945892652e-06,
+      "loss": 1.3536,
+      "step": 2512
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7866279239790984e-06,
+      "loss": 1.4596,
+      "step": 2513
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7805037346378384e-06,
+      "loss": 0.96,
+      "step": 2514
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7743890336350723e-06,
+      "loss": 1.1131,
+      "step": 2515
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7682838280294224e-06,
+      "loss": 1.5239,
+      "step": 2516
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7621881248685569e-06,
+      "loss": 1.2377,
+      "step": 2517
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.756101931189169e-06,
+      "loss": 1.2082,
+      "step": 2518
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7500252540169782e-06,
+      "loss": 1.1942,
+      "step": 2519
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7439581003667161e-06,
+      "loss": 1.3535,
+      "step": 2520
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7379004772421215e-06,
+      "loss": 1.399,
+      "step": 2521
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7318523916359376e-06,
+      "loss": 1.3683,
+      "step": 2522
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7258138505298817e-06,
+      "loss": 1.1289,
+      "step": 2523
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7197848608946722e-06,
+      "loss": 1.5497,
+      "step": 2524
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7137654296899918e-06,
+      "loss": 1.3604,
+      "step": 2525
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7077555638644838e-06,
+      "loss": 1.4106,
+      "step": 2526
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7017552703557626e-06,
+      "loss": 1.4646,
+      "step": 2527
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6957645560903834e-06,
+      "loss": 1.6527,
+      "step": 2528
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6897834279838466e-06,
+      "loss": 1.2875,
+      "step": 2529
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6838118929405856e-06,
+      "loss": 1.1573,
+      "step": 2530
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6778499578539597e-06,
+      "loss": 1.195,
+      "step": 2531
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6718976296062484e-06,
+      "loss": 1.4534,
+      "step": 2532
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6659549150686373e-06,
+      "loss": 1.339,
+      "step": 2533
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.660021821101222e-06,
+      "loss": 1.4729,
+      "step": 2534
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.654098354552982e-06,
+      "loss": 1.279,
+      "step": 2535
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6481845222617875e-06,
+      "loss": 1.1192,
+      "step": 2536
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6422803310543954e-06,
+      "loss": 1.5088,
+      "step": 2537
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6363857877464161e-06,
+      "loss": 1.3397,
+      "step": 2538
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.63050089914234e-06,
+      "loss": 1.2984,
+      "step": 2539
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6246256720355047e-06,
+      "loss": 1.2336,
+      "step": 2540
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6187601132080888e-06,
+      "loss": 1.3496,
+      "step": 2541
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6129042294311227e-06,
+      "loss": 1.4351,
+      "step": 2542
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6070580274644599e-06,
+      "loss": 1.2279,
+      "step": 2543
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6012215140567789e-06,
+      "loss": 1.2667,
+      "step": 2544
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5953946959455756e-06,
+      "loss": 1.2945,
+      "step": 2545
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5895775798571523e-06,
+      "loss": 1.5061,
+      "step": 2546
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5837701725066123e-06,
+      "loss": 1.146,
+      "step": 2547
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5779724805978503e-06,
+      "loss": 1.3147,
+      "step": 2548
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5721845108235522e-06,
+      "loss": 1.6055,
+      "step": 2549
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5664062698651706e-06,
+      "loss": 1.2098,
+      "step": 2550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5606377643929305e-06,
+      "loss": 1.627,
+      "step": 2551
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5548790010658287e-06,
+      "loss": 1.3484,
+      "step": 2552
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.549129986531598e-06,
+      "loss": 1.2896,
+      "step": 2553
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5433907274267357e-06,
+      "loss": 1.2678,
+      "step": 2554
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.537661230376465e-06,
+      "loss": 1.2482,
+      "step": 2555
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5319415019947449e-06,
+      "loss": 1.2785,
+      "step": 2556
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5262315488842583e-06,
+      "loss": 1.5992,
+      "step": 2557
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5205313776364028e-06,
+      "loss": 1.2298,
+      "step": 2558
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5148409948312825e-06,
+      "loss": 1.2509,
+      "step": 2559
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.509160407037703e-06,
+      "loss": 1.0965,
+      "step": 2560
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.50348962081317e-06,
+      "loss": 1.4349,
+      "step": 2561
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4978286427038602e-06,
+      "loss": 1.489,
+      "step": 2562
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4921774792446365e-06,
+      "loss": 1.1246,
+      "step": 2563
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4865361369590392e-06,
+      "loss": 1.586,
+      "step": 2564
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4809046223592526e-06,
+      "loss": 1.2495,
+      "step": 2565
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4752829419461357e-06,
+      "loss": 1.3096,
+      "step": 2566
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4696711022091836e-06,
+      "loss": 1.1754,
+      "step": 2567
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4640691096265358e-06,
+      "loss": 1.525,
+      "step": 2568
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4584769706649626e-06,
+      "loss": 1.3613,
+      "step": 2569
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4528946917798603e-06,
+      "loss": 1.6165,
+      "step": 2570
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4473222794152443e-06,
+      "loss": 1.4984,
+      "step": 2571
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4417597400037386e-06,
+      "loss": 1.1309,
+      "step": 2572
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4362070799665705e-06,
+      "loss": 1.363,
+      "step": 2573
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4306643057135638e-06,
+      "loss": 1.386,
+      "step": 2574
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4251314236431278e-06,
+      "loss": 1.176,
+      "step": 2575
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4196084401422584e-06,
+      "loss": 1.3616,
+      "step": 2576
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4140953615865216e-06,
+      "loss": 1.4839,
+      "step": 2577
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4085921943400416e-06,
+      "loss": 1.5035,
+      "step": 2578
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4030989447555165e-06,
+      "loss": 1.2497,
+      "step": 2579
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.397615619174184e-06,
+      "loss": 1.3451,
+      "step": 2580
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.392142223925832e-06,
+      "loss": 1.3517,
+      "step": 2581
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3866787653287804e-06,
+      "loss": 1.2906,
+      "step": 2582
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3812252496898825e-06,
+      "loss": 1.5351,
+      "step": 2583
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3757816833045123e-06,
+      "loss": 1.2007,
+      "step": 2584
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3703480724565577e-06,
+      "loss": 1.2104,
+      "step": 2585
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3649244234184157e-06,
+      "loss": 1.2684,
+      "step": 2586
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3595107424509801e-06,
+      "loss": 1.3832,
+      "step": 2587
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3541070358036457e-06,
+      "loss": 1.5019,
+      "step": 2588
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3487133097142891e-06,
+      "loss": 1.1706,
+      "step": 2589
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3433295704092586e-06,
+      "loss": 1.3676,
+      "step": 2590
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3379558241033852e-06,
+      "loss": 1.1519,
+      "step": 2591
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.332592076999959e-06,
+      "loss": 1.3786,
+      "step": 2592
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.327238335290727e-06,
+      "loss": 1.2558,
+      "step": 2593
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3218946051558867e-06,
+      "loss": 1.4375,
+      "step": 2594
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3165608927640805e-06,
+      "loss": 1.544,
+      "step": 2595
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.311237204272383e-06,
+      "loss": 1.3591,
+      "step": 2596
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.305923545826302e-06,
+      "loss": 1.503,
+      "step": 2597
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3006199235597628e-06,
+      "loss": 1.1858,
+      "step": 2598
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.295326343595107e-06,
+      "loss": 1.4693,
+      "step": 2599
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2900428120430807e-06,
+      "loss": 1.316,
+      "step": 2600
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2847693350028422e-06,
+      "loss": 1.3022,
+      "step": 2601
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.279505918561923e-06,
+      "loss": 1.4677,
+      "step": 2602
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2742525687962615e-06,
+      "loss": 1.4128,
+      "step": 2603
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2690092917701636e-06,
+      "loss": 1.3809,
+      "step": 2604
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2637760935363053e-06,
+      "loss": 1.4159,
+      "step": 2605
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2585529801357377e-06,
+      "loss": 1.1574,
+      "step": 2606
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2533399575978644e-06,
+      "loss": 1.2838,
+      "step": 2607
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2481370319404407e-06,
+      "loss": 1.3414,
+      "step": 2608
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2429442091695677e-06,
+      "loss": 1.2157,
+      "step": 2609
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2377614952796825e-06,
+      "loss": 1.5108,
+      "step": 2610
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2325888962535536e-06,
+      "loss": 1.6073,
+      "step": 2611
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.227426418062273e-06,
+      "loss": 1.0034,
+      "step": 2612
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2222740666652532e-06,
+      "loss": 1.3755,
+      "step": 2613
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.217131848010209e-06,
+      "loss": 1.0961,
+      "step": 2614
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.211999768033162e-06,
+      "loss": 1.066,
+      "step": 2615
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2068778326584363e-06,
+      "loss": 1.2083,
+      "step": 2616
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2017660477986314e-06,
+      "loss": 0.8679,
+      "step": 2617
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.196664419354644e-06,
+      "loss": 1.2129,
+      "step": 2618
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1915729532156372e-06,
+      "loss": 1.354,
+      "step": 2619
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1864916552590467e-06,
+      "loss": 1.4909,
+      "step": 2620
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.18142053135057e-06,
+      "loss": 1.3966,
+      "step": 2621
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.176359587344158e-06,
+      "loss": 1.2884,
+      "step": 2622
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.171308829082013e-06,
+      "loss": 1.4897,
+      "step": 2623
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1662682623945753e-06,
+      "loss": 1.4529,
+      "step": 2624
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1612378931005296e-06,
+      "loss": 1.1059,
+      "step": 2625
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1562177270067766e-06,
+      "loss": 1.1902,
+      "step": 2626
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1512077699084435e-06,
+      "loss": 1.4603,
+      "step": 2627
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1462080275888799e-06,
+      "loss": 1.3359,
+      "step": 2628
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.14121850581963e-06,
+      "loss": 1.1298,
+      "step": 2629
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1362392103604536e-06,
+      "loss": 1.3936,
+      "step": 2630
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1312701469592978e-06,
+      "loss": 1.1067,
+      "step": 2631
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1263113213522948e-06,
+      "loss": 1.3448,
+      "step": 2632
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1213627392637671e-06,
+      "loss": 1.2833,
+      "step": 2633
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1164244064062101e-06,
+      "loss": 1.4893,
+      "step": 2634
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1114963284802826e-06,
+      "loss": 1.1514,
+      "step": 2635
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1065785111748117e-06,
+      "loss": 1.4632,
+      "step": 2636
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1016709601667764e-06,
+      "loss": 0.9996,
+      "step": 2637
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0967736811213048e-06,
+      "loss": 1.4497,
+      "step": 2638
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0918866796916671e-06,
+      "loss": 1.2752,
+      "step": 2639
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0870099615192765e-06,
+      "loss": 1.4437,
+      "step": 2640
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0821435322336616e-06,
+      "loss": 1.3854,
+      "step": 2641
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0772873974524833e-06,
+      "loss": 1.3988,
+      "step": 2642
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.072441562781522e-06,
+      "loss": 1.5402,
+      "step": 2643
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0676060338146555e-06,
+      "loss": 1.4782,
+      "step": 2644
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0627808161338794e-06,
+      "loss": 1.3631,
+      "step": 2645
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0579659153092759e-06,
+      "loss": 1.2488,
+      "step": 2646
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0531613368990211e-06,
+      "loss": 1.3453,
+      "step": 2647
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0483670864493777e-06,
+      "loss": 1.3787,
+      "step": 2648
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0435831694946807e-06,
+      "loss": 1.3678,
+      "step": 2649
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0388095915573427e-06,
+      "loss": 1.3098,
+      "step": 2650
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0340463581478333e-06,
+      "loss": 1.2779,
+      "step": 2651
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0292934747646921e-06,
+      "loss": 1.2566,
+      "step": 2652
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0245509468944992e-06,
+      "loss": 1.4866,
+      "step": 2653
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0198187800118842e-06,
+      "loss": 1.2636,
+      "step": 2654
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.015096979579525e-06,
+      "loss": 1.0431,
+      "step": 2655
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0103855510481186e-06,
+      "loss": 1.2232,
+      "step": 2656
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0056844998563998e-06,
+      "loss": 1.3522,
+      "step": 2657
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0009938314311186e-06,
+      "loss": 1.1809,
+      "step": 2658
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.963135511870414e-07,
+      "loss": 1.1264,
+      "step": 2659
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.916436645269434e-07,
+      "loss": 1.1923,
+      "step": 2660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.869841768415988e-07,
+      "loss": 1.4355,
+      "step": 2661
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.82335093509782e-07,
+      "loss": 1.4176,
+      "step": 2662
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.77696419898253e-07,
+      "loss": 1.4356,
+      "step": 2663
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.730681613617577e-07,
+      "loss": 1.3858,
+      "step": 2664
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.684503232430176e-07,
+      "loss": 1.3454,
+      "step": 2665
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.638429108727232e-07,
+      "loss": 1.4664,
+      "step": 2666
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.592459295695366e-07,
+      "loss": 1.257,
+      "step": 2667
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.546593846400742e-07,
+      "loss": 1.685,
+      "step": 2668
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.500832813788996e-07,
+      "loss": 1.1669,
+      "step": 2669
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.455176250685338e-07,
+      "loss": 1.6388,
+      "step": 2670
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.40962420979431e-07,
+      "loss": 1.6227,
+      "step": 2671
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.364176743699816e-07,
+      "loss": 1.6423,
+      "step": 2672
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.318833904865032e-07,
+      "loss": 1.2868,
+      "step": 2673
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.27359574563238e-07,
+      "loss": 1.0601,
+      "step": 2674
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.228462318223397e-07,
+      "loss": 1.1154,
+      "step": 2675
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.183433674738773e-07,
+      "loss": 1.2909,
+      "step": 2676
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.138509867158218e-07,
+      "loss": 1.5078,
+      "step": 2677
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.093690947340406e-07,
+      "loss": 1.3473,
+      "step": 2678
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.048976967022926e-07,
+      "loss": 1.4068,
+      "step": 2679
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.004367977822315e-07,
+      "loss": 1.2548,
+      "step": 2680
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.959864031233767e-07,
+      "loss": 1.3311,
+      "step": 2681
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.915465178631344e-07,
+      "loss": 1.7165,
+      "step": 2682
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.871171471267726e-07,
+      "loss": 1.1827,
+      "step": 2683
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.826982960274244e-07,
+      "loss": 1.6362,
+      "step": 2684
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.782899696660763e-07,
+      "loss": 1.5173,
+      "step": 2685
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.738921731315686e-07,
+      "loss": 0.9031,
+      "step": 2686
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.695049115005838e-07,
+      "loss": 1.3422,
+      "step": 2687
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.651281898376451e-07,
+      "loss": 1.2449,
+      "step": 2688
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.607620131951089e-07,
+      "loss": 1.3213,
+      "step": 2689
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.564063866131567e-07,
+      "loss": 1.4465,
+      "step": 2690
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.520613151197899e-07,
+      "loss": 1.2099,
+      "step": 2691
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.47726803730834e-07,
+      "loss": 1.4388,
+      "step": 2692
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.434028574499121e-07,
+      "loss": 1.3148,
+      "step": 2693
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.390894812684602e-07,
+      "loss": 1.2207,
+      "step": 2694
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.347866801657123e-07,
+      "loss": 1.3937,
+      "step": 2695
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.30494459108685e-07,
+      "loss": 1.392,
+      "step": 2696
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.262128230521949e-07,
+      "loss": 1.2449,
+      "step": 2697
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.219417769388316e-07,
+      "loss": 1.2626,
+      "step": 2698
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.176813256989613e-07,
+      "loss": 1.3935,
+      "step": 2699
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.134314742507221e-07,
+      "loss": 1.3667,
+      "step": 2700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.091922275000119e-07,
+      "loss": 1.2696,
+      "step": 2701
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.049635903404907e-07,
+      "loss": 1.5098,
+      "step": 2702
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.00745567653568e-07,
+      "loss": 1.3983,
+      "step": 2703
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.965381643084069e-07,
+      "loss": 1.5094,
+      "step": 2704
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.92341385161901e-07,
+      "loss": 1.2559,
+      "step": 2705
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.881552350586863e-07,
+      "loss": 1.3789,
+      "step": 2706
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.839797188311327e-07,
+      "loss": 1.284,
+      "step": 2707
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.798148412993234e-07,
+      "loss": 1.2795,
+      "step": 2708
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.756606072710737e-07,
+      "loss": 1.2699,
+      "step": 2709
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.715170215419043e-07,
+      "loss": 1.1096,
+      "step": 2710
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.673840888950424e-07,
+      "loss": 1.0773,
+      "step": 2711
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.632618141014237e-07,
+      "loss": 1.4543,
+      "step": 2712
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.591502019196773e-07,
+      "loss": 1.115,
+      "step": 2713
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.550492570961243e-07,
+      "loss": 1.6108,
+      "step": 2714
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.509589843647691e-07,
+      "loss": 1.4234,
+      "step": 2715
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.468793884473058e-07,
+      "loss": 1.216,
+      "step": 2716
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.428104740530906e-07,
+      "loss": 0.9836,
+      "step": 2717
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.387522458791552e-07,
+      "loss": 1.2867,
+      "step": 2718
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.34704708610201e-07,
+      "loss": 1.3161,
+      "step": 2719
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.306678669185774e-07,
+      "loss": 1.0458,
+      "step": 2720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.266417254642966e-07,
+      "loss": 1.2854,
+      "step": 2721
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.226262888950153e-07,
+      "loss": 1.3577,
+      "step": 2722
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.186215618460268e-07,
+      "loss": 1.2408,
+      "step": 2723
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.146275489402721e-07,
+      "loss": 1.4875,
+      "step": 2724
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.106442547883186e-07,
+      "loss": 1.3218,
+      "step": 2725
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.066716839883592e-07,
+      "loss": 1.1632,
+      "step": 2726
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.027098411262112e-07,
+      "loss": 1.295,
+      "step": 2727
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.987587307753052e-07,
+      "loss": 1.0619,
+      "step": 2728
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.94818357496686e-07,
+      "loss": 1.4265,
+      "step": 2729
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.908887258389974e-07,
+      "loss": 1.3887,
+      "step": 2730
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.869698403384939e-07,
+      "loss": 1.4124,
+      "step": 2731
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.830617055190136e-07,
+      "loss": 1.3867,
+      "step": 2732
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.791643258919889e-07,
+      "loss": 1.2196,
+      "step": 2733
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.752777059564431e-07,
+      "loss": 1.2242,
+      "step": 2734
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.714018501989661e-07,
+      "loss": 1.294,
+      "step": 2735
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.675367630937335e-07,
+      "loss": 1.3876,
+      "step": 2736
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.636824491024841e-07,
+      "loss": 1.35,
+      "step": 2737
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.598389126745209e-07,
+      "loss": 1.3246,
+      "step": 2738
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.560061582467059e-07,
+      "loss": 1.4282,
+      "step": 2739
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.521841902434567e-07,
+      "loss": 1.5409,
+      "step": 2740
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.483730130767374e-07,
+      "loss": 1.5143,
+      "step": 2741
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.445726311460553e-07,
+      "loss": 1.2164,
+      "step": 2742
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.407830488384581e-07,
+      "loss": 1.2928,
+      "step": 2743
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.370042705285229e-07,
+      "loss": 1.2581,
+      "step": 2744
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.332363005783581e-07,
+      "loss": 1.3291,
+      "step": 2745
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.29479143337598e-07,
+      "loss": 1.2346,
+      "step": 2746
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.257328031433851e-07,
+      "loss": 1.4231,
+      "step": 2747
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.21997284320387e-07,
+      "loss": 1.2647,
+      "step": 2748
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.182725911807718e-07,
+      "loss": 1.4215,
+      "step": 2749
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.145587280242138e-07,
+      "loss": 1.3959,
+      "step": 2750
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.108556991378834e-07,
+      "loss": 1.2339,
+      "step": 2751
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.071635087964456e-07,
+      "loss": 1.3744,
+      "step": 2752
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.034821612620534e-07,
+      "loss": 1.6327,
+      "step": 2753
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.99811660784344e-07,
+      "loss": 1.3937,
+      "step": 2754
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.961520116004326e-07,
+      "loss": 1.2855,
+      "step": 2755
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.925032179349066e-07,
+      "loss": 1.3606,
+      "step": 2756
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.888652839998232e-07,
+      "loss": 1.073,
+      "step": 2757
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.852382139947077e-07,
+      "loss": 1.3689,
+      "step": 2758
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.816220121065352e-07,
+      "loss": 1.4321,
+      "step": 2759
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.780166825097399e-07,
+      "loss": 1.2922,
+      "step": 2760
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.744222293662094e-07,
+      "loss": 1.3955,
+      "step": 2761
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.708386568252688e-07,
+      "loss": 1.3448,
+      "step": 2762
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.672659690236892e-07,
+      "loss": 1.5288,
+      "step": 2763
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.637041700856705e-07,
+      "loss": 1.0266,
+      "step": 2764
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.601532641228469e-07,
+      "loss": 1.5093,
+      "step": 2765
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.566132552342784e-07,
+      "loss": 1.3968,
+      "step": 2766
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.53084147506443e-07,
+      "loss": 1.7025,
+      "step": 2767
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.495659450132362e-07,
+      "loss": 1.4238,
+      "step": 2768
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.460586518159649e-07,
+      "loss": 1.5938,
+      "step": 2769
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.425622719633428e-07,
+      "loss": 1.4279,
+      "step": 2770
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.390768094914866e-07,
+      "loss": 1.3579,
+      "step": 2771
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.35602268423906e-07,
+      "loss": 1.3276,
+      "step": 2772
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.3213865277151e-07,
+      "loss": 1.3018,
+      "step": 2773
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.286859665325905e-07,
+      "loss": 1.2204,
+      "step": 2774
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.252442136928215e-07,
+      "loss": 1.3873,
+      "step": 2775
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.218133982252626e-07,
+      "loss": 1.3271,
+      "step": 2776
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.183935240903415e-07,
+      "loss": 1.3478,
+      "step": 2777
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.149845952358589e-07,
+      "loss": 1.293,
+      "step": 2778
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.115866155969796e-07,
+      "loss": 1.4468,
+      "step": 2779
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.081995890962266e-07,
+      "loss": 1.2268,
+      "step": 2780
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.048235196434836e-07,
+      "loss": 1.5364,
+      "step": 2781
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.014584111359811e-07,
+      "loss": 1.1762,
+      "step": 2782
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.981042674583014e-07,
+      "loss": 1.1931,
+      "step": 2783
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.947610924823643e-07,
+      "loss": 1.1653,
+      "step": 2784
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.914288900674313e-07,
+      "loss": 1.2873,
+      "step": 2785
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.881076640600979e-07,
+      "loss": 1.2991,
+      "step": 2786
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.847974182942839e-07,
+      "loss": 1.4948,
+      "step": 2787
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.814981565912391e-07,
+      "loss": 1.5929,
+      "step": 2788
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.782098827595305e-07,
+      "loss": 1.1272,
+      "step": 2789
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7493260059504497e-07,
+      "loss": 1.3009,
+      "step": 2790
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7166631388097515e-07,
+      "loss": 1.248,
+      "step": 2791
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.684110263878239e-07,
+      "loss": 1.4013,
+      "step": 2792
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.651667418733996e-07,
+      "loss": 1.2092,
+      "step": 2793
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6193346408280216e-07,
+      "loss": 1.4415,
+      "step": 2794
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.587111967484359e-07,
+      "loss": 1.3939,
+      "step": 2795
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.554999435899843e-07,
+      "loss": 1.0745,
+      "step": 2796
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5229970831442095e-07,
+      "loss": 1.2284,
+      "step": 2797
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.491104946160052e-07,
+      "loss": 1.5713,
+      "step": 2798
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4593230617626324e-07,
+      "loss": 1.4132,
+      "step": 2799
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.427651466640037e-07,
+      "loss": 1.2573,
+      "step": 2800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.396090197353009e-07,
+      "loss": 1.2896,
+      "step": 2801
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3646392903348823e-07,
+      "loss": 1.4793,
+      "step": 2802
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3332987818916814e-07,
+      "loss": 1.4177,
+      "step": 2803
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.302068708201912e-07,
+      "loss": 1.1994,
+      "step": 2804
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2709491053166465e-07,
+      "loss": 1.413,
+      "step": 2805
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2399400091594154e-07,
+      "loss": 1.2706,
+      "step": 2806
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.209041455526164e-07,
+      "loss": 1.4266,
+      "step": 2807
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.17825348008527e-07,
+      "loss": 1.3097,
+      "step": 2808
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.1475761183774275e-07,
+      "loss": 1.266,
+      "step": 2809
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.117009405815686e-07,
+      "loss": 1.3697,
+      "step": 2810
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0865533776852986e-07,
+      "loss": 1.3158,
+      "step": 2811
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0562080691438076e-07,
+      "loss": 1.2453,
+      "step": 2812
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.025973515220938e-07,
+      "loss": 1.4779,
+      "step": 2813
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9958497508185036e-07,
+      "loss": 1.2347,
+      "step": 2814
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9658368107104995e-07,
+      "loss": 1.348,
+      "step": 2815
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.935934729542956e-07,
+      "loss": 1.4117,
+      "step": 2816
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.906143541833918e-07,
+      "loss": 1.1606,
+      "step": 2817
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8764632819734526e-07,
+      "loss": 1.4968,
+      "step": 2818
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.846893984223543e-07,
+      "loss": 1.1265,
+      "step": 2819
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.817435682718096e-07,
+      "loss": 1.2827,
+      "step": 2820
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.788088411462887e-07,
+      "loss": 1.1688,
+      "step": 2821
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.758852204335539e-07,
+      "loss": 1.3651,
+      "step": 2822
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7297270950854224e-07,
+      "loss": 1.5693,
+      "step": 2823
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7007131173336766e-07,
+      "loss": 1.6414,
+      "step": 2824
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6718103045731977e-07,
+      "loss": 1.1036,
+      "step": 2825
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.643018690168487e-07,
+      "loss": 1.3781,
+      "step": 2826
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.614338307355747e-07,
+      "loss": 1.1516,
+      "step": 2827
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.585769189242716e-07,
+      "loss": 1.0591,
+      "step": 2828
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.557311368808747e-07,
+      "loss": 1.2362,
+      "step": 2829
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.5289648789046616e-07,
+      "loss": 1.2806,
+      "step": 2830
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.500729752252818e-07,
+      "loss": 1.3344,
+      "step": 2831
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.472606021446989e-07,
+      "loss": 1.2779,
+      "step": 2832
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4445937189523494e-07,
+      "loss": 1.0414,
+      "step": 2833
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4166928771054653e-07,
+      "loss": 1.28,
+      "step": 2834
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3889035281142293e-07,
+      "loss": 1.3941,
+      "step": 2835
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.361225704057813e-07,
+      "loss": 1.2626,
+      "step": 2836
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3336594368866825e-07,
+      "loss": 1.12,
+      "step": 2837
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3062047584224934e-07,
+      "loss": 1.5095,
+      "step": 2838
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2788617003580825e-07,
+      "loss": 0.9954,
+      "step": 2839
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2516302942574794e-07,
+      "loss": 1.2786,
+      "step": 2840
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2245105715557835e-07,
+      "loss": 1.346,
+      "step": 2841
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.197502563559185e-07,
+      "loss": 1.1548,
+      "step": 2842
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.170606301444912e-07,
+      "loss": 1.3541,
+      "step": 2843
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1438218162612057e-07,
+      "loss": 1.3184,
+      "step": 2844
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.117149138927267e-07,
+      "loss": 1.2862,
+      "step": 2845
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0905883002332213e-07,
+      "loss": 1.2123,
+      "step": 2846
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.064139330840099e-07,
+      "loss": 1.3282,
+      "step": 2847
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0378022612797875e-07,
+      "loss": 1.2676,
+      "step": 2848
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.011577121955034e-07,
+      "loss": 1.3765,
+      "step": 2849
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.985463943139322e-07,
+      "loss": 1.255,
+      "step": 2850
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.959462754976916e-07,
+      "loss": 1.2161,
+      "step": 2851
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9335735874828185e-07,
+      "loss": 1.3419,
+      "step": 2852
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.907796470542701e-07,
+      "loss": 1.4892,
+      "step": 2853
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.882131433912883e-07,
+      "loss": 1.4053,
+      "step": 2854
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.856578507220298e-07,
+      "loss": 1.4569,
+      "step": 2855
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8311377199624957e-07,
+      "loss": 1.4057,
+      "step": 2856
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.805809101507539e-07,
+      "loss": 1.1829,
+      "step": 2857
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7805926810940297e-07,
+      "loss": 1.2645,
+      "step": 2858
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.755488487831037e-07,
+      "loss": 1.3144,
+      "step": 2859
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7304965506980805e-07,
+      "loss": 1.3929,
+      "step": 2860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.705616898545105e-07,
+      "loss": 1.3418,
+      "step": 2861
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6808495600924355e-07,
+      "loss": 1.2462,
+      "step": 2862
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.656194563930714e-07,
+      "loss": 1.1697,
+      "step": 2863
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6316519385209627e-07,
+      "loss": 1.0823,
+      "step": 2864
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.60722171219443e-07,
+      "loss": 1.1808,
+      "step": 2865
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.582903913152612e-07,
+      "loss": 1.2288,
+      "step": 2866
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.558698569467266e-07,
+      "loss": 1.1951,
+      "step": 2867
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5346057090803065e-07,
+      "loss": 1.271,
+      "step": 2868
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.510625359803809e-07,
+      "loss": 1.3745,
+      "step": 2869
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4867575493199515e-07,
+      "loss": 1.3558,
+      "step": 2870
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.463002305181028e-07,
+      "loss": 1.5647,
+      "step": 2871
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.43935965480937e-07,
+      "loss": 1.2554,
+      "step": 2872
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4158296254973346e-07,
+      "loss": 1.1783,
+      "step": 2873
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.392412244407294e-07,
+      "loss": 1.4686,
+      "step": 2874
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3691075385715356e-07,
+      "loss": 1.2956,
+      "step": 2875
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.34591553489234e-07,
+      "loss": 1.3484,
+      "step": 2876
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3228362601418474e-07,
+      "loss": 1.3457,
+      "step": 2877
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2998697409620573e-07,
+      "loss": 1.5168,
+      "step": 2878
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2770160038648625e-07,
+      "loss": 1.4888,
+      "step": 2879
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.254275075231893e-07,
+      "loss": 1.4345,
+      "step": 2880
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.231646981314617e-07,
+      "loss": 1.1294,
+      "step": 2881
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2091317482342056e-07,
+      "loss": 1.1904,
+      "step": 2882
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1867294019815688e-07,
+      "loss": 1.238,
+      "step": 2883
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.164439968417298e-07,
+      "loss": 1.4471,
+      "step": 2884
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.142263473271644e-07,
+      "loss": 1.2423,
+      "step": 2885
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1201999421445074e-07,
+      "loss": 1.3565,
+      "step": 2886
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0982494005053256e-07,
+      "loss": 1.2758,
+      "step": 2887
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0764118736931406e-07,
+      "loss": 1.4338,
+      "step": 2888
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0546873869165762e-07,
+      "loss": 1.2584,
+      "step": 2889
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0330759652536835e-07,
+      "loss": 1.3589,
+      "step": 2890
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0115776336520622e-07,
+      "loss": 1.3055,
+      "step": 2891
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9901924169287268e-07,
+      "loss": 1.2429,
+      "step": 2892
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9689203397701084e-07,
+      "loss": 1.2976,
+      "step": 2893
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9477614267320867e-07,
+      "loss": 1.2024,
+      "step": 2894
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9267157022398452e-07,
+      "loss": 1.4007,
+      "step": 2895
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.905783190587951e-07,
+      "loss": 1.3061,
+      "step": 2896
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8849639159402522e-07,
+      "loss": 1.4823,
+      "step": 2897
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8642579023298913e-07,
+      "loss": 1.4686,
+      "step": 2898
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8436651736592704e-07,
+      "loss": 1.1814,
+      "step": 2899
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8231857537000075e-07,
+      "loss": 1.2503,
+      "step": 2900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8028196660929476e-07,
+      "loss": 1.3132,
+      "step": 2901
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7825669343480624e-07,
+      "loss": 1.377,
+      "step": 2902
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7624275818444947e-07,
+      "loss": 1.3826,
+      "step": 2903
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.742401631830526e-07,
+      "loss": 1.2218,
+      "step": 2904
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7224891074234972e-07,
+      "loss": 1.4471,
+      "step": 2905
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7026900316098217e-07,
+      "loss": 1.4364,
+      "step": 2906
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6830044272449609e-07,
+      "loss": 1.2391,
+      "step": 2907
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6634323170533928e-07,
+      "loss": 1.1006,
+      "step": 2908
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6439737236285447e-07,
+      "loss": 1.3736,
+      "step": 2909
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6246286694328594e-07,
+      "loss": 1.5759,
+      "step": 2910
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.605397176797663e-07,
+      "loss": 1.2885,
+      "step": 2911
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5862792679232074e-07,
+      "loss": 1.2768,
+      "step": 2912
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5672749648786511e-07,
+      "loss": 1.2014,
+      "step": 2913
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5483842896019675e-07,
+      "loss": 1.1765,
+      "step": 2914
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5296072638999904e-07,
+      "loss": 1.4009,
+      "step": 2915
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.510943909448359e-07,
+      "loss": 1.2554,
+      "step": 2916
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.492394247791473e-07,
+      "loss": 1.362,
+      "step": 2917
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.473958300342504e-07,
+      "loss": 1.3752,
+      "step": 2918
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.455636088383372e-07,
+      "loss": 1.4093,
+      "step": 2919
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4374276330646587e-07,
+      "loss": 1.1977,
+      "step": 2920
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4193329554056834e-07,
+      "loss": 1.3186,
+      "step": 2921
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.401352076294371e-07,
+      "loss": 1.3645,
+      "step": 2922
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3834850164873182e-07,
+      "loss": 1.177,
+      "step": 2923
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.365731796609715e-07,
+      "loss": 1.4222,
+      "step": 2924
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.348092437155346e-07,
+      "loss": 1.4146,
+      "step": 2925
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3305669584865565e-07,
+      "loss": 1.1082,
+      "step": 2926
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.313155380834219e-07,
+      "loss": 1.3801,
+      "step": 2927
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2958577242977666e-07,
+      "loss": 1.2997,
+      "step": 2928
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2786740088450488e-07,
+      "loss": 1.3609,
+      "step": 2929
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.261604254312454e-07,
+      "loss": 1.3902,
+      "step": 2930
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2446484804047754e-07,
+      "loss": 1.3799,
+      "step": 2931
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.227806706695278e-07,
+      "loss": 1.3723,
+      "step": 2932
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2110789526255662e-07,
+      "loss": 1.1631,
+      "step": 2933
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1944652375056597e-07,
+      "loss": 1.4594,
+      "step": 2934
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.17796558051394e-07,
+      "loss": 1.1806,
+      "step": 2935
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1615800006971045e-07,
+      "loss": 1.1607,
+      "step": 2936
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1453085169701782e-07,
+      "loss": 1.3029,
+      "step": 2937
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1291511481164807e-07,
+      "loss": 1.1731,
+      "step": 2938
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1131079127875588e-07,
+      "loss": 1.3215,
+      "step": 2939
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.097178829503276e-07,
+      "loss": 1.5948,
+      "step": 2940
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0813639166516677e-07,
+      "loss": 1.152,
+      "step": 2941
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0656631924889749e-07,
+      "loss": 1.3328,
+      "step": 2942
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0500766751396774e-07,
+      "loss": 1.0938,
+      "step": 2943
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0346043825963603e-07,
+      "loss": 1.3495,
+      "step": 2944
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0192463327197699e-07,
+      "loss": 1.2589,
+      "step": 2945
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0040025432387801e-07,
+      "loss": 1.3637,
+      "step": 2946
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.888730317503592e-08,
+      "loss": 1.2074,
+      "step": 2947
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.738578157195588e-08,
+      "loss": 1.189,
+      "step": 2948
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.589569124794918e-08,
+      "loss": 1.4438,
+      "step": 2949
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.441703392313095e-08,
+      "loss": 1.3176,
+      "step": 2950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.294981130441915e-08,
+      "loss": 1.2371,
+      "step": 2951
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.149402508553007e-08,
+      "loss": 1.3817,
+      "step": 2952
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.004967694698164e-08,
+      "loss": 1.3131,
+      "step": 2953
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.861676855608237e-08,
+      "loss": 1.2445,
+      "step": 2954
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.719530156694245e-08,
+      "loss": 1.5249,
+      "step": 2955
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.578527762045818e-08,
+      "loss": 1.5001,
+      "step": 2956
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.438669834431756e-08,
+      "loss": 1.038,
+      "step": 2957
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.299956535300135e-08,
+      "loss": 1.3638,
+      "step": 2958
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.162388024777202e-08,
+      "loss": 1.3877,
+      "step": 2959
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.025964461667924e-08,
+      "loss": 1.4172,
+      "step": 2960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.89068600345555e-08,
+      "loss": 1.2632,
+      "step": 2961
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.756552806301498e-08,
+      "loss": 1.2799,
+      "step": 2962
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.623565025045021e-08,
+      "loss": 1.451,
+      "step": 2963
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.491722813203206e-08,
+      "loss": 1.5416,
+      "step": 2964
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.361026322970866e-08,
+      "loss": 1.2674,
+      "step": 2965
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.23147570521987e-08,
+      "loss": 1.2977,
+      "step": 2966
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.103071109499482e-08,
+      "loss": 1.2161,
+      "step": 2967
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.975812684036464e-08,
+      "loss": 1.358,
+      "step": 2968
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.849700575733753e-08,
+      "loss": 1.4869,
+      "step": 2969
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.724734930171561e-08,
+      "loss": 1.2691,
+      "step": 2970
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.600915891606385e-08,
+      "loss": 1.3743,
+      "step": 2971
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.478243602971223e-08,
+      "loss": 1.3749,
+      "step": 2972
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.356718205875246e-08,
+      "loss": 1.3831,
+      "step": 2973
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.236339840603677e-08,
+      "loss": 1.3496,
+      "step": 2974
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.117108646117587e-08,
+      "loss": 1.2957,
+      "step": 2975
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.999024760054095e-08,
+      "loss": 1.5453,
+      "step": 2976
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8820883187254983e-08,
+      "loss": 1.1694,
+      "step": 2977
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.766299457119817e-08,
+      "loss": 1.3449,
+      "step": 2978
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.651658308900021e-08,
+      "loss": 1.0801,
+      "step": 2979
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5381650064045835e-08,
+      "loss": 1.3216,
+      "step": 2980
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.425819680646704e-08,
+      "loss": 1.3303,
+      "step": 2981
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3146224613144225e-08,
+      "loss": 1.4011,
+      "step": 2982
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.2045734767705026e-08,
+      "loss": 1.1832,
+      "step": 2983
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.095672854051992e-08,
+      "loss": 1.4232,
+      "step": 2984
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.987920718870665e-08,
+      "loss": 1.4733,
+      "step": 2985
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8813171956123565e-08,
+      "loss": 1.4079,
+      "step": 2986
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.77586240733674e-08,
+      "loss": 1.4166,
+      "step": 2987
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.671556475777883e-08,
+      "loss": 1.2354,
+      "step": 2988
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5683995213433584e-08,
+      "loss": 1.278,
+      "step": 2989
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4663916631143554e-08,
+      "loss": 1.3097,
+      "step": 2990
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3655330188455694e-08,
+      "loss": 1.2042,
+      "step": 2991
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2658237049655325e-08,
+      "loss": 1.1266,
+      "step": 2992
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.167263836575286e-08,
+      "loss": 1.2124,
+      "step": 2993
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.069853527449596e-08,
+      "loss": 1.2365,
+      "step": 2994
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9735928900359597e-08,
+      "loss": 1.5254,
+      "step": 2995
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8784820354546004e-08,
+      "loss": 1.3922,
+      "step": 2996
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.784521073498804e-08,
+      "loss": 1.1821,
+      "step": 2997
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.691710112634139e-08,
+      "loss": 1.1222,
+      "step": 2998
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6000492599989054e-08,
+      "loss": 1.488,
+      "step": 2999
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.509538621403463e-08,
+      "loss": 1.1743,
+      "step": 3000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.42017830133079e-08,
+      "loss": 1.3106,
+      "step": 3001
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3319684029354815e-08,
+      "loss": 1.4583,
+      "step": 3002
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2449090280445294e-08,
+      "loss": 1.3686,
+      "step": 3003
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.159000277156654e-08,
+      "loss": 1.1827,
+      "step": 3004
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.074242249442305e-08,
+      "loss": 1.1919,
+      "step": 3005
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9906350427435505e-08,
+      "loss": 1.4537,
+      "step": 3006
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9081787535741867e-08,
+      "loss": 1.291,
+      "step": 3007
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.826873477118963e-08,
+      "loss": 1.4726,
+      "step": 3008
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7467193072345798e-08,
+      "loss": 1.2885,
+      "step": 3009
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.667716336448356e-08,
+      "loss": 1.6707,
+      "step": 3010
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.589864655959007e-08,
+      "loss": 1.1435,
+      "step": 3011
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5131643556361996e-08,
+      "loss": 1.3806,
+      "step": 3012
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4376155240203313e-08,
+      "loss": 1.2776,
+      "step": 3013
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3632182483228628e-08,
+      "loss": 1.4064,
+      "step": 3014
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2899726144256506e-08,
+      "loss": 1.2568,
+      "step": 3015
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2178787068812822e-08,
+      "loss": 1.4361,
+      "step": 3016
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.146936608912853e-08,
+      "loss": 1.206,
+      "step": 3017
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.077146402413521e-08,
+      "loss": 1.164,
+      "step": 3018
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0085081679473982e-08,
+      "loss": 1.5211,
+      "step": 3019
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9410219847479926e-08,
+      "loss": 1.43,
+      "step": 3020
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.874687930719543e-08,
+      "loss": 1.3033,
+      "step": 3021
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.80950608243613e-08,
+      "loss": 1.479,
+      "step": 3022
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7454765151415644e-08,
+      "loss": 1.2778,
+      "step": 3023
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6825993027497213e-08,
+      "loss": 1.2988,
+      "step": 3024
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6208745178440955e-08,
+      "loss": 1.4242,
+      "step": 3025
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5603022316780235e-08,
+      "loss": 1.204,
+      "step": 3026
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.50088251417424e-08,
+      "loss": 1.2815,
+      "step": 3027
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4426154339250986e-08,
+      "loss": 1.358,
+      "step": 3028
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3855010581922402e-08,
+      "loss": 1.28,
+      "step": 3029
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.329539452907036e-08,
+      "loss": 1.3935,
+      "step": 3030
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2747306826697003e-08,
+      "loss": 1.3187,
+      "step": 3031
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2210748107498449e-08,
+      "loss": 1.2983,
+      "step": 3032
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1685718990861462e-08,
+      "loss": 1.481,
+      "step": 3033
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.117222008286456e-08,
+      "loss": 1.4963,
+      "step": 3034
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0670251976275803e-08,
+      "loss": 1.1308,
+      "step": 3035
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0179815250550562e-08,
+      "loss": 1.1673,
+      "step": 3036
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.700910471835967e-09,
+      "loss": 1.3461,
+      "step": 3037
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.233538192963132e-09,
+      "loss": 1.2813,
+      "step": 3038
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.777698953454927e-09,
+      "loss": 1.3723,
+      "step": 3039
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.33339327951599e-09,
+      "loss": 1.5213,
+      "step": 3040
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.900621684041599e-09,
+      "loss": 1.2869,
+      "step": 3041
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.479384666608802e-09,
+      "loss": 1.137,
+      "step": 3042
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.069682713480852e-09,
+      "loss": 1.0495,
+      "step": 3043
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.671516297606095e-09,
+      "loss": 1.2672,
+      "step": 3044
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.284885878615754e-09,
+      "loss": 1.1805,
+      "step": 3045
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.909791902823925e-09,
+      "loss": 1.3118,
+      "step": 3046
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.546234803227579e-09,
+      "loss": 1.3739,
+      "step": 3047
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.1942149995065635e-09,
+      "loss": 1.3098,
+      "step": 3048
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.85373289802249e-09,
+      "loss": 1.2643,
+      "step": 3049
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.524788891816512e-09,
+      "loss": 1.2796,
+      "step": 3050
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.20738336061155e-09,
+      "loss": 1.5313,
+      "step": 3051
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.901516670811178e-09,
+      "loss": 1.1196,
+      "step": 3052
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6071891754996258e-09,
+      "loss": 1.1904,
+      "step": 3053
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3244012144395545e-09,
+      "loss": 1.5327,
+      "step": 3054
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0531531140709503e-09,
+      "loss": 1.1975,
+      "step": 3055
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7934451875166747e-09,
+      "loss": 1.5974,
+      "step": 3056
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5452777345746915e-09,
+      "loss": 1.3612,
+      "step": 3057
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3086510417225093e-09,
+      "loss": 1.3328,
+      "step": 3058
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0835653821138503e-09,
+      "loss": 1.4179,
+      "step": 3059
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8700210155819797e-09,
+      "loss": 1.3455,
+      "step": 3060
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6680181886352676e-09,
+      "loss": 1.2922,
+      "step": 3061
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4775571344605167e-09,
+      "loss": 1.3924,
+      "step": 3062
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2986380729185232e-09,
+      "loss": 1.3837,
+      "step": 3063
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1312612105507382e-09,
+      "loss": 1.4901,
+      "step": 3064
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.754267405692741e-10,
+      "loss": 1.1579,
+      "step": 3065
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.311348428657884e-10,
+      "loss": 1.2489,
+      "step": 3066
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.983856840070413e-10,
+      "loss": 1.5465,
+      "step": 3067
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.771794172348966e-10,
+      "loss": 1.3687,
+      "step": 3068
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.675161824641006e-10,
+      "loss": 1.3581,
+      "step": 3069
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.6939610628894396e-10,
+      "loss": 1.664,
+      "step": 3070
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.828193019754899e-10,
+      "loss": 1.432,
+      "step": 3071
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0778586946490487e-10,
+      "loss": 1.2313,
+      "step": 3072
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4429589537345855e-10,
+      "loss": 1.5874,
+      "step": 3073
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.234945299363418e-11,
+      "loss": 1.4343,
+      "step": 3074
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.194660228857729e-11,
+      "loss": 1.3749,
+      "step": 3075
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.3087389899867364e-11,
+      "loss": 1.2478,
+      "step": 3076
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.7718491408564895e-12,
+      "loss": 1.2528,
+      "step": 3077
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 1.2187,
+      "step": 3078
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3078,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 4.144027688224424e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}