diff --git "a/checkpoint-6925/trainer_state.json" "b/checkpoint-6925/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-6925/trainer_state.json"
@@ -0,0 +1,41791 @@
+{
+  "best_metric": 3.3698325157165527,
+  "best_model_checkpoint": "bird_sounds_classification/checkpoint-6925",
+  "epoch": 25.0,
+  "global_step": 6925,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 3.9265,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.9201,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.9157,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.9419,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.9165,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.9169,
+      "step": 6
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.9317,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.8781,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.9222,
+      "step": 9
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.9165,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1e-06,
+      "loss": 3.9313,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 3.9205,
+      "step": 12
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3e-06,
+      "loss": 3.8988,
+      "step": 13
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.9248,
+      "step": 14
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5e-06,
+      "loss": 3.9184,
+      "step": 15
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 3.9186,
+      "step": 16
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 3.9203,
+      "step": 17
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 3.9052,
+      "step": 18
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 3.9009,
+      "step": 19
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.8969,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 3.8832,
+      "step": 21
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2e-06,
+      "loss": 3.8743,
+      "step": 22
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 3.9247,
+      "step": 23
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.8468,
+      "step": 24
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5e-06,
+      "loss": 3.9204,
+      "step": 25
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.6e-06,
+      "loss": 3.9259,
+      "step": 26
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 3.9085,
+      "step": 27
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 3.9264,
+      "step": 28
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9e-06,
+      "loss": 3.9245,
+      "step": 29
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3e-06,
+      "loss": 3.9032,
+      "step": 30
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 3.9212,
+      "step": 31
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 3.9171,
+      "step": 32
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 3.9293,
+      "step": 33
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 3.8338,
+      "step": 34
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5e-06,
+      "loss": 3.8863,
+      "step": 35
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 3.9327,
+      "step": 36
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.7e-06,
+      "loss": 3.927,
+      "step": 37
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 3.9278,
+      "step": 38
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 3.9291,
+      "step": 39
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.9077,
+      "step": 40
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.1e-06,
+      "loss": 3.865,
+      "step": 41
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 3.8786,
+      "step": 42
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3e-06,
+      "loss": 3.9032,
+      "step": 43
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4e-06,
+      "loss": 3.9205,
+      "step": 44
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5e-06,
+      "loss": 3.933,
+      "step": 45
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 3.894,
+      "step": 46
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7e-06,
+      "loss": 3.898,
+      "step": 47
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.0067,
+      "step": 48
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 3.9013,
+      "step": 49
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5e-06,
+      "loss": 3.9172,
+      "step": 50
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999272727272727e-06,
+      "loss": 3.902,
+      "step": 51
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.998545454545455e-06,
+      "loss": 3.91,
+      "step": 52
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.997818181818182e-06,
+      "loss": 3.9242,
+      "step": 53
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9970909090909094e-06,
+      "loss": 3.9252,
+      "step": 54
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.996363636363637e-06,
+      "loss": 3.9287,
+      "step": 55
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.995636363636364e-06,
+      "loss": 3.9083,
+      "step": 56
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.994909090909092e-06,
+      "loss": 3.854,
+      "step": 57
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9941818181818185e-06,
+      "loss": 3.9083,
+      "step": 58
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.993454545454546e-06,
+      "loss": 3.915,
+      "step": 59
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.992727272727273e-06,
+      "loss": 3.8907,
+      "step": 60
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.992e-06,
+      "loss": 3.8879,
+      "step": 61
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.991272727272728e-06,
+      "loss": 3.8331,
+      "step": 62
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.990545454545455e-06,
+      "loss": 3.9241,
+      "step": 63
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.989818181818182e-06,
+      "loss": 3.914,
+      "step": 64
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.98909090909091e-06,
+      "loss": 3.8941,
+      "step": 65
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9883636363636365e-06,
+      "loss": 3.9278,
+      "step": 66
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.987636363636364e-06,
+      "loss": 3.9053,
+      "step": 67
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.986909090909091e-06,
+      "loss": 3.9214,
+      "step": 68
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.986181818181819e-06,
+      "loss": 3.9208,
+      "step": 69
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.985454545454546e-06,
+      "loss": 3.9192,
+      "step": 70
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.984727272727273e-06,
+      "loss": 3.8952,
+      "step": 71
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.984000000000001e-06,
+      "loss": 3.9311,
+      "step": 72
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.983272727272728e-06,
+      "loss": 3.9254,
+      "step": 73
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9825454545454546e-06,
+      "loss": 3.9017,
+      "step": 74
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.981818181818182e-06,
+      "loss": 3.922,
+      "step": 75
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.981090909090909e-06,
+      "loss": 3.9247,
+      "step": 76
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.980363636363637e-06,
+      "loss": 3.9199,
+      "step": 77
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9796363636363644e-06,
+      "loss": 3.9254,
+      "step": 78
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.978909090909091e-06,
+      "loss": 3.8985,
+      "step": 79
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.978181818181819e-06,
+      "loss": 3.9184,
+      "step": 80
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.977454545454546e-06,
+      "loss": 3.9101,
+      "step": 81
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9767272727272735e-06,
+      "loss": 3.9187,
+      "step": 82
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.976e-06,
+      "loss": 3.9222,
+      "step": 83
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.975272727272727e-06,
+      "loss": 3.9222,
+      "step": 84
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.974545454545455e-06,
+      "loss": 3.9381,
+      "step": 85
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9738181818181825e-06,
+      "loss": 3.8843,
+      "step": 86
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.973090909090909e-06,
+      "loss": 3.896,
+      "step": 87
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.972363636363637e-06,
+      "loss": 3.9127,
+      "step": 88
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.971636363636364e-06,
+      "loss": 3.9317,
+      "step": 89
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9709090909090915e-06,
+      "loss": 3.9047,
+      "step": 90
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.970181818181818e-06,
+      "loss": 3.9127,
+      "step": 91
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.969454545454546e-06,
+      "loss": 3.908,
+      "step": 92
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.968727272727274e-06,
+      "loss": 3.9101,
+      "step": 93
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9680000000000005e-06,
+      "loss": 3.8955,
+      "step": 94
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.967272727272727e-06,
+      "loss": 3.9286,
+      "step": 95
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.966545454545455e-06,
+      "loss": 3.9202,
+      "step": 96
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.965818181818182e-06,
+      "loss": 3.9166,
+      "step": 97
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9650909090909096e-06,
+      "loss": 3.9435,
+      "step": 98
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.964363636363636e-06,
+      "loss": 3.9178,
+      "step": 99
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.963636363636364e-06,
+      "loss": 3.8977,
+      "step": 100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.962909090909092e-06,
+      "loss": 3.9095,
+      "step": 101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.962181818181819e-06,
+      "loss": 3.9003,
+      "step": 102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.961454545454546e-06,
+      "loss": 3.9009,
+      "step": 103
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.960727272727273e-06,
+      "loss": 3.9072,
+      "step": 104
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 3.9061,
+      "step": 105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.959272727272728e-06,
+      "loss": 3.9149,
+      "step": 106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9585454545454545e-06,
+      "loss": 3.9218,
+      "step": 107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.957818181818182e-06,
+      "loss": 3.9295,
+      "step": 108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.95709090909091e-06,
+      "loss": 3.8893,
+      "step": 109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.956363636363637e-06,
+      "loss": 3.9168,
+      "step": 110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.955636363636364e-06,
+      "loss": 3.8953,
+      "step": 111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.954909090909091e-06,
+      "loss": 3.9087,
+      "step": 112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.954181818181819e-06,
+      "loss": 3.9077,
+      "step": 113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.953454545454546e-06,
+      "loss": 3.9165,
+      "step": 114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.952727272727273e-06,
+      "loss": 3.8944,
+      "step": 115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.952e-06,
+      "loss": 3.9315,
+      "step": 116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.951272727272728e-06,
+      "loss": 3.9167,
+      "step": 117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.950545454545455e-06,
+      "loss": 3.8719,
+      "step": 118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.949818181818182e-06,
+      "loss": 3.8878,
+      "step": 119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.949090909090909e-06,
+      "loss": 3.8811,
+      "step": 120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.948363636363637e-06,
+      "loss": 3.9155,
+      "step": 121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.947636363636364e-06,
+      "loss": 3.9106,
+      "step": 122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.946909090909091e-06,
+      "loss": 3.9236,
+      "step": 123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.946181818181819e-06,
+      "loss": 3.8969,
+      "step": 124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.945454545454546e-06,
+      "loss": 3.9314,
+      "step": 125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.944727272727274e-06,
+      "loss": 3.8898,
+      "step": 126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9440000000000004e-06,
+      "loss": 3.8947,
+      "step": 127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.943272727272727e-06,
+      "loss": 3.8962,
+      "step": 128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.942545454545455e-06,
+      "loss": 3.9269,
+      "step": 129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.941818181818182e-06,
+      "loss": 3.9115,
+      "step": 130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9410909090909095e-06,
+      "loss": 3.9276,
+      "step": 131
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.940363636363637e-06,
+      "loss": 3.8972,
+      "step": 132
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.939636363636364e-06,
+      "loss": 3.9359,
+      "step": 133
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.938909090909092e-06,
+      "loss": 3.9165,
+      "step": 134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9381818181818185e-06,
+      "loss": 3.9054,
+      "step": 135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.937454545454546e-06,
+      "loss": 3.9232,
+      "step": 136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.936727272727273e-06,
+      "loss": 3.8697,
+      "step": 137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936e-06,
+      "loss": 3.9301,
+      "step": 138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9352727272727275e-06,
+      "loss": 3.9063,
+      "step": 139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.934545454545455e-06,
+      "loss": 3.921,
+      "step": 140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933818181818182e-06,
+      "loss": 3.8996,
+      "step": 141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93309090909091e-06,
+      "loss": 3.9248,
+      "step": 142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9323636363636365e-06,
+      "loss": 3.8895,
+      "step": 143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.931636363636364e-06,
+      "loss": 3.9256,
+      "step": 144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930909090909091e-06,
+      "loss": 3.8939,
+      "step": 145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.930181818181819e-06,
+      "loss": 3.9251,
+      "step": 146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9294545454545456e-06,
+      "loss": 3.9231,
+      "step": 147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.928727272727273e-06,
+      "loss": 3.9121,
+      "step": 148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.928000000000001e-06,
+      "loss": 3.9261,
+      "step": 149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.927272727272728e-06,
+      "loss": 3.8999,
+      "step": 150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.926545454545455e-06,
+      "loss": 3.9201,
+      "step": 151
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.925818181818182e-06,
+      "loss": 3.9024,
+      "step": 152
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.925090909090909e-06,
+      "loss": 3.9118,
+      "step": 153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.924363636363637e-06,
+      "loss": 3.9047,
+      "step": 154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.923636363636364e-06,
+      "loss": 3.9042,
+      "step": 155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.922909090909091e-06,
+      "loss": 3.9365,
+      "step": 156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.922181818181819e-06,
+      "loss": 3.9004,
+      "step": 157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.921454545454546e-06,
+      "loss": 3.9051,
+      "step": 158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9207272727272735e-06,
+      "loss": 3.9094,
+      "step": 159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.92e-06,
+      "loss": 3.9147,
+      "step": 160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.919272727272727e-06,
+      "loss": 3.9373,
+      "step": 161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.918545454545455e-06,
+      "loss": 3.9201,
+      "step": 162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9178181818181825e-06,
+      "loss": 3.9037,
+      "step": 163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.917090909090909e-06,
+      "loss": 3.8845,
+      "step": 164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.916363636363637e-06,
+      "loss": 3.9335,
+      "step": 165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.915636363636364e-06,
+      "loss": 3.9109,
+      "step": 166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9149090909090915e-06,
+      "loss": 3.8829,
+      "step": 167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.914181818181818e-06,
+      "loss": 3.9118,
+      "step": 168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.913454545454546e-06,
+      "loss": 3.8975,
+      "step": 169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.912727272727273e-06,
+      "loss": 3.9051,
+      "step": 170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9120000000000006e-06,
+      "loss": 3.8961,
+      "step": 171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.911272727272728e-06,
+      "loss": 3.9022,
+      "step": 172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.910545454545455e-06,
+      "loss": 3.9115,
+      "step": 173
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.909818181818182e-06,
+      "loss": 3.9039,
+      "step": 174
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.90909090909091e-06,
+      "loss": 3.9303,
+      "step": 175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9083636363636364e-06,
+      "loss": 3.9203,
+      "step": 176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.907636363636364e-06,
+      "loss": 3.9135,
+      "step": 177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.906909090909091e-06,
+      "loss": 3.9449,
+      "step": 178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.906181818181819e-06,
+      "loss": 3.9068,
+      "step": 179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.905454545454546e-06,
+      "loss": 3.9283,
+      "step": 180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.904727272727273e-06,
+      "loss": 3.9375,
+      "step": 181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.904000000000001e-06,
+      "loss": 3.903,
+      "step": 182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.903272727272728e-06,
+      "loss": 3.9362,
+      "step": 183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.9025454545454545e-06,
+      "loss": 3.9123,
+      "step": 184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.901818181818182e-06,
+      "loss": 3.9127,
+      "step": 185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.901090909090909e-06,
+      "loss": 3.9145,
+      "step": 186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900363636363637e-06,
+      "loss": 3.8922,
+      "step": 187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.899636363636364e-06,
+      "loss": 4.0623,
+      "step": 188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898909090909091e-06,
+      "loss": 3.8935,
+      "step": 189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.898181818181819e-06,
+      "loss": 3.9204,
+      "step": 190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.897454545454546e-06,
+      "loss": 3.9001,
+      "step": 191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.896727272727273e-06,
+      "loss": 3.9173,
+      "step": 192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.896e-06,
+      "loss": 3.9123,
+      "step": 193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.895272727272727e-06,
+      "loss": 3.9002,
+      "step": 194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.894545454545455e-06,
+      "loss": 3.9245,
+      "step": 195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.893818181818182e-06,
+      "loss": 3.9073,
+      "step": 196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.893090909090909e-06,
+      "loss": 3.9074,
+      "step": 197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.892363636363637e-06,
+      "loss": 3.9098,
+      "step": 198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.891636363636364e-06,
+      "loss": 3.9156,
+      "step": 199
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8909090909090914e-06,
+      "loss": 3.9249,
+      "step": 200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.890181818181818e-06,
+      "loss": 3.9204,
+      "step": 201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.889454545454546e-06,
+      "loss": 3.9128,
+      "step": 202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.888727272727274e-06,
+      "loss": 3.923,
+      "step": 203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8880000000000005e-06,
+      "loss": 3.9127,
+      "step": 204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.887272727272728e-06,
+      "loss": 3.9147,
+      "step": 205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.886545454545455e-06,
+      "loss": 3.921,
+      "step": 206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.885818181818182e-06,
+      "loss": 3.9135,
+      "step": 207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8850909090909095e-06,
+      "loss": 3.9236,
+      "step": 208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.884363636363636e-06,
+      "loss": 3.9031,
+      "step": 209
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.883636363636364e-06,
+      "loss": 3.9285,
+      "step": 210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.882909090909092e-06,
+      "loss": 3.908,
+      "step": 211
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8821818181818185e-06,
+      "loss": 3.9109,
+      "step": 212
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.881454545454546e-06,
+      "loss": 3.921,
+      "step": 213
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.880727272727273e-06,
+      "loss": 3.8796,
+      "step": 214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 3.9159,
+      "step": 215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8792727272727275e-06,
+      "loss": 3.9119,
+      "step": 216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.878545454545454e-06,
+      "loss": 3.9247,
+      "step": 217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.877818181818182e-06,
+      "loss": 3.9147,
+      "step": 218
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.87709090909091e-06,
+      "loss": 3.9055,
+      "step": 219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8763636363636366e-06,
+      "loss": 3.9062,
+      "step": 220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.875636363636364e-06,
+      "loss": 3.9047,
+      "step": 221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.874909090909091e-06,
+      "loss": 3.9217,
+      "step": 222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.874181818181819e-06,
+      "loss": 3.9084,
+      "step": 223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.873454545454546e-06,
+      "loss": 3.922,
+      "step": 224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.872727272727273e-06,
+      "loss": 3.8994,
+      "step": 225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.872000000000001e-06,
+      "loss": 3.9275,
+      "step": 226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.871272727272728e-06,
+      "loss": 3.9015,
+      "step": 227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.870545454545455e-06,
+      "loss": 3.9174,
+      "step": 228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.869818181818182e-06,
+      "loss": 3.8829,
+      "step": 229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.869090909090909e-06,
+      "loss": 3.9242,
+      "step": 230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.868363636363637e-06,
+      "loss": 3.895,
+      "step": 231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.867636363636364e-06,
+      "loss": 3.9128,
+      "step": 232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.866909090909091e-06,
+      "loss": 3.9043,
+      "step": 233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.866181818181819e-06,
+      "loss": 3.9037,
+      "step": 234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.865454545454546e-06,
+      "loss": 3.9127,
+      "step": 235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8647272727272735e-06,
+      "loss": 3.9235,
+      "step": 236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.864e-06,
+      "loss": 4.0046,
+      "step": 237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.863272727272727e-06,
+      "loss": 3.9057,
+      "step": 238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.862545454545455e-06,
+      "loss": 3.9166,
+      "step": 239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.861818181818182e-06,
+      "loss": 3.9028,
+      "step": 240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.861090909090909e-06,
+      "loss": 3.9073,
+      "step": 241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.860363636363637e-06,
+      "loss": 3.9031,
+      "step": 242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.859636363636364e-06,
+      "loss": 3.908,
+      "step": 243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8589090909090916e-06,
+      "loss": 3.9138,
+      "step": 244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.858181818181818e-06,
+      "loss": 3.9227,
+      "step": 245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.857454545454546e-06,
+      "loss": 3.8929,
+      "step": 246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.856727272727273e-06,
+      "loss": 3.9067,
+      "step": 247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.856e-06,
+      "loss": 3.9126,
+      "step": 248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.855272727272728e-06,
+      "loss": 3.8889,
+      "step": 249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.854545454545455e-06,
+      "loss": 3.9037,
+      "step": 250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.853818181818182e-06,
+      "loss": 3.9124,
+      "step": 251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.85309090909091e-06,
+      "loss": 3.8984,
+      "step": 252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8523636363636365e-06,
+      "loss": 3.9069,
+      "step": 253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.851636363636364e-06,
+      "loss": 3.8829,
+      "step": 254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.850909090909091e-06,
+      "loss": 3.9294,
+      "step": 255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.850181818181819e-06,
+      "loss": 3.9118,
+      "step": 256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.849454545454546e-06,
+      "loss": 3.906,
+      "step": 257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848727272727273e-06,
+      "loss": 3.887,
+      "step": 258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.848000000000001e-06,
+      "loss": 3.9136,
+      "step": 259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847272727272728e-06,
+      "loss": 3.9012,
+      "step": 260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8465454545454545e-06,
+      "loss": 3.9209,
+      "step": 261
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.845818181818182e-06,
+      "loss": 3.9256,
+      "step": 262
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.845090909090909e-06,
+      "loss": 3.9068,
+      "step": 263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.844363636363637e-06,
+      "loss": 3.9123,
+      "step": 264
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.843636363636364e-06,
+      "loss": 3.9265,
+      "step": 265
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.842909090909091e-06,
+      "loss": 3.9042,
+      "step": 266
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.842181818181819e-06,
+      "loss": 3.9004,
+      "step": 267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.841454545454546e-06,
+      "loss": 3.8786,
+      "step": 268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.840727272727273e-06,
+      "loss": 3.9081,
+      "step": 269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.84e-06,
+      "loss": 3.9049,
+      "step": 270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.839272727272727e-06,
+      "loss": 3.8521,
+      "step": 271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.838545454545456e-06,
+      "loss": 3.9125,
+      "step": 272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8378181818181824e-06,
+      "loss": 3.9186,
+      "step": 273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.837090909090909e-06,
+      "loss": 3.9307,
+      "step": 274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.836363636363637e-06,
+      "loss": 3.902,
+      "step": 275
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.835636363636364e-06,
+      "loss": 3.9042,
+      "step": 276
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8349090909090915e-06,
+      "loss": 3.8846,
+      "step": 277
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.02972972972972973,
+      "eval_loss": 3.9061622619628906,
+      "eval_runtime": 142.8246,
+      "eval_samples_per_second": 2.591,
+      "eval_steps_per_second": 0.651,
+      "step": 277
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.834181818181818e-06,
+      "loss": 3.909,
+      "step": 278
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.833454545454546e-06,
+      "loss": 3.8994,
+      "step": 279
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.832727272727274e-06,
+      "loss": 3.9185,
+      "step": 280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8320000000000005e-06,
+      "loss": 3.9028,
+      "step": 281
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.831272727272728e-06,
+      "loss": 3.8976,
+      "step": 282
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830545454545455e-06,
+      "loss": 3.913,
+      "step": 283
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.829818181818182e-06,
+      "loss": 3.9109,
+      "step": 284
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8290909090909095e-06,
+      "loss": 3.91,
+      "step": 285
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828363636363636e-06,
+      "loss": 3.8683,
+      "step": 286
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.827636363636364e-06,
+      "loss": 3.8721,
+      "step": 287
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826909090909092e-06,
+      "loss": 3.9038,
+      "step": 288
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8261818181818185e-06,
+      "loss": 3.9525,
+      "step": 289
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.825454545454546e-06,
+      "loss": 3.8995,
+      "step": 290
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.824727272727273e-06,
+      "loss": 3.8952,
+      "step": 291
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.824000000000001e-06,
+      "loss": 3.9123,
+      "step": 292
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8232727272727276e-06,
+      "loss": 3.9209,
+      "step": 293
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.822545454545454e-06,
+      "loss": 3.8994,
+      "step": 294
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.821818181818182e-06,
+      "loss": 3.9025,
+      "step": 295
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.82109090909091e-06,
+      "loss": 3.9145,
+      "step": 296
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.820363636363637e-06,
+      "loss": 3.9085,
+      "step": 297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.819636363636364e-06,
+      "loss": 3.9048,
+      "step": 298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.818909090909091e-06,
+      "loss": 3.9095,
+      "step": 299
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.818181818181819e-06,
+      "loss": 3.9011,
+      "step": 300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.817454545454546e-06,
+      "loss": 3.8569,
+      "step": 301
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.816727272727273e-06,
+      "loss": 3.9222,
+      "step": 302
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.816e-06,
+      "loss": 3.8826,
+      "step": 303
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.815272727272728e-06,
+      "loss": 3.9114,
+      "step": 304
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8145454545454555e-06,
+      "loss": 3.9088,
+      "step": 305
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.813818181818182e-06,
+      "loss": 3.9024,
+      "step": 306
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.813090909090909e-06,
+      "loss": 3.9045,
+      "step": 307
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.812363636363637e-06,
+      "loss": 3.8678,
+      "step": 308
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.811636363636364e-06,
+      "loss": 3.8923,
+      "step": 309
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.810909090909091e-06,
+      "loss": 3.897,
+      "step": 310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.810181818181819e-06,
+      "loss": 3.9036,
+      "step": 311
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.809454545454546e-06,
+      "loss": 3.9216,
+      "step": 312
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8087272727272736e-06,
+      "loss": 3.912,
+      "step": 313
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.808e-06,
+      "loss": 3.8979,
+      "step": 314
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.807272727272728e-06,
+      "loss": 3.8887,
+      "step": 315
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.806545454545455e-06,
+      "loss": 3.9242,
+      "step": 316
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.805818181818182e-06,
+      "loss": 3.9123,
+      "step": 317
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.805090909090909e-06,
+      "loss": 3.907,
+      "step": 318
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.804363636363637e-06,
+      "loss": 3.9167,
+      "step": 319
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.803636363636364e-06,
+      "loss": 3.9047,
+      "step": 320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.802909090909092e-06,
+      "loss": 3.9096,
+      "step": 321
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8021818181818184e-06,
+      "loss": 3.9094,
+      "step": 322
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.801454545454546e-06,
+      "loss": 3.918,
+      "step": 323
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.800727272727273e-06,
+      "loss": 3.9152,
+      "step": 324
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 3.9224,
+      "step": 325
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7992727272727275e-06,
+      "loss": 3.905,
+      "step": 326
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.798545454545455e-06,
+      "loss": 3.9235,
+      "step": 327
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.797818181818182e-06,
+      "loss": 3.9108,
+      "step": 328
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.79709090909091e-06,
+      "loss": 3.8739,
+      "step": 329
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7963636363636365e-06,
+      "loss": 3.9019,
+      "step": 330
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.795636363636364e-06,
+      "loss": 3.9238,
+      "step": 331
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.794909090909091e-06,
+      "loss": 3.9037,
+      "step": 332
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.794181818181819e-06,
+      "loss": 3.9181,
+      "step": 333
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7934545454545455e-06,
+      "loss": 3.9174,
+      "step": 334
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.792727272727273e-06,
+      "loss": 3.8725,
+      "step": 335
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.792000000000001e-06,
+      "loss": 3.8995,
+      "step": 336
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.791272727272728e-06,
+      "loss": 3.9178,
+      "step": 337
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7905454545454545e-06,
+      "loss": 3.9178,
+      "step": 338
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.789818181818182e-06,
+      "loss": 3.8768,
+      "step": 339
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.789090909090909e-06,
+      "loss": 3.8868,
+      "step": 340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.788363636363637e-06,
+      "loss": 3.9106,
+      "step": 341
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.787636363636364e-06,
+      "loss": 3.9031,
+      "step": 342
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.786909090909091e-06,
+      "loss": 3.9266,
+      "step": 343
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.786181818181819e-06,
+      "loss": 3.9,
+      "step": 344
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.785454545454546e-06,
+      "loss": 3.8307,
+      "step": 345
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7847272727272734e-06,
+      "loss": 3.8722,
+      "step": 346
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.784e-06,
+      "loss": 3.9129,
+      "step": 347
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.783272727272727e-06,
+      "loss": 3.897,
+      "step": 348
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.782545454545455e-06,
+      "loss": 3.9156,
+      "step": 349
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7818181818181825e-06,
+      "loss": 3.9274,
+      "step": 350
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.781090909090909e-06,
+      "loss": 3.9049,
+      "step": 351
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.780363636363637e-06,
+      "loss": 3.9009,
+      "step": 352
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.779636363636364e-06,
+      "loss": 3.908,
+      "step": 353
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7789090909090915e-06,
+      "loss": 3.9076,
+      "step": 354
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.778181818181818e-06,
+      "loss": 3.885,
+      "step": 355
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.777454545454546e-06,
+      "loss": 3.8845,
+      "step": 356
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.776727272727273e-06,
+      "loss": 3.9211,
+      "step": 357
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7760000000000005e-06,
+      "loss": 3.9011,
+      "step": 358
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.775272727272728e-06,
+      "loss": 3.8897,
+      "step": 359
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.774545454545455e-06,
+      "loss": 3.9179,
+      "step": 360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.773818181818182e-06,
+      "loss": 3.8694,
+      "step": 361
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7730909090909095e-06,
+      "loss": 3.9136,
+      "step": 362
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.772363636363636e-06,
+      "loss": 3.8902,
+      "step": 363
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.771636363636364e-06,
+      "loss": 3.8992,
+      "step": 364
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.770909090909091e-06,
+      "loss": 3.9169,
+      "step": 365
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7701818181818186e-06,
+      "loss": 3.96,
+      "step": 366
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.769454545454546e-06,
+      "loss": 3.9065,
+      "step": 367
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.768727272727273e-06,
+      "loss": 3.9389,
+      "step": 368
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.768000000000001e-06,
+      "loss": 3.907,
+      "step": 369
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.767272727272728e-06,
+      "loss": 3.8857,
+      "step": 370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7665454545454544e-06,
+      "loss": 3.9323,
+      "step": 371
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.765818181818182e-06,
+      "loss": 3.9148,
+      "step": 372
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.765090909090909e-06,
+      "loss": 3.8998,
+      "step": 373
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.764363636363637e-06,
+      "loss": 3.8721,
+      "step": 374
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.763636363636364e-06,
+      "loss": 3.9105,
+      "step": 375
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.762909090909091e-06,
+      "loss": 3.9286,
+      "step": 376
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.762181818181819e-06,
+      "loss": 3.871,
+      "step": 377
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.761454545454546e-06,
+      "loss": 3.8858,
+      "step": 378
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.760727272727273e-06,
+      "loss": 3.93,
+      "step": 379
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.76e-06,
+      "loss": 3.8988,
+      "step": 380
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.759272727272728e-06,
+      "loss": 3.9111,
+      "step": 381
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7585454545454555e-06,
+      "loss": 3.8782,
+      "step": 382
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.757818181818182e-06,
+      "loss": 3.9107,
+      "step": 383
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.757090909090909e-06,
+      "loss": 3.8879,
+      "step": 384
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.756363636363637e-06,
+      "loss": 3.8997,
+      "step": 385
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.755636363636364e-06,
+      "loss": 3.9126,
+      "step": 386
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.754909090909091e-06,
+      "loss": 3.9083,
+      "step": 387
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.754181818181818e-06,
+      "loss": 3.8682,
+      "step": 388
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.753454545454546e-06,
+      "loss": 3.8792,
+      "step": 389
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.752727272727274e-06,
+      "loss": 3.8787,
+      "step": 390
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.752e-06,
+      "loss": 3.8908,
+      "step": 391
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.751272727272728e-06,
+      "loss": 3.9179,
+      "step": 392
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.750545454545455e-06,
+      "loss": 3.8888,
+      "step": 393
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.749818181818182e-06,
+      "loss": 3.865,
+      "step": 394
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7490909090909094e-06,
+      "loss": 3.9138,
+      "step": 395
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.748363636363636e-06,
+      "loss": 3.9243,
+      "step": 396
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.747636363636364e-06,
+      "loss": 3.935,
+      "step": 397
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.746909090909092e-06,
+      "loss": 3.927,
+      "step": 398
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7461818181818185e-06,
+      "loss": 3.9007,
+      "step": 399
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.745454545454546e-06,
+      "loss": 3.8794,
+      "step": 400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.744727272727273e-06,
+      "loss": 3.902,
+      "step": 401
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.744000000000001e-06,
+      "loss": 3.9318,
+      "step": 402
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7432727272727275e-06,
+      "loss": 3.8679,
+      "step": 403
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.742545454545454e-06,
+      "loss": 3.8899,
+      "step": 404
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.741818181818183e-06,
+      "loss": 3.9077,
+      "step": 405
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.74109090909091e-06,
+      "loss": 3.9319,
+      "step": 406
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7403636363636365e-06,
+      "loss": 3.9026,
+      "step": 407
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.739636363636364e-06,
+      "loss": 3.8943,
+      "step": 408
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.738909090909091e-06,
+      "loss": 3.9493,
+      "step": 409
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.738181818181819e-06,
+      "loss": 3.8977,
+      "step": 410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7374545454545455e-06,
+      "loss": 3.8742,
+      "step": 411
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.736727272727273e-06,
+      "loss": 3.9153,
+      "step": 412
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.736000000000001e-06,
+      "loss": 3.8857,
+      "step": 413
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.735272727272728e-06,
+      "loss": 3.9084,
+      "step": 414
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.734545454545455e-06,
+      "loss": 3.8954,
+      "step": 415
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.733818181818182e-06,
+      "loss": 3.9019,
+      "step": 416
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.733090909090909e-06,
+      "loss": 3.8603,
+      "step": 417
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.732363636363637e-06,
+      "loss": 3.9134,
+      "step": 418
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.731636363636364e-06,
+      "loss": 3.9013,
+      "step": 419
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.730909090909091e-06,
+      "loss": 3.9178,
+      "step": 420
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.730181818181819e-06,
+      "loss": 3.9062,
+      "step": 421
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.729454545454546e-06,
+      "loss": 3.9037,
+      "step": 422
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7287272727272735e-06,
+      "loss": 3.903,
+      "step": 423
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.728e-06,
+      "loss": 3.8698,
+      "step": 424
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.727272727272728e-06,
+      "loss": 3.9047,
+      "step": 425
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.726545454545455e-06,
+      "loss": 3.8614,
+      "step": 426
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.725818181818182e-06,
+      "loss": 3.8938,
+      "step": 427
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.725090909090909e-06,
+      "loss": 3.8966,
+      "step": 428
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.724363636363637e-06,
+      "loss": 3.8887,
+      "step": 429
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.723636363636364e-06,
+      "loss": 3.868,
+      "step": 430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7229090909090915e-06,
+      "loss": 3.9049,
+      "step": 431
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.722181818181818e-06,
+      "loss": 3.9103,
+      "step": 432
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.721454545454546e-06,
+      "loss": 3.9487,
+      "step": 433
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.720727272727273e-06,
+      "loss": 3.9246,
+      "step": 434
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 3.8505,
+      "step": 435
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.719272727272728e-06,
+      "loss": 3.9064,
+      "step": 436
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.718545454545455e-06,
+      "loss": 3.895,
+      "step": 437
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.717818181818182e-06,
+      "loss": 3.8607,
+      "step": 438
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7170909090909096e-06,
+      "loss": 3.8795,
+      "step": 439
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.716363636363636e-06,
+      "loss": 3.8725,
+      "step": 440
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.715636363636364e-06,
+      "loss": 3.9443,
+      "step": 441
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.714909090909091e-06,
+      "loss": 3.9115,
+      "step": 442
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.714181818181819e-06,
+      "loss": 3.8786,
+      "step": 443
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.713454545454546e-06,
+      "loss": 3.9169,
+      "step": 444
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.712727272727273e-06,
+      "loss": 3.8754,
+      "step": 445
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.712000000000001e-06,
+      "loss": 3.859,
+      "step": 446
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.711272727272728e-06,
+      "loss": 3.8866,
+      "step": 447
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7105454545454545e-06,
+      "loss": 3.9118,
+      "step": 448
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.709818181818182e-06,
+      "loss": 3.9306,
+      "step": 449
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.709090909090909e-06,
+      "loss": 3.9149,
+      "step": 450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.708363636363637e-06,
+      "loss": 3.8893,
+      "step": 451
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.707636363636364e-06,
+      "loss": 3.8806,
+      "step": 452
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.706909090909091e-06,
+      "loss": 3.9297,
+      "step": 453
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.706181818181819e-06,
+      "loss": 3.8894,
+      "step": 454
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.705454545454546e-06,
+      "loss": 3.8661,
+      "step": 455
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.704727272727273e-06,
+      "loss": 3.8727,
+      "step": 456
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.704e-06,
+      "loss": 3.9008,
+      "step": 457
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.703272727272727e-06,
+      "loss": 3.9415,
+      "step": 458
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7025454545454556e-06,
+      "loss": 3.9099,
+      "step": 459
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.701818181818182e-06,
+      "loss": 3.8579,
+      "step": 460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.701090909090909e-06,
+      "loss": 3.9021,
+      "step": 461
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.700363636363637e-06,
+      "loss": 3.8948,
+      "step": 462
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.699636363636364e-06,
+      "loss": 3.9064,
+      "step": 463
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.698909090909091e-06,
+      "loss": 3.8619,
+      "step": 464
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.698181818181818e-06,
+      "loss": 3.8951,
+      "step": 465
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.697454545454546e-06,
+      "loss": 3.8839,
+      "step": 466
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.696727272727274e-06,
+      "loss": 3.9029,
+      "step": 467
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.6960000000000004e-06,
+      "loss": 3.862,
+      "step": 468
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.695272727272728e-06,
+      "loss": 3.8691,
+      "step": 469
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.694545454545455e-06,
+      "loss": 3.935,
+      "step": 470
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.693818181818182e-06,
+      "loss": 3.9106,
+      "step": 471
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.6930909090909095e-06,
+      "loss": 3.8841,
+      "step": 472
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.692363636363636e-06,
+      "loss": 3.8826,
+      "step": 473
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.691636363636364e-06,
+      "loss": 3.8477,
+      "step": 474
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.690909090909092e-06,
+      "loss": 3.914,
+      "step": 475
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.6901818181818185e-06,
+      "loss": 3.9217,
+      "step": 476
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.689454545454546e-06,
+      "loss": 3.883,
+      "step": 477
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.688727272727273e-06,
+      "loss": 3.9184,
+      "step": 478
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.688000000000001e-06,
+      "loss": 3.9244,
+      "step": 479
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6872727272727275e-06,
+      "loss": 3.916,
+      "step": 480
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.686545454545454e-06,
+      "loss": 3.9356,
+      "step": 481
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.685818181818182e-06,
+      "loss": 3.9046,
+      "step": 482
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.68509090909091e-06,
+      "loss": 3.9181,
+      "step": 483
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.6843636363636365e-06,
+      "loss": 3.9203,
+      "step": 484
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.683636363636364e-06,
+      "loss": 3.9051,
+      "step": 485
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.682909090909091e-06,
+      "loss": 3.909,
+      "step": 486
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.682181818181819e-06,
+      "loss": 3.9126,
+      "step": 487
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.6814545454545456e-06,
+      "loss": 3.9149,
+      "step": 488
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.680727272727273e-06,
+      "loss": 3.9014,
+      "step": 489
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 3.8809,
+      "step": 490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.679272727272728e-06,
+      "loss": 3.8879,
+      "step": 491
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.6785454545454554e-06,
+      "loss": 3.8962,
+      "step": 492
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.677818181818182e-06,
+      "loss": 3.8642,
+      "step": 493
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.677090909090909e-06,
+      "loss": 3.9381,
+      "step": 494
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.676363636363637e-06,
+      "loss": 3.9058,
+      "step": 495
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.675636363636364e-06,
+      "loss": 3.9116,
+      "step": 496
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.674909090909091e-06,
+      "loss": 3.9024,
+      "step": 497
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.674181818181819e-06,
+      "loss": 3.8828,
+      "step": 498
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.673454545454546e-06,
+      "loss": 3.932,
+      "step": 499
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6727272727272735e-06,
+      "loss": 3.9237,
+      "step": 500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.672e-06,
+      "loss": 3.9138,
+      "step": 501
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.671272727272728e-06,
+      "loss": 3.9075,
+      "step": 502
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.670545454545455e-06,
+      "loss": 3.9187,
+      "step": 503
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.669818181818182e-06,
+      "loss": 3.9379,
+      "step": 504
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.669090909090909e-06,
+      "loss": 3.9164,
+      "step": 505
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.668363636363637e-06,
+      "loss": 3.8817,
+      "step": 506
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.667636363636364e-06,
+      "loss": 3.8753,
+      "step": 507
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6669090909090916e-06,
+      "loss": 3.8724,
+      "step": 508
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.666181818181818e-06,
+      "loss": 3.8801,
+      "step": 509
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.665454545454546e-06,
+      "loss": 3.8996,
+      "step": 510
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.664727272727273e-06,
+      "loss": 3.9085,
+      "step": 511
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.664000000000001e-06,
+      "loss": 3.8762,
+      "step": 512
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.663272727272727e-06,
+      "loss": 3.9169,
+      "step": 513
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.662545454545455e-06,
+      "loss": 3.8652,
+      "step": 514
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.661818181818183e-06,
+      "loss": 3.8957,
+      "step": 515
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.66109090909091e-06,
+      "loss": 3.892,
+      "step": 516
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6603636363636364e-06,
+      "loss": 3.9002,
+      "step": 517
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.659636363636364e-06,
+      "loss": 3.9093,
+      "step": 518
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.658909090909091e-06,
+      "loss": 3.8977,
+      "step": 519
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.658181818181819e-06,
+      "loss": 3.8823,
+      "step": 520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6574545454545455e-06,
+      "loss": 3.9172,
+      "step": 521
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.656727272727273e-06,
+      "loss": 3.8686,
+      "step": 522
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.656000000000001e-06,
+      "loss": 3.881,
+      "step": 523
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.655272727272728e-06,
+      "loss": 3.8618,
+      "step": 524
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.654545454545455e-06,
+      "loss": 3.92,
+      "step": 525
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.653818181818182e-06,
+      "loss": 3.9215,
+      "step": 526
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.653090909090909e-06,
+      "loss": 3.931,
+      "step": 527
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.652363636363637e-06,
+      "loss": 3.9365,
+      "step": 528
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.651636363636364e-06,
+      "loss": 3.8618,
+      "step": 529
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.650909090909091e-06,
+      "loss": 3.8733,
+      "step": 530
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.650181818181819e-06,
+      "loss": 3.9092,
+      "step": 531
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.649454545454546e-06,
+      "loss": 3.8911,
+      "step": 532
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.648727272727273e-06,
+      "loss": 3.8584,
+      "step": 533
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.648e-06,
+      "loss": 3.9065,
+      "step": 534
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.647272727272728e-06,
+      "loss": 3.9013,
+      "step": 535
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.646545454545455e-06,
+      "loss": 3.928,
+      "step": 536
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.645818181818182e-06,
+      "loss": 3.9077,
+      "step": 537
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.645090909090909e-06,
+      "loss": 3.9013,
+      "step": 538
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.644363636363637e-06,
+      "loss": 3.9513,
+      "step": 539
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.643636363636364e-06,
+      "loss": 3.9096,
+      "step": 540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6429090909090914e-06,
+      "loss": 3.9511,
+      "step": 541
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.642181818181818e-06,
+      "loss": 3.8867,
+      "step": 542
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.641454545454546e-06,
+      "loss": 3.834,
+      "step": 543
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.640727272727273e-06,
+      "loss": 3.9558,
+      "step": 544
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 3.8893,
+      "step": 545
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.639272727272728e-06,
+      "loss": 3.8796,
+      "step": 546
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.638545454545455e-06,
+      "loss": 3.8919,
+      "step": 547
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.637818181818182e-06,
+      "loss": 3.9041,
+      "step": 548
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6370909090909095e-06,
+      "loss": 3.9177,
+      "step": 549
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.636363636363636e-06,
+      "loss": 3.8667,
+      "step": 550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.635636363636364e-06,
+      "loss": 3.9442,
+      "step": 551
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.634909090909091e-06,
+      "loss": 3.8523,
+      "step": 552
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6341818181818185e-06,
+      "loss": 3.8601,
+      "step": 553
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.633454545454546e-06,
+      "loss": 3.8719,
+      "step": 554
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.05405405405405406,
+      "eval_loss": 3.882791042327881,
+      "eval_runtime": 142.9203,
+      "eval_samples_per_second": 2.589,
+      "eval_steps_per_second": 0.651,
+      "step": 554
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.632727272727273e-06,
+      "loss": 3.9102,
+      "step": 555
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.632000000000001e-06,
+      "loss": 3.864,
+      "step": 556
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6312727272727275e-06,
+      "loss": 3.8992,
+      "step": 557
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.630545454545454e-06,
+      "loss": 3.8851,
+      "step": 558
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.629818181818182e-06,
+      "loss": 3.8495,
+      "step": 559
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.62909090909091e-06,
+      "loss": 3.9134,
+      "step": 560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6283636363636366e-06,
+      "loss": 3.8806,
+      "step": 561
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.627636363636364e-06,
+      "loss": 3.8626,
+      "step": 562
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.626909090909091e-06,
+      "loss": 3.9368,
+      "step": 563
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.626181818181819e-06,
+      "loss": 3.8738,
+      "step": 564
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.625454545454546e-06,
+      "loss": 3.8801,
+      "step": 565
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.624727272727273e-06,
+      "loss": 3.9409,
+      "step": 566
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.624e-06,
+      "loss": 3.8897,
+      "step": 567
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.623272727272728e-06,
+      "loss": 3.8796,
+      "step": 568
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6225454545454555e-06,
+      "loss": 3.909,
+      "step": 569
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.621818181818182e-06,
+      "loss": 3.8904,
+      "step": 570
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.621090909090909e-06,
+      "loss": 3.8803,
+      "step": 571
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.620363636363637e-06,
+      "loss": 3.8174,
+      "step": 572
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.619636363636364e-06,
+      "loss": 3.8893,
+      "step": 573
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.618909090909091e-06,
+      "loss": 3.8279,
+      "step": 574
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.618181818181818e-06,
+      "loss": 3.9018,
+      "step": 575
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.617454545454546e-06,
+      "loss": 3.8742,
+      "step": 576
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6167272727272735e-06,
+      "loss": 3.8598,
+      "step": 577
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.616e-06,
+      "loss": 3.8961,
+      "step": 578
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.615272727272728e-06,
+      "loss": 3.8683,
+      "step": 579
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.614545454545455e-06,
+      "loss": 3.8625,
+      "step": 580
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.613818181818182e-06,
+      "loss": 3.9589,
+      "step": 581
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.613090909090909e-06,
+      "loss": 3.8643,
+      "step": 582
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.612363636363636e-06,
+      "loss": 3.8929,
+      "step": 583
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.611636363636364e-06,
+      "loss": 3.9015,
+      "step": 584
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.610909090909092e-06,
+      "loss": 3.8894,
+      "step": 585
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.610181818181818e-06,
+      "loss": 3.8885,
+      "step": 586
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.609454545454546e-06,
+      "loss": 3.9304,
+      "step": 587
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.608727272727273e-06,
+      "loss": 3.8266,
+      "step": 588
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.608000000000001e-06,
+      "loss": 3.8992,
+      "step": 589
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6072727272727274e-06,
+      "loss": 3.925,
+      "step": 590
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.606545454545454e-06,
+      "loss": 3.895,
+      "step": 591
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.605818181818183e-06,
+      "loss": 3.8577,
+      "step": 592
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.60509090909091e-06,
+      "loss": 3.9004,
+      "step": 593
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6043636363636365e-06,
+      "loss": 3.8617,
+      "step": 594
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.603636363636364e-06,
+      "loss": 3.8822,
+      "step": 595
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.602909090909091e-06,
+      "loss": 3.919,
+      "step": 596
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.602181818181819e-06,
+      "loss": 3.9032,
+      "step": 597
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6014545454545455e-06,
+      "loss": 3.9086,
+      "step": 598
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.600727272727273e-06,
+      "loss": 3.8981,
+      "step": 599
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 3.8561,
+      "step": 600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.599272727272728e-06,
+      "loss": 3.886,
+      "step": 601
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.598545454545455e-06,
+      "loss": 3.9169,
+      "step": 602
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.597818181818182e-06,
+      "loss": 3.8554,
+      "step": 603
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.597090909090909e-06,
+      "loss": 3.9502,
+      "step": 604
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.596363636363637e-06,
+      "loss": 3.8862,
+      "step": 605
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.5956363636363635e-06,
+      "loss": 3.9039,
+      "step": 606
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.594909090909091e-06,
+      "loss": 3.8804,
+      "step": 607
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.594181818181819e-06,
+      "loss": 3.8177,
+      "step": 608
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.593454545454546e-06,
+      "loss": 3.9675,
+      "step": 609
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.592727272727273e-06,
+      "loss": 3.9022,
+      "step": 610
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.592e-06,
+      "loss": 3.8722,
+      "step": 611
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.591272727272728e-06,
+      "loss": 3.824,
+      "step": 612
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.590545454545455e-06,
+      "loss": 3.8576,
+      "step": 613
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.589818181818182e-06,
+      "loss": 3.802,
+      "step": 614
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.58909090909091e-06,
+      "loss": 3.88,
+      "step": 615
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.588363636363637e-06,
+      "loss": 3.8854,
+      "step": 616
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.587636363636364e-06,
+      "loss": 3.9252,
+      "step": 617
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.5869090909090915e-06,
+      "loss": 3.9106,
+      "step": 618
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.586181818181818e-06,
+      "loss": 3.9405,
+      "step": 619
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.585454545454546e-06,
+      "loss": 3.8884,
+      "step": 620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.584727272727273e-06,
+      "loss": 3.8962,
+      "step": 621
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.5840000000000005e-06,
+      "loss": 3.935,
+      "step": 622
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.583272727272728e-06,
+      "loss": 3.8898,
+      "step": 623
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.582545454545455e-06,
+      "loss": 3.9164,
+      "step": 624
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.581818181818183e-06,
+      "loss": 3.974,
+      "step": 625
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.5810909090909095e-06,
+      "loss": 3.8879,
+      "step": 626
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.580363636363636e-06,
+      "loss": 3.8993,
+      "step": 627
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.579636363636364e-06,
+      "loss": 3.8695,
+      "step": 628
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.578909090909091e-06,
+      "loss": 3.8652,
+      "step": 629
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.5781818181818185e-06,
+      "loss": 3.9007,
+      "step": 630
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.577454545454546e-06,
+      "loss": 3.929,
+      "step": 631
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.576727272727273e-06,
+      "loss": 3.9511,
+      "step": 632
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.576000000000001e-06,
+      "loss": 3.8877,
+      "step": 633
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.5752727272727276e-06,
+      "loss": 3.8861,
+      "step": 634
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.574545454545455e-06,
+      "loss": 3.8799,
+      "step": 635
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.573818181818182e-06,
+      "loss": 3.8449,
+      "step": 636
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.573090909090909e-06,
+      "loss": 3.8837,
+      "step": 637
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.572363636363637e-06,
+      "loss": 3.8921,
+      "step": 638
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.571636363636364e-06,
+      "loss": 3.8768,
+      "step": 639
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.570909090909091e-06,
+      "loss": 3.8962,
+      "step": 640
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.570181818181819e-06,
+      "loss": 3.841,
+      "step": 641
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.569454545454546e-06,
+      "loss": 3.8233,
+      "step": 642
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.568727272727273e-06,
+      "loss": 3.8931,
+      "step": 643
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.568e-06,
+      "loss": 3.8502,
+      "step": 644
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.567272727272728e-06,
+      "loss": 3.8631,
+      "step": 645
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.5665454545454555e-06,
+      "loss": 3.9177,
+      "step": 646
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.565818181818182e-06,
+      "loss": 3.8371,
+      "step": 647
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.565090909090909e-06,
+      "loss": 3.8894,
+      "step": 648
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.564363636363637e-06,
+      "loss": 3.8419,
+      "step": 649
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.563636363636364e-06,
+      "loss": 3.8784,
+      "step": 650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.562909090909091e-06,
+      "loss": 3.8885,
+      "step": 651
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.562181818181818e-06,
+      "loss": 3.8362,
+      "step": 652
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.561454545454546e-06,
+      "loss": 3.9536,
+      "step": 653
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5607272727272736e-06,
+      "loss": 3.8919,
+      "step": 654
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.56e-06,
+      "loss": 3.9442,
+      "step": 655
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.559272727272728e-06,
+      "loss": 3.8448,
+      "step": 656
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.558545454545455e-06,
+      "loss": 3.9149,
+      "step": 657
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.557818181818182e-06,
+      "loss": 3.9126,
+      "step": 658
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.557090909090909e-06,
+      "loss": 3.9171,
+      "step": 659
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.556363636363636e-06,
+      "loss": 3.9003,
+      "step": 660
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.555636363636364e-06,
+      "loss": 3.8806,
+      "step": 661
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.554909090909092e-06,
+      "loss": 3.9026,
+      "step": 662
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.5541818181818184e-06,
+      "loss": 3.8807,
+      "step": 663
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.553454545454546e-06,
+      "loss": 3.8857,
+      "step": 664
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.552727272727273e-06,
+      "loss": 3.9012,
+      "step": 665
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.552000000000001e-06,
+      "loss": 3.8956,
+      "step": 666
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5512727272727275e-06,
+      "loss": 3.9132,
+      "step": 667
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.550545454545454e-06,
+      "loss": 3.8559,
+      "step": 668
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.549818181818183e-06,
+      "loss": 3.8792,
+      "step": 669
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.54909090909091e-06,
+      "loss": 3.8799,
+      "step": 670
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5483636363636365e-06,
+      "loss": 3.8996,
+      "step": 671
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.547636363636364e-06,
+      "loss": 3.9425,
+      "step": 672
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.546909090909091e-06,
+      "loss": 3.8846,
+      "step": 673
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.546181818181819e-06,
+      "loss": 3.9815,
+      "step": 674
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 3.9425,
+      "step": 675
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.544727272727273e-06,
+      "loss": 3.8685,
+      "step": 676
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.544000000000001e-06,
+      "loss": 3.9048,
+      "step": 677
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.543272727272728e-06,
+      "loss": 3.9198,
+      "step": 678
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.542545454545455e-06,
+      "loss": 3.9145,
+      "step": 679
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.541818181818182e-06,
+      "loss": 3.9086,
+      "step": 680
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.541090909090909e-06,
+      "loss": 3.8297,
+      "step": 681
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.540363636363637e-06,
+      "loss": 3.843,
+      "step": 682
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5396363636363636e-06,
+      "loss": 3.857,
+      "step": 683
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.538909090909091e-06,
+      "loss": 3.8027,
+      "step": 684
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.538181818181819e-06,
+      "loss": 3.8529,
+      "step": 685
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.537454545454546e-06,
+      "loss": 3.9687,
+      "step": 686
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5367272727272734e-06,
+      "loss": 3.8873,
+      "step": 687
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.536e-06,
+      "loss": 3.8454,
+      "step": 688
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.535272727272728e-06,
+      "loss": 3.879,
+      "step": 689
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.534545454545455e-06,
+      "loss": 3.8841,
+      "step": 690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.533818181818182e-06,
+      "loss": 3.9195,
+      "step": 691
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.533090909090909e-06,
+      "loss": 3.8947,
+      "step": 692
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.532363636363637e-06,
+      "loss": 3.9107,
+      "step": 693
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.531636363636364e-06,
+      "loss": 3.8248,
+      "step": 694
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5309090909090915e-06,
+      "loss": 3.8907,
+      "step": 695
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.530181818181818e-06,
+      "loss": 3.9048,
+      "step": 696
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.529454545454546e-06,
+      "loss": 3.8562,
+      "step": 697
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.528727272727273e-06,
+      "loss": 3.8915,
+      "step": 698
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5280000000000005e-06,
+      "loss": 3.9478,
+      "step": 699
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.527272727272727e-06,
+      "loss": 3.873,
+      "step": 700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.526545454545455e-06,
+      "loss": 3.8976,
+      "step": 701
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.525818181818183e-06,
+      "loss": 3.9271,
+      "step": 702
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5250909090909095e-06,
+      "loss": 3.8795,
+      "step": 703
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.524363636363636e-06,
+      "loss": 3.9244,
+      "step": 704
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.523636363636364e-06,
+      "loss": 3.9049,
+      "step": 705
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.522909090909091e-06,
+      "loss": 3.8584,
+      "step": 706
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5221818181818186e-06,
+      "loss": 3.8561,
+      "step": 707
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.521454545454546e-06,
+      "loss": 3.9561,
+      "step": 708
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.520727272727273e-06,
+      "loss": 3.8516,
+      "step": 709
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 3.8893,
+      "step": 710
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.519272727272728e-06,
+      "loss": 3.8942,
+      "step": 711
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.518545454545455e-06,
+      "loss": 3.8855,
+      "step": 712
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.517818181818182e-06,
+      "loss": 3.8834,
+      "step": 713
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.517090909090909e-06,
+      "loss": 3.8734,
+      "step": 714
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.516363636363637e-06,
+      "loss": 3.7919,
+      "step": 715
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.515636363636364e-06,
+      "loss": 3.8651,
+      "step": 716
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.514909090909091e-06,
+      "loss": 3.8867,
+      "step": 717
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.514181818181819e-06,
+      "loss": 3.8874,
+      "step": 718
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.513454545454546e-06,
+      "loss": 3.8853,
+      "step": 719
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.512727272727273e-06,
+      "loss": 3.9143,
+      "step": 720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.512e-06,
+      "loss": 3.8614,
+      "step": 721
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.511272727272728e-06,
+      "loss": 3.8087,
+      "step": 722
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.510545454545455e-06,
+      "loss": 3.89,
+      "step": 723
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.509818181818182e-06,
+      "loss": 3.9101,
+      "step": 724
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.50909090909091e-06,
+      "loss": 3.8921,
+      "step": 725
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.508363636363637e-06,
+      "loss": 3.9149,
+      "step": 726
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.507636363636364e-06,
+      "loss": 3.9064,
+      "step": 727
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.506909090909091e-06,
+      "loss": 3.9202,
+      "step": 728
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.506181818181818e-06,
+      "loss": 3.8954,
+      "step": 729
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.505454545454546e-06,
+      "loss": 3.9117,
+      "step": 730
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.504727272727273e-06,
+      "loss": 3.8312,
+      "step": 731
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.504e-06,
+      "loss": 3.8846,
+      "step": 732
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.503272727272728e-06,
+      "loss": 3.8372,
+      "step": 733
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.502545454545455e-06,
+      "loss": 3.8793,
+      "step": 734
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.501818181818183e-06,
+      "loss": 3.8916,
+      "step": 735
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5010909090909094e-06,
+      "loss": 3.8278,
+      "step": 736
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.500363636363636e-06,
+      "loss": 3.7696,
+      "step": 737
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.499636363636364e-06,
+      "loss": 3.8849,
+      "step": 738
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.498909090909091e-06,
+      "loss": 3.8996,
+      "step": 739
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.4981818181818185e-06,
+      "loss": 3.8674,
+      "step": 740
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.497454545454546e-06,
+      "loss": 3.8896,
+      "step": 741
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.496727272727273e-06,
+      "loss": 3.8679,
+      "step": 742
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.496000000000001e-06,
+      "loss": 3.8979,
+      "step": 743
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.4952727272727275e-06,
+      "loss": 3.8804,
+      "step": 744
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.494545454545455e-06,
+      "loss": 3.8415,
+      "step": 745
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.493818181818182e-06,
+      "loss": 3.8833,
+      "step": 746
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.49309090909091e-06,
+      "loss": 3.8504,
+      "step": 747
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.4923636363636365e-06,
+      "loss": 3.9012,
+      "step": 748
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.491636363636364e-06,
+      "loss": 3.9181,
+      "step": 749
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.490909090909091e-06,
+      "loss": 3.9199,
+      "step": 750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.490181818181819e-06,
+      "loss": 3.8501,
+      "step": 751
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.4894545454545455e-06,
+      "loss": 3.8544,
+      "step": 752
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.488727272727273e-06,
+      "loss": 3.8195,
+      "step": 753
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.488e-06,
+      "loss": 3.9135,
+      "step": 754
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.487272727272728e-06,
+      "loss": 3.8046,
+      "step": 755
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.486545454545455e-06,
+      "loss": 3.8316,
+      "step": 756
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.485818181818182e-06,
+      "loss": 3.8809,
+      "step": 757
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.485090909090909e-06,
+      "loss": 3.8916,
+      "step": 758
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.484363636363637e-06,
+      "loss": 3.8406,
+      "step": 759
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.483636363636364e-06,
+      "loss": 3.8864,
+      "step": 760
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.482909090909091e-06,
+      "loss": 3.8434,
+      "step": 761
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.482181818181818e-06,
+      "loss": 3.9561,
+      "step": 762
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.481454545454546e-06,
+      "loss": 3.8686,
+      "step": 763
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.4807272727272735e-06,
+      "loss": 3.8791,
+      "step": 764
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.48e-06,
+      "loss": 3.839,
+      "step": 765
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.479272727272728e-06,
+      "loss": 3.887,
+      "step": 766
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.478545454545455e-06,
+      "loss": 3.8679,
+      "step": 767
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.477818181818182e-06,
+      "loss": 3.8646,
+      "step": 768
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.477090909090909e-06,
+      "loss": 3.9309,
+      "step": 769
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.476363636363636e-06,
+      "loss": 3.8607,
+      "step": 770
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.475636363636364e-06,
+      "loss": 3.9249,
+      "step": 771
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.4749090909090915e-06,
+      "loss": 3.8708,
+      "step": 772
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.474181818181818e-06,
+      "loss": 3.911,
+      "step": 773
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.473454545454546e-06,
+      "loss": 3.9395,
+      "step": 774
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.472727272727273e-06,
+      "loss": 3.877,
+      "step": 775
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.4720000000000006e-06,
+      "loss": 3.8635,
+      "step": 776
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.471272727272727e-06,
+      "loss": 3.8517,
+      "step": 777
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.470545454545455e-06,
+      "loss": 3.8332,
+      "step": 778
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.469818181818183e-06,
+      "loss": 3.8812,
+      "step": 779
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.46909090909091e-06,
+      "loss": 3.8378,
+      "step": 780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.468363636363636e-06,
+      "loss": 3.8268,
+      "step": 781
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.467636363636364e-06,
+      "loss": 3.8327,
+      "step": 782
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.466909090909091e-06,
+      "loss": 3.8908,
+      "step": 783
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.466181818181819e-06,
+      "loss": 3.9248,
+      "step": 784
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.4654545454545454e-06,
+      "loss": 3.8855,
+      "step": 785
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.464727272727273e-06,
+      "loss": 3.8693,
+      "step": 786
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.464000000000001e-06,
+      "loss": 3.9312,
+      "step": 787
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.463272727272728e-06,
+      "loss": 3.8926,
+      "step": 788
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.462545454545455e-06,
+      "loss": 3.9209,
+      "step": 789
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.461818181818182e-06,
+      "loss": 3.8065,
+      "step": 790
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.461090909090909e-06,
+      "loss": 3.8802,
+      "step": 791
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.460363636363637e-06,
+      "loss": 3.8683,
+      "step": 792
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.4596363636363635e-06,
+      "loss": 3.9006,
+      "step": 793
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.458909090909091e-06,
+      "loss": 3.8327,
+      "step": 794
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.458181818181819e-06,
+      "loss": 3.8319,
+      "step": 795
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.457454545454546e-06,
+      "loss": 3.8652,
+      "step": 796
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.456727272727273e-06,
+      "loss": 3.8936,
+      "step": 797
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.456e-06,
+      "loss": 3.8585,
+      "step": 798
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.455272727272728e-06,
+      "loss": 3.9426,
+      "step": 799
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.454545454545455e-06,
+      "loss": 3.9191,
+      "step": 800
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.4538181818181815e-06,
+      "loss": 3.872,
+      "step": 801
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.45309090909091e-06,
+      "loss": 3.8672,
+      "step": 802
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.452363636363637e-06,
+      "loss": 3.9207,
+      "step": 803
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.451636363636364e-06,
+      "loss": 3.8219,
+      "step": 804
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.450909090909091e-06,
+      "loss": 3.8451,
+      "step": 805
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.450181818181818e-06,
+      "loss": 3.9553,
+      "step": 806
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.449454545454546e-06,
+      "loss": 3.8318,
+      "step": 807
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.448727272727273e-06,
+      "loss": 3.9289,
+      "step": 808
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.4480000000000004e-06,
+      "loss": 3.916,
+      "step": 809
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.447272727272728e-06,
+      "loss": 3.9454,
+      "step": 810
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.446545454545455e-06,
+      "loss": 3.8038,
+      "step": 811
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.445818181818183e-06,
+      "loss": 3.9006,
+      "step": 812
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4450909090909095e-06,
+      "loss": 3.8822,
+      "step": 813
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.444363636363636e-06,
+      "loss": 3.8789,
+      "step": 814
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.443636363636364e-06,
+      "loss": 3.779,
+      "step": 815
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.442909090909091e-06,
+      "loss": 3.9051,
+      "step": 816
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.4421818181818185e-06,
+      "loss": 3.8599,
+      "step": 817
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.441454545454546e-06,
+      "loss": 3.8848,
+      "step": 818
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.440727272727273e-06,
+      "loss": 3.8414,
+      "step": 819
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 3.9096,
+      "step": 820
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4392727272727275e-06,
+      "loss": 3.9157,
+      "step": 821
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.438545454545455e-06,
+      "loss": 3.886,
+      "step": 822
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.437818181818182e-06,
+      "loss": 3.9034,
+      "step": 823
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.437090909090909e-06,
+      "loss": 3.8901,
+      "step": 824
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.436363636363637e-06,
+      "loss": 3.856,
+      "step": 825
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.435636363636364e-06,
+      "loss": 3.9088,
+      "step": 826
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.434909090909091e-06,
+      "loss": 3.8323,
+      "step": 827
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.434181818181819e-06,
+      "loss": 3.8862,
+      "step": 828
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.4334545454545456e-06,
+      "loss": 3.7848,
+      "step": 829
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.432727272727273e-06,
+      "loss": 3.88,
+      "step": 830
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.432e-06,
+      "loss": 3.8902,
+      "step": 831
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.05945945945945946,
+      "eval_loss": 3.8751111030578613,
+      "eval_runtime": 142.824,
+      "eval_samples_per_second": 2.591,
+      "eval_steps_per_second": 0.651,
+      "step": 831
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.431272727272728e-06,
+      "loss": 3.9444,
+      "step": 832
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.4305454545454554e-06,
+      "loss": 3.8806,
+      "step": 833
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.429818181818182e-06,
+      "loss": 3.9102,
+      "step": 834
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.42909090909091e-06,
+      "loss": 3.9089,
+      "step": 835
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.428363636363637e-06,
+      "loss": 3.9321,
+      "step": 836
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.427636363636364e-06,
+      "loss": 3.8408,
+      "step": 837
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.426909090909091e-06,
+      "loss": 3.8695,
+      "step": 838
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.426181818181818e-06,
+      "loss": 3.7935,
+      "step": 839
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.425454545454546e-06,
+      "loss": 3.9127,
+      "step": 840
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.4247272727272735e-06,
+      "loss": 3.8643,
+      "step": 841
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.424e-06,
+      "loss": 3.9109,
+      "step": 842
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.423272727272728e-06,
+      "loss": 3.9815,
+      "step": 843
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.422545454545455e-06,
+      "loss": 3.8068,
+      "step": 844
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.4218181818181825e-06,
+      "loss": 3.8829,
+      "step": 845
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.421090909090909e-06,
+      "loss": 3.8664,
+      "step": 846
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.420363636363636e-06,
+      "loss": 3.8284,
+      "step": 847
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.419636363636364e-06,
+      "loss": 3.8327,
+      "step": 848
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.4189090909090916e-06,
+      "loss": 3.9424,
+      "step": 849
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.418181818181818e-06,
+      "loss": 3.8573,
+      "step": 850
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.417454545454546e-06,
+      "loss": 3.9048,
+      "step": 851
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.416727272727273e-06,
+      "loss": 3.8512,
+      "step": 852
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.416000000000001e-06,
+      "loss": 3.9081,
+      "step": 853
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.415272727272727e-06,
+      "loss": 3.8414,
+      "step": 854
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.414545454545455e-06,
+      "loss": 3.8117,
+      "step": 855
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.413818181818183e-06,
+      "loss": 3.8355,
+      "step": 856
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.41309090909091e-06,
+      "loss": 3.8756,
+      "step": 857
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.4123636363636364e-06,
+      "loss": 3.9092,
+      "step": 858
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.411636363636364e-06,
+      "loss": 3.9408,
+      "step": 859
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.410909090909091e-06,
+      "loss": 3.8407,
+      "step": 860
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.410181818181819e-06,
+      "loss": 3.8651,
+      "step": 861
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4094545454545455e-06,
+      "loss": 3.8281,
+      "step": 862
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.408727272727273e-06,
+      "loss": 3.8563,
+      "step": 863
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.408000000000001e-06,
+      "loss": 3.9231,
+      "step": 864
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.407272727272728e-06,
+      "loss": 3.8752,
+      "step": 865
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.406545454545455e-06,
+      "loss": 3.9419,
+      "step": 866
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.405818181818182e-06,
+      "loss": 3.8776,
+      "step": 867
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.405090909090909e-06,
+      "loss": 3.8992,
+      "step": 868
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.404363636363637e-06,
+      "loss": 3.8842,
+      "step": 869
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.4036363636363635e-06,
+      "loss": 3.8778,
+      "step": 870
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.402909090909091e-06,
+      "loss": 3.8651,
+      "step": 871
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.402181818181819e-06,
+      "loss": 3.9135,
+      "step": 872
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.401454545454546e-06,
+      "loss": 3.7934,
+      "step": 873
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.400727272727273e-06,
+      "loss": 3.9092,
+      "step": 874
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.4e-06,
+      "loss": 3.7859,
+      "step": 875
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.399272727272728e-06,
+      "loss": 3.849,
+      "step": 876
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.398545454545455e-06,
+      "loss": 3.8678,
+      "step": 877
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.3978181818181816e-06,
+      "loss": 3.8593,
+      "step": 878
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.397090909090909e-06,
+      "loss": 3.7794,
+      "step": 879
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.396363636363637e-06,
+      "loss": 3.7587,
+      "step": 880
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.395636363636364e-06,
+      "loss": 3.8461,
+      "step": 881
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.3949090909090914e-06,
+      "loss": 3.8716,
+      "step": 882
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.394181818181818e-06,
+      "loss": 3.9101,
+      "step": 883
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.393454545454546e-06,
+      "loss": 3.8866,
+      "step": 884
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.392727272727273e-06,
+      "loss": 3.785,
+      "step": 885
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.3920000000000005e-06,
+      "loss": 3.8148,
+      "step": 886
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.391272727272728e-06,
+      "loss": 3.8329,
+      "step": 887
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.390545454545455e-06,
+      "loss": 3.7697,
+      "step": 888
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.389818181818183e-06,
+      "loss": 3.8707,
+      "step": 889
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.3890909090909095e-06,
+      "loss": 3.8379,
+      "step": 890
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.388363636363636e-06,
+      "loss": 3.8751,
+      "step": 891
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.387636363636364e-06,
+      "loss": 3.8926,
+      "step": 892
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.386909090909091e-06,
+      "loss": 3.8656,
+      "step": 893
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.3861818181818185e-06,
+      "loss": 3.8858,
+      "step": 894
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.385454545454546e-06,
+      "loss": 3.8864,
+      "step": 895
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.384727272727273e-06,
+      "loss": 3.8654,
+      "step": 896
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.384000000000001e-06,
+      "loss": 3.9014,
+      "step": 897
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.3832727272727275e-06,
+      "loss": 3.8938,
+      "step": 898
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.382545454545455e-06,
+      "loss": 3.8485,
+      "step": 899
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.381818181818182e-06,
+      "loss": 3.8874,
+      "step": 900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.381090909090909e-06,
+      "loss": 3.8792,
+      "step": 901
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.3803636363636366e-06,
+      "loss": 3.8921,
+      "step": 902
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.379636363636364e-06,
+      "loss": 3.7999,
+      "step": 903
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.378909090909091e-06,
+      "loss": 3.8697,
+      "step": 904
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.378181818181819e-06,
+      "loss": 3.8764,
+      "step": 905
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.377454545454546e-06,
+      "loss": 3.8593,
+      "step": 906
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.376727272727273e-06,
+      "loss": 3.9116,
+      "step": 907
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.376e-06,
+      "loss": 3.9352,
+      "step": 908
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.375272727272728e-06,
+      "loss": 3.869,
+      "step": 909
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.374545454545455e-06,
+      "loss": 3.8899,
+      "step": 910
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.373818181818182e-06,
+      "loss": 3.8456,
+      "step": 911
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.37309090909091e-06,
+      "loss": 3.8708,
+      "step": 912
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.372363636363637e-06,
+      "loss": 3.8639,
+      "step": 913
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.371636363636364e-06,
+      "loss": 3.8275,
+      "step": 914
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.370909090909091e-06,
+      "loss": 3.9345,
+      "step": 915
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.370181818181818e-06,
+      "loss": 3.8308,
+      "step": 916
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.369454545454546e-06,
+      "loss": 3.8658,
+      "step": 917
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.368727272727273e-06,
+      "loss": 3.8509,
+      "step": 918
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.368e-06,
+      "loss": 3.8576,
+      "step": 919
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.367272727272728e-06,
+      "loss": 3.8779,
+      "step": 920
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.366545454545455e-06,
+      "loss": 3.8837,
+      "step": 921
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.3658181818181826e-06,
+      "loss": 3.7757,
+      "step": 922
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.365090909090909e-06,
+      "loss": 3.7944,
+      "step": 923
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.364363636363636e-06,
+      "loss": 3.9095,
+      "step": 924
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.363636363636364e-06,
+      "loss": 3.836,
+      "step": 925
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.362909090909092e-06,
+      "loss": 3.8402,
+      "step": 926
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.362181818181818e-06,
+      "loss": 3.8227,
+      "step": 927
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.361454545454546e-06,
+      "loss": 3.8,
+      "step": 928
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.360727272727273e-06,
+      "loss": 3.8769,
+      "step": 929
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 3.9247,
+      "step": 930
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.3592727272727274e-06,
+      "loss": 3.927,
+      "step": 931
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.358545454545455e-06,
+      "loss": 3.8517,
+      "step": 932
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.357818181818182e-06,
+      "loss": 3.8198,
+      "step": 933
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.35709090909091e-06,
+      "loss": 3.8565,
+      "step": 934
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.356363636363637e-06,
+      "loss": 3.8949,
+      "step": 935
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.355636363636364e-06,
+      "loss": 3.8521,
+      "step": 936
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.354909090909091e-06,
+      "loss": 3.8278,
+      "step": 937
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.354181818181819e-06,
+      "loss": 3.8606,
+      "step": 938
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.3534545454545455e-06,
+      "loss": 3.8338,
+      "step": 939
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.352727272727273e-06,
+      "loss": 3.9183,
+      "step": 940
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.352e-06,
+      "loss": 3.8212,
+      "step": 941
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.351272727272728e-06,
+      "loss": 3.8,
+      "step": 942
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.350545454545455e-06,
+      "loss": 3.8615,
+      "step": 943
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.349818181818182e-06,
+      "loss": 3.8897,
+      "step": 944
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.34909090909091e-06,
+      "loss": 3.7659,
+      "step": 945
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.348363636363637e-06,
+      "loss": 3.8911,
+      "step": 946
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.3476363636363635e-06,
+      "loss": 3.8555,
+      "step": 947
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.346909090909091e-06,
+      "loss": 3.9018,
+      "step": 948
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.346181818181818e-06,
+      "loss": 3.8513,
+      "step": 949
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.345454545454546e-06,
+      "loss": 3.8435,
+      "step": 950
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.344727272727273e-06,
+      "loss": 3.9144,
+      "step": 951
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.344e-06,
+      "loss": 3.892,
+      "step": 952
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.343272727272728e-06,
+      "loss": 3.9087,
+      "step": 953
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.342545454545455e-06,
+      "loss": 3.8673,
+      "step": 954
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.3418181818181824e-06,
+      "loss": 3.9253,
+      "step": 955
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.341090909090909e-06,
+      "loss": 3.9312,
+      "step": 956
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.340363636363636e-06,
+      "loss": 3.801,
+      "step": 957
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.339636363636364e-06,
+      "loss": 3.7984,
+      "step": 958
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.3389090909090915e-06,
+      "loss": 3.7365,
+      "step": 959
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.338181818181818e-06,
+      "loss": 3.8032,
+      "step": 960
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.337454545454546e-06,
+      "loss": 3.7993,
+      "step": 961
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.336727272727273e-06,
+      "loss": 3.9172,
+      "step": 962
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.3360000000000005e-06,
+      "loss": 3.8704,
+      "step": 963
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.335272727272727e-06,
+      "loss": 3.946,
+      "step": 964
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.334545454545455e-06,
+      "loss": 3.8557,
+      "step": 965
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.333818181818183e-06,
+      "loss": 3.9646,
+      "step": 966
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3330909090909095e-06,
+      "loss": 3.8421,
+      "step": 967
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.332363636363636e-06,
+      "loss": 3.9361,
+      "step": 968
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.331636363636364e-06,
+      "loss": 3.8949,
+      "step": 969
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.330909090909091e-06,
+      "loss": 3.9292,
+      "step": 970
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3301818181818186e-06,
+      "loss": 3.8526,
+      "step": 971
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.329454545454545e-06,
+      "loss": 3.9111,
+      "step": 972
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.328727272727273e-06,
+      "loss": 3.8661,
+      "step": 973
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.328000000000001e-06,
+      "loss": 3.8389,
+      "step": 974
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.327272727272728e-06,
+      "loss": 3.8433,
+      "step": 975
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.326545454545455e-06,
+      "loss": 3.8242,
+      "step": 976
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.325818181818182e-06,
+      "loss": 3.77,
+      "step": 977
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.325090909090909e-06,
+      "loss": 3.893,
+      "step": 978
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.324363636363637e-06,
+      "loss": 3.8549,
+      "step": 979
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.3236363636363634e-06,
+      "loss": 3.7394,
+      "step": 980
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.322909090909091e-06,
+      "loss": 3.864,
+      "step": 981
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.322181818181819e-06,
+      "loss": 3.832,
+      "step": 982
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.321454545454546e-06,
+      "loss": 3.8424,
+      "step": 983
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.320727272727273e-06,
+      "loss": 3.8517,
+      "step": 984
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.32e-06,
+      "loss": 3.8202,
+      "step": 985
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.319272727272728e-06,
+      "loss": 3.8712,
+      "step": 986
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.318545454545455e-06,
+      "loss": 3.9517,
+      "step": 987
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.3178181818181815e-06,
+      "loss": 3.8694,
+      "step": 988
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.31709090909091e-06,
+      "loss": 3.9184,
+      "step": 989
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.316363636363637e-06,
+      "loss": 3.8101,
+      "step": 990
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.315636363636364e-06,
+      "loss": 3.8505,
+      "step": 991
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.314909090909091e-06,
+      "loss": 3.8809,
+      "step": 992
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.314181818181818e-06,
+      "loss": 3.9009,
+      "step": 993
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.313454545454546e-06,
+      "loss": 3.8802,
+      "step": 994
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.312727272727273e-06,
+      "loss": 3.8377,
+      "step": 995
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.312e-06,
+      "loss": 3.8646,
+      "step": 996
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.311272727272728e-06,
+      "loss": 3.8204,
+      "step": 997
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.310545454545455e-06,
+      "loss": 3.8533,
+      "step": 998
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.309818181818183e-06,
+      "loss": 3.8914,
+      "step": 999
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.309090909090909e-06,
+      "loss": 3.85,
+      "step": 1000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.308363636363636e-06,
+      "loss": 3.8875,
+      "step": 1001
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.307636363636364e-06,
+      "loss": 3.8762,
+      "step": 1002
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.306909090909091e-06,
+      "loss": 3.8107,
+      "step": 1003
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.3061818181818184e-06,
+      "loss": 3.9037,
+      "step": 1004
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.305454545454546e-06,
+      "loss": 3.9219,
+      "step": 1005
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.304727272727273e-06,
+      "loss": 3.8988,
+      "step": 1006
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.304000000000001e-06,
+      "loss": 3.9123,
+      "step": 1007
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3032727272727275e-06,
+      "loss": 3.8793,
+      "step": 1008
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.302545454545455e-06,
+      "loss": 3.7978,
+      "step": 1009
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.301818181818182e-06,
+      "loss": 3.874,
+      "step": 1010
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.301090909090909e-06,
+      "loss": 3.8764,
+      "step": 1011
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.300363636363637e-06,
+      "loss": 3.85,
+      "step": 1012
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.299636363636364e-06,
+      "loss": 3.8861,
+      "step": 1013
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.298909090909091e-06,
+      "loss": 3.919,
+      "step": 1014
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.298181818181819e-06,
+      "loss": 3.8604,
+      "step": 1015
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.2974545454545455e-06,
+      "loss": 3.7657,
+      "step": 1016
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.296727272727273e-06,
+      "loss": 3.8671,
+      "step": 1017
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.296e-06,
+      "loss": 3.8448,
+      "step": 1018
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.295272727272728e-06,
+      "loss": 3.9069,
+      "step": 1019
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.294545454545455e-06,
+      "loss": 3.891,
+      "step": 1020
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.293818181818182e-06,
+      "loss": 3.8132,
+      "step": 1021
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.29309090909091e-06,
+      "loss": 3.8227,
+      "step": 1022
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.292363636363637e-06,
+      "loss": 3.9631,
+      "step": 1023
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.2916363636363636e-06,
+      "loss": 3.8501,
+      "step": 1024
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.290909090909091e-06,
+      "loss": 3.874,
+      "step": 1025
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.290181818181818e-06,
+      "loss": 3.8356,
+      "step": 1026
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.289454545454546e-06,
+      "loss": 3.8978,
+      "step": 1027
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.2887272727272734e-06,
+      "loss": 3.8621,
+      "step": 1028
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.288e-06,
+      "loss": 3.8687,
+      "step": 1029
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.287272727272728e-06,
+      "loss": 3.8517,
+      "step": 1030
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.286545454545455e-06,
+      "loss": 3.768,
+      "step": 1031
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.2858181818181825e-06,
+      "loss": 3.7273,
+      "step": 1032
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.285090909090909e-06,
+      "loss": 3.845,
+      "step": 1033
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.284363636363636e-06,
+      "loss": 3.8752,
+      "step": 1034
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.283636363636365e-06,
+      "loss": 3.8216,
+      "step": 1035
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.2829090909090915e-06,
+      "loss": 3.8327,
+      "step": 1036
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.282181818181818e-06,
+      "loss": 3.8519,
+      "step": 1037
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.281454545454546e-06,
+      "loss": 3.8433,
+      "step": 1038
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.280727272727273e-06,
+      "loss": 3.839,
+      "step": 1039
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 3.8435,
+      "step": 1040
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.279272727272727e-06,
+      "loss": 3.8308,
+      "step": 1041
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.278545454545455e-06,
+      "loss": 3.8696,
+      "step": 1042
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.277818181818183e-06,
+      "loss": 3.7771,
+      "step": 1043
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.2770909090909096e-06,
+      "loss": 3.882,
+      "step": 1044
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.276363636363637e-06,
+      "loss": 3.9078,
+      "step": 1045
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.275636363636364e-06,
+      "loss": 3.8951,
+      "step": 1046
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.274909090909091e-06,
+      "loss": 3.8012,
+      "step": 1047
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.274181818181819e-06,
+      "loss": 3.8842,
+      "step": 1048
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.273454545454545e-06,
+      "loss": 3.831,
+      "step": 1049
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.272727272727273e-06,
+      "loss": 3.7911,
+      "step": 1050
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.272000000000001e-06,
+      "loss": 3.799,
+      "step": 1051
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.271272727272728e-06,
+      "loss": 3.8417,
+      "step": 1052
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.270545454545455e-06,
+      "loss": 3.8219,
+      "step": 1053
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.269818181818182e-06,
+      "loss": 3.7666,
+      "step": 1054
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.26909090909091e-06,
+      "loss": 3.8022,
+      "step": 1055
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.268363636363637e-06,
+      "loss": 3.87,
+      "step": 1056
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.2676363636363635e-06,
+      "loss": 3.8899,
+      "step": 1057
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.266909090909091e-06,
+      "loss": 3.9658,
+      "step": 1058
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.266181818181819e-06,
+      "loss": 3.8305,
+      "step": 1059
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.265454545454546e-06,
+      "loss": 3.9279,
+      "step": 1060
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.264727272727273e-06,
+      "loss": 3.8691,
+      "step": 1061
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.264e-06,
+      "loss": 3.8095,
+      "step": 1062
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.263272727272728e-06,
+      "loss": 3.8743,
+      "step": 1063
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.262545454545455e-06,
+      "loss": 3.841,
+      "step": 1064
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.261818181818182e-06,
+      "loss": 3.8539,
+      "step": 1065
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.261090909090909e-06,
+      "loss": 3.8704,
+      "step": 1066
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.260363636363637e-06,
+      "loss": 3.857,
+      "step": 1067
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.259636363636364e-06,
+      "loss": 3.8347,
+      "step": 1068
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.258909090909091e-06,
+      "loss": 3.8654,
+      "step": 1069
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.258181818181818e-06,
+      "loss": 3.8945,
+      "step": 1070
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.257454545454546e-06,
+      "loss": 3.8761,
+      "step": 1071
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.256727272727273e-06,
+      "loss": 3.8487,
+      "step": 1072
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.256e-06,
+      "loss": 3.8748,
+      "step": 1073
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.255272727272728e-06,
+      "loss": 3.8711,
+      "step": 1074
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.254545454545455e-06,
+      "loss": 3.7664,
+      "step": 1075
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.253818181818183e-06,
+      "loss": 3.7964,
+      "step": 1076
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.2530909090909094e-06,
+      "loss": 3.8252,
+      "step": 1077
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.252363636363636e-06,
+      "loss": 3.8652,
+      "step": 1078
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.251636363636364e-06,
+      "loss": 3.9265,
+      "step": 1079
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.250909090909091e-06,
+      "loss": 3.8359,
+      "step": 1080
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.2501818181818185e-06,
+      "loss": 3.902,
+      "step": 1081
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.249454545454546e-06,
+      "loss": 3.8484,
+      "step": 1082
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.248727272727273e-06,
+      "loss": 3.7861,
+      "step": 1083
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.248000000000001e-06,
+      "loss": 3.9398,
+      "step": 1084
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.2472727272727275e-06,
+      "loss": 3.8337,
+      "step": 1085
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.246545454545455e-06,
+      "loss": 3.824,
+      "step": 1086
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.245818181818182e-06,
+      "loss": 3.848,
+      "step": 1087
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.245090909090909e-06,
+      "loss": 3.8588,
+      "step": 1088
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2443636363636365e-06,
+      "loss": 3.8033,
+      "step": 1089
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.243636363636364e-06,
+      "loss": 3.8709,
+      "step": 1090
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.242909090909091e-06,
+      "loss": 3.909,
+      "step": 1091
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.242181818181819e-06,
+      "loss": 3.816,
+      "step": 1092
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.2414545454545455e-06,
+      "loss": 3.923,
+      "step": 1093
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.240727272727273e-06,
+      "loss": 3.8657,
+      "step": 1094
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.24e-06,
+      "loss": 3.8211,
+      "step": 1095
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.239272727272728e-06,
+      "loss": 3.8992,
+      "step": 1096
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.2385454545454546e-06,
+      "loss": 3.8732,
+      "step": 1097
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.237818181818182e-06,
+      "loss": 3.8594,
+      "step": 1098
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.23709090909091e-06,
+      "loss": 3.7911,
+      "step": 1099
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.236363636363637e-06,
+      "loss": 3.8121,
+      "step": 1100
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.235636363636364e-06,
+      "loss": 3.8499,
+      "step": 1101
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.234909090909091e-06,
+      "loss": 3.7437,
+      "step": 1102
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.234181818181818e-06,
+      "loss": 3.8832,
+      "step": 1103
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.233454545454546e-06,
+      "loss": 3.8752,
+      "step": 1104
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.2327272727272735e-06,
+      "loss": 3.9161,
+      "step": 1105
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.232e-06,
+      "loss": 3.8752,
+      "step": 1106
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.231272727272728e-06,
+      "loss": 3.8892,
+      "step": 1107
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.230545454545455e-06,
+      "loss": 3.81,
+      "step": 1108
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.08378378378378379,
+      "eval_loss": 3.8348114490509033,
+      "eval_runtime": 143.3498,
+      "eval_samples_per_second": 2.581,
+      "eval_steps_per_second": 0.649,
+      "step": 1108
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.2298181818181825e-06,
+      "loss": 3.8409,
+      "step": 1109
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.229090909090909e-06,
+      "loss": 3.8453,
+      "step": 1110
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.228363636363636e-06,
+      "loss": 3.7484,
+      "step": 1111
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.227636363636364e-06,
+      "loss": 3.8465,
+      "step": 1112
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.2269090909090915e-06,
+      "loss": 3.8198,
+      "step": 1113
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.226181818181818e-06,
+      "loss": 3.9175,
+      "step": 1114
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.225454545454546e-06,
+      "loss": 3.7387,
+      "step": 1115
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.224727272727273e-06,
+      "loss": 3.8572,
+      "step": 1116
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.2240000000000006e-06,
+      "loss": 3.8898,
+      "step": 1117
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.223272727272727e-06,
+      "loss": 3.7961,
+      "step": 1118
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.222545454545455e-06,
+      "loss": 3.8364,
+      "step": 1119
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.221818181818182e-06,
+      "loss": 3.8127,
+      "step": 1120
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.22109090909091e-06,
+      "loss": 3.7845,
+      "step": 1121
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.220363636363637e-06,
+      "loss": 3.8194,
+      "step": 1122
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.219636363636364e-06,
+      "loss": 3.757,
+      "step": 1123
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.218909090909091e-06,
+      "loss": 3.8952,
+      "step": 1124
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.218181818181819e-06,
+      "loss": 3.7853,
+      "step": 1125
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.2174545454545454e-06,
+      "loss": 3.8333,
+      "step": 1126
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.216727272727273e-06,
+      "loss": 3.769,
+      "step": 1127
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.216e-06,
+      "loss": 3.7935,
+      "step": 1128
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.215272727272728e-06,
+      "loss": 3.8984,
+      "step": 1129
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.214545454545455e-06,
+      "loss": 3.8594,
+      "step": 1130
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.213818181818182e-06,
+      "loss": 3.843,
+      "step": 1131
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.21309090909091e-06,
+      "loss": 3.8079,
+      "step": 1132
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.212363636363637e-06,
+      "loss": 3.9208,
+      "step": 1133
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.2116363636363635e-06,
+      "loss": 3.9333,
+      "step": 1134
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.210909090909091e-06,
+      "loss": 3.8808,
+      "step": 1135
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.210181818181818e-06,
+      "loss": 3.8082,
+      "step": 1136
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.209454545454546e-06,
+      "loss": 3.7824,
+      "step": 1137
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.208727272727273e-06,
+      "loss": 3.8544,
+      "step": 1138
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.208e-06,
+      "loss": 3.8521,
+      "step": 1139
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.207272727272728e-06,
+      "loss": 3.7967,
+      "step": 1140
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.206545454545455e-06,
+      "loss": 3.8141,
+      "step": 1141
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.205818181818182e-06,
+      "loss": 3.8778,
+      "step": 1142
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.205090909090909e-06,
+      "loss": 3.8499,
+      "step": 1143
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.204363636363637e-06,
+      "loss": 3.7311,
+      "step": 1144
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.203636363636365e-06,
+      "loss": 3.7813,
+      "step": 1145
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.202909090909091e-06,
+      "loss": 3.8287,
+      "step": 1146
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.202181818181818e-06,
+      "loss": 3.826,
+      "step": 1147
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.201454545454546e-06,
+      "loss": 3.8472,
+      "step": 1148
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.200727272727273e-06,
+      "loss": 3.876,
+      "step": 1149
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 3.8822,
+      "step": 1150
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.199272727272727e-06,
+      "loss": 3.6911,
+      "step": 1151
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.198545454545455e-06,
+      "loss": 3.8623,
+      "step": 1152
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.197818181818183e-06,
+      "loss": 3.8439,
+      "step": 1153
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.1970909090909095e-06,
+      "loss": 3.9183,
+      "step": 1154
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.196363636363637e-06,
+      "loss": 3.6305,
+      "step": 1155
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.195636363636364e-06,
+      "loss": 3.8569,
+      "step": 1156
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.194909090909091e-06,
+      "loss": 3.8075,
+      "step": 1157
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.1941818181818185e-06,
+      "loss": 3.7878,
+      "step": 1158
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.193454545454545e-06,
+      "loss": 3.8139,
+      "step": 1159
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.192727272727273e-06,
+      "loss": 3.9093,
+      "step": 1160
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.192000000000001e-06,
+      "loss": 3.8227,
+      "step": 1161
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.1912727272727275e-06,
+      "loss": 3.7112,
+      "step": 1162
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.190545454545455e-06,
+      "loss": 3.7961,
+      "step": 1163
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.189818181818182e-06,
+      "loss": 3.8507,
+      "step": 1164
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.18909090909091e-06,
+      "loss": 3.8559,
+      "step": 1165
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.1883636363636366e-06,
+      "loss": 3.779,
+      "step": 1166
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.187636363636363e-06,
+      "loss": 3.862,
+      "step": 1167
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.186909090909091e-06,
+      "loss": 3.685,
+      "step": 1168
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.186181818181819e-06,
+      "loss": 3.9902,
+      "step": 1169
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.185454545454546e-06,
+      "loss": 3.8827,
+      "step": 1170
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.184727272727273e-06,
+      "loss": 3.8729,
+      "step": 1171
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.184e-06,
+      "loss": 3.8608,
+      "step": 1172
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.183272727272728e-06,
+      "loss": 3.8672,
+      "step": 1173
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.182545454545455e-06,
+      "loss": 3.8197,
+      "step": 1174
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.181818181818182e-06,
+      "loss": 3.7883,
+      "step": 1175
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.18109090909091e-06,
+      "loss": 3.8558,
+      "step": 1176
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.180363636363637e-06,
+      "loss": 3.869,
+      "step": 1177
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.179636363636364e-06,
+      "loss": 3.8104,
+      "step": 1178
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.178909090909091e-06,
+      "loss": 3.7676,
+      "step": 1179
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.178181818181818e-06,
+      "loss": 3.8851,
+      "step": 1180
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.177454545454546e-06,
+      "loss": 3.8028,
+      "step": 1181
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.176727272727273e-06,
+      "loss": 3.8355,
+      "step": 1182
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.176e-06,
+      "loss": 3.7957,
+      "step": 1183
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.175272727272728e-06,
+      "loss": 3.895,
+      "step": 1184
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.174545454545455e-06,
+      "loss": 3.8195,
+      "step": 1185
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.1738181818181825e-06,
+      "loss": 3.8806,
+      "step": 1186
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.173090909090909e-06,
+      "loss": 3.8906,
+      "step": 1187
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.172363636363636e-06,
+      "loss": 3.8701,
+      "step": 1188
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.171636363636364e-06,
+      "loss": 3.943,
+      "step": 1189
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.170909090909091e-06,
+      "loss": 3.7781,
+      "step": 1190
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.170181818181818e-06,
+      "loss": 3.8527,
+      "step": 1191
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.169454545454546e-06,
+      "loss": 3.7687,
+      "step": 1192
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.168727272727273e-06,
+      "loss": 3.8558,
+      "step": 1193
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.168000000000001e-06,
+      "loss": 3.9114,
+      "step": 1194
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.167272727272727e-06,
+      "loss": 3.8503,
+      "step": 1195
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.166545454545455e-06,
+      "loss": 3.8372,
+      "step": 1196
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.165818181818182e-06,
+      "loss": 3.7829,
+      "step": 1197
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.165090909090909e-06,
+      "loss": 3.8038,
+      "step": 1198
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.164363636363637e-06,
+      "loss": 3.8972,
+      "step": 1199
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.163636363636364e-06,
+      "loss": 3.8138,
+      "step": 1200
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.162909090909091e-06,
+      "loss": 3.7809,
+      "step": 1201
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.162181818181819e-06,
+      "loss": 3.8336,
+      "step": 1202
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.1614545454545455e-06,
+      "loss": 3.804,
+      "step": 1203
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.160727272727273e-06,
+      "loss": 3.8043,
+      "step": 1204
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.16e-06,
+      "loss": 3.8724,
+      "step": 1205
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.159272727272728e-06,
+      "loss": 3.8343,
+      "step": 1206
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.158545454545455e-06,
+      "loss": 3.8066,
+      "step": 1207
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.157818181818182e-06,
+      "loss": 3.9617,
+      "step": 1208
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.15709090909091e-06,
+      "loss": 3.7712,
+      "step": 1209
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.156363636363637e-06,
+      "loss": 3.7576,
+      "step": 1210
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.1556363636363635e-06,
+      "loss": 3.8274,
+      "step": 1211
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.154909090909091e-06,
+      "loss": 3.8352,
+      "step": 1212
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.154181818181818e-06,
+      "loss": 3.7606,
+      "step": 1213
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.153454545454546e-06,
+      "loss": 3.8452,
+      "step": 1214
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.152727272727273e-06,
+      "loss": 3.7624,
+      "step": 1215
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.152e-06,
+      "loss": 3.8791,
+      "step": 1216
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.151272727272728e-06,
+      "loss": 3.7806,
+      "step": 1217
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.150545454545455e-06,
+      "loss": 3.8443,
+      "step": 1218
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.149818181818182e-06,
+      "loss": 3.7378,
+      "step": 1219
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.149090909090909e-06,
+      "loss": 3.7679,
+      "step": 1220
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.148363636363636e-06,
+      "loss": 3.8465,
+      "step": 1221
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.147636363636365e-06,
+      "loss": 3.7539,
+      "step": 1222
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.1469090909090914e-06,
+      "loss": 3.8852,
+      "step": 1223
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.146181818181818e-06,
+      "loss": 3.8503,
+      "step": 1224
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.145454545454546e-06,
+      "loss": 3.8114,
+      "step": 1225
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.144727272727273e-06,
+      "loss": 3.8254,
+      "step": 1226
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.1440000000000005e-06,
+      "loss": 3.8684,
+      "step": 1227
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.143272727272727e-06,
+      "loss": 3.7725,
+      "step": 1228
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.142545454545455e-06,
+      "loss": 3.866,
+      "step": 1229
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.141818181818183e-06,
+      "loss": 3.8381,
+      "step": 1230
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.1410909090909095e-06,
+      "loss": 3.8193,
+      "step": 1231
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.140363636363637e-06,
+      "loss": 3.8485,
+      "step": 1232
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.139636363636364e-06,
+      "loss": 3.8181,
+      "step": 1233
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.138909090909091e-06,
+      "loss": 3.8289,
+      "step": 1234
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.1381818181818185e-06,
+      "loss": 3.8061,
+      "step": 1235
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.137454545454545e-06,
+      "loss": 3.7525,
+      "step": 1236
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.136727272727273e-06,
+      "loss": 3.6942,
+      "step": 1237
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.136000000000001e-06,
+      "loss": 3.8785,
+      "step": 1238
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.1352727272727276e-06,
+      "loss": 3.8323,
+      "step": 1239
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.134545454545455e-06,
+      "loss": 3.8461,
+      "step": 1240
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.133818181818182e-06,
+      "loss": 3.8961,
+      "step": 1241
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.13309090909091e-06,
+      "loss": 3.935,
+      "step": 1242
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.132363636363637e-06,
+      "loss": 3.8098,
+      "step": 1243
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.131636363636363e-06,
+      "loss": 3.7349,
+      "step": 1244
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.130909090909091e-06,
+      "loss": 3.7371,
+      "step": 1245
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.130181818181819e-06,
+      "loss": 3.8776,
+      "step": 1246
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.129454545454546e-06,
+      "loss": 3.798,
+      "step": 1247
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.128727272727273e-06,
+      "loss": 3.8244,
+      "step": 1248
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.128e-06,
+      "loss": 3.8335,
+      "step": 1249
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.127272727272728e-06,
+      "loss": 3.8421,
+      "step": 1250
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.126545454545455e-06,
+      "loss": 3.9341,
+      "step": 1251
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.125818181818182e-06,
+      "loss": 3.7331,
+      "step": 1252
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.12509090909091e-06,
+      "loss": 3.8408,
+      "step": 1253
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.124363636363637e-06,
+      "loss": 3.8321,
+      "step": 1254
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.1236363636363645e-06,
+      "loss": 3.9264,
+      "step": 1255
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.122909090909091e-06,
+      "loss": 3.9633,
+      "step": 1256
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.122181818181818e-06,
+      "loss": 3.9406,
+      "step": 1257
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.121454545454546e-06,
+      "loss": 3.867,
+      "step": 1258
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.120727272727273e-06,
+      "loss": 3.9053,
+      "step": 1259
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.12e-06,
+      "loss": 3.8268,
+      "step": 1260
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.119272727272728e-06,
+      "loss": 3.9297,
+      "step": 1261
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.118545454545455e-06,
+      "loss": 3.8336,
+      "step": 1262
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1178181818181826e-06,
+      "loss": 3.8698,
+      "step": 1263
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.117090909090909e-06,
+      "loss": 3.9629,
+      "step": 1264
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.116363636363637e-06,
+      "loss": 3.9155,
+      "step": 1265
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.115636363636364e-06,
+      "loss": 3.8056,
+      "step": 1266
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.114909090909091e-06,
+      "loss": 3.7753,
+      "step": 1267
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.114181818181818e-06,
+      "loss": 3.9015,
+      "step": 1268
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.113454545454546e-06,
+      "loss": 3.7253,
+      "step": 1269
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.112727272727273e-06,
+      "loss": 3.7162,
+      "step": 1270
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.112000000000001e-06,
+      "loss": 3.8535,
+      "step": 1271
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.1112727272727274e-06,
+      "loss": 3.7332,
+      "step": 1272
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.110545454545455e-06,
+      "loss": 3.8009,
+      "step": 1273
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.109818181818182e-06,
+      "loss": 3.8877,
+      "step": 1274
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.10909090909091e-06,
+      "loss": 3.8011,
+      "step": 1275
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.1083636363636365e-06,
+      "loss": 3.9033,
+      "step": 1276
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.107636363636364e-06,
+      "loss": 3.7952,
+      "step": 1277
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.106909090909091e-06,
+      "loss": 3.9107,
+      "step": 1278
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.106181818181819e-06,
+      "loss": 3.9237,
+      "step": 1279
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.1054545454545455e-06,
+      "loss": 3.7911,
+      "step": 1280
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.104727272727273e-06,
+      "loss": 3.9034,
+      "step": 1281
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.104e-06,
+      "loss": 3.8529,
+      "step": 1282
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.103272727272728e-06,
+      "loss": 3.8138,
+      "step": 1283
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.1025454545454545e-06,
+      "loss": 3.6897,
+      "step": 1284
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.101818181818182e-06,
+      "loss": 3.711,
+      "step": 1285
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.10109090909091e-06,
+      "loss": 3.906,
+      "step": 1286
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.100363636363637e-06,
+      "loss": 3.9411,
+      "step": 1287
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.0996363636363635e-06,
+      "loss": 3.8118,
+      "step": 1288
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.098909090909091e-06,
+      "loss": 3.733,
+      "step": 1289
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.098181818181818e-06,
+      "loss": 3.8136,
+      "step": 1290
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.097454545454546e-06,
+      "loss": 3.8488,
+      "step": 1291
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.096727272727273e-06,
+      "loss": 3.8162,
+      "step": 1292
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.096e-06,
+      "loss": 3.7635,
+      "step": 1293
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.095272727272728e-06,
+      "loss": 3.804,
+      "step": 1294
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.094545454545455e-06,
+      "loss": 3.6552,
+      "step": 1295
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.0938181818181824e-06,
+      "loss": 3.8576,
+      "step": 1296
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.093090909090909e-06,
+      "loss": 3.8618,
+      "step": 1297
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.092363636363637e-06,
+      "loss": 3.7723,
+      "step": 1298
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.091636363636364e-06,
+      "loss": 3.8256,
+      "step": 1299
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.0909090909090915e-06,
+      "loss": 3.8161,
+      "step": 1300
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.090181818181818e-06,
+      "loss": 3.8302,
+      "step": 1301
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.089454545454546e-06,
+      "loss": 3.8031,
+      "step": 1302
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.088727272727273e-06,
+      "loss": 3.7976,
+      "step": 1303
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.0880000000000005e-06,
+      "loss": 3.8534,
+      "step": 1304
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.087272727272727e-06,
+      "loss": 3.8429,
+      "step": 1305
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.086545454545455e-06,
+      "loss": 3.7731,
+      "step": 1306
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.085818181818182e-06,
+      "loss": 3.8908,
+      "step": 1307
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.0850909090909095e-06,
+      "loss": 3.6212,
+      "step": 1308
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.084363636363637e-06,
+      "loss": 3.8486,
+      "step": 1309
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.083636363636364e-06,
+      "loss": 3.8039,
+      "step": 1310
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.082909090909091e-06,
+      "loss": 3.75,
+      "step": 1311
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.0821818181818186e-06,
+      "loss": 3.8053,
+      "step": 1312
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.081454545454545e-06,
+      "loss": 3.8159,
+      "step": 1313
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.080727272727273e-06,
+      "loss": 3.8689,
+      "step": 1314
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.08e-06,
+      "loss": 3.788,
+      "step": 1315
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.079272727272728e-06,
+      "loss": 3.8624,
+      "step": 1316
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.078545454545455e-06,
+      "loss": 3.6769,
+      "step": 1317
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.077818181818182e-06,
+      "loss": 3.7944,
+      "step": 1318
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.07709090909091e-06,
+      "loss": 3.7377,
+      "step": 1319
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.076363636363637e-06,
+      "loss": 3.7651,
+      "step": 1320
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.0756363636363634e-06,
+      "loss": 3.843,
+      "step": 1321
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.074909090909091e-06,
+      "loss": 3.8688,
+      "step": 1322
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.074181818181818e-06,
+      "loss": 3.8428,
+      "step": 1323
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.073454545454546e-06,
+      "loss": 3.8737,
+      "step": 1324
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.072727272727273e-06,
+      "loss": 3.8319,
+      "step": 1325
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.072e-06,
+      "loss": 3.8884,
+      "step": 1326
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.071272727272728e-06,
+      "loss": 3.8141,
+      "step": 1327
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.070545454545455e-06,
+      "loss": 3.8045,
+      "step": 1328
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.069818181818182e-06,
+      "loss": 3.7939,
+      "step": 1329
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.069090909090909e-06,
+      "loss": 3.8299,
+      "step": 1330
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.068363636363637e-06,
+      "loss": 3.859,
+      "step": 1331
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.0676363636363645e-06,
+      "loss": 3.8535,
+      "step": 1332
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.066909090909091e-06,
+      "loss": 3.8162,
+      "step": 1333
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.066181818181818e-06,
+      "loss": 3.8818,
+      "step": 1334
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.065454545454546e-06,
+      "loss": 3.8121,
+      "step": 1335
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.064727272727273e-06,
+      "loss": 3.8694,
+      "step": 1336
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.064e-06,
+      "loss": 3.7416,
+      "step": 1337
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.063272727272727e-06,
+      "loss": 3.8807,
+      "step": 1338
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.062545454545455e-06,
+      "loss": 3.7796,
+      "step": 1339
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.061818181818183e-06,
+      "loss": 3.759,
+      "step": 1340
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.061090909090909e-06,
+      "loss": 3.8564,
+      "step": 1341
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.060363636363637e-06,
+      "loss": 3.7685,
+      "step": 1342
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.059636363636364e-06,
+      "loss": 3.9033,
+      "step": 1343
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.058909090909091e-06,
+      "loss": 3.7912,
+      "step": 1344
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0581818181818184e-06,
+      "loss": 3.7496,
+      "step": 1345
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.057454545454545e-06,
+      "loss": 3.8889,
+      "step": 1346
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.056727272727273e-06,
+      "loss": 3.7803,
+      "step": 1347
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.056000000000001e-06,
+      "loss": 3.6842,
+      "step": 1348
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.0552727272727275e-06,
+      "loss": 3.8325,
+      "step": 1349
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.054545454545455e-06,
+      "loss": 3.7545,
+      "step": 1350
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.053818181818182e-06,
+      "loss": 3.7612,
+      "step": 1351
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.05309090909091e-06,
+      "loss": 3.8922,
+      "step": 1352
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.0523636363636365e-06,
+      "loss": 3.8527,
+      "step": 1353
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.051636363636363e-06,
+      "loss": 3.807,
+      "step": 1354
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.050909090909092e-06,
+      "loss": 3.9204,
+      "step": 1355
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.050181818181819e-06,
+      "loss": 3.8301,
+      "step": 1356
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.0494545454545455e-06,
+      "loss": 3.8764,
+      "step": 1357
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.048727272727273e-06,
+      "loss": 3.7478,
+      "step": 1358
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.048e-06,
+      "loss": 3.977,
+      "step": 1359
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.047272727272728e-06,
+      "loss": 3.9203,
+      "step": 1360
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.0465454545454546e-06,
+      "loss": 3.8857,
+      "step": 1361
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.045818181818182e-06,
+      "loss": 3.8449,
+      "step": 1362
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.04509090909091e-06,
+      "loss": 3.8616,
+      "step": 1363
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.044363636363637e-06,
+      "loss": 3.8276,
+      "step": 1364
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.0436363636363644e-06,
+      "loss": 3.7776,
+      "step": 1365
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.042909090909091e-06,
+      "loss": 3.7205,
+      "step": 1366
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.042181818181818e-06,
+      "loss": 3.8231,
+      "step": 1367
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.041454545454546e-06,
+      "loss": 3.7995,
+      "step": 1368
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.040727272727273e-06,
+      "loss": 3.7867,
+      "step": 1369
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.04e-06,
+      "loss": 3.8204,
+      "step": 1370
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.039272727272728e-06,
+      "loss": 3.8063,
+      "step": 1371
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.038545454545455e-06,
+      "loss": 3.7414,
+      "step": 1372
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.0378181818181825e-06,
+      "loss": 3.7792,
+      "step": 1373
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.037090909090909e-06,
+      "loss": 3.9355,
+      "step": 1374
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.036363636363637e-06,
+      "loss": 3.851,
+      "step": 1375
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.035636363636364e-06,
+      "loss": 3.8911,
+      "step": 1376
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.034909090909091e-06,
+      "loss": 3.9047,
+      "step": 1377
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.034181818181818e-06,
+      "loss": 3.8611,
+      "step": 1378
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.033454545454546e-06,
+      "loss": 3.8002,
+      "step": 1379
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.032727272727273e-06,
+      "loss": 3.7923,
+      "step": 1380
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.0320000000000005e-06,
+      "loss": 3.82,
+      "step": 1381
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.031272727272727e-06,
+      "loss": 3.7392,
+      "step": 1382
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.030545454545455e-06,
+      "loss": 3.8179,
+      "step": 1383
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.029818181818182e-06,
+      "loss": 3.8029,
+      "step": 1384
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.0290909090909096e-06,
+      "loss": 3.6694,
+      "step": 1385
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.0972972972972973,
+      "eval_loss": 3.7997894287109375,
+      "eval_runtime": 143.963,
+      "eval_samples_per_second": 2.57,
+      "eval_steps_per_second": 0.646,
+      "step": 1385
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.028363636363637e-06,
+      "loss": 3.9129,
+      "step": 1386
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.027636363636364e-06,
+      "loss": 3.9077,
+      "step": 1387
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.026909090909091e-06,
+      "loss": 3.7144,
+      "step": 1388
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.026181818181819e-06,
+      "loss": 3.8523,
+      "step": 1389
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.025454545454545e-06,
+      "loss": 3.7837,
+      "step": 1390
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.024727272727273e-06,
+      "loss": 3.8104,
+      "step": 1391
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.024e-06,
+      "loss": 3.7803,
+      "step": 1392
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.023272727272728e-06,
+      "loss": 3.7638,
+      "step": 1393
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.022545454545455e-06,
+      "loss": 3.6368,
+      "step": 1394
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.021818181818182e-06,
+      "loss": 3.8787,
+      "step": 1395
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.02109090909091e-06,
+      "loss": 3.7711,
+      "step": 1396
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.020363636363637e-06,
+      "loss": 3.8125,
+      "step": 1397
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.019636363636364e-06,
+      "loss": 3.8334,
+      "step": 1398
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.018909090909091e-06,
+      "loss": 3.7352,
+      "step": 1399
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.018181818181818e-06,
+      "loss": 3.6368,
+      "step": 1400
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.017454545454546e-06,
+      "loss": 3.7621,
+      "step": 1401
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.016727272727273e-06,
+      "loss": 3.7601,
+      "step": 1402
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.016e-06,
+      "loss": 3.7446,
+      "step": 1403
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.015272727272728e-06,
+      "loss": 3.8261,
+      "step": 1404
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.014545454545455e-06,
+      "loss": 3.8154,
+      "step": 1405
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.013818181818182e-06,
+      "loss": 3.6487,
+      "step": 1406
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.013090909090909e-06,
+      "loss": 3.7837,
+      "step": 1407
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.012363636363637e-06,
+      "loss": 3.7234,
+      "step": 1408
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.0116363636363646e-06,
+      "loss": 3.8936,
+      "step": 1409
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.010909090909091e-06,
+      "loss": 3.8535,
+      "step": 1410
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.010181818181818e-06,
+      "loss": 3.8725,
+      "step": 1411
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.009454545454546e-06,
+      "loss": 3.8642,
+      "step": 1412
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.008727272727273e-06,
+      "loss": 3.8115,
+      "step": 1413
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.008e-06,
+      "loss": 3.7944,
+      "step": 1414
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.007272727272727e-06,
+      "loss": 3.7067,
+      "step": 1415
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.006545454545455e-06,
+      "loss": 3.9332,
+      "step": 1416
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.005818181818183e-06,
+      "loss": 3.8142,
+      "step": 1417
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.0050909090909094e-06,
+      "loss": 3.7029,
+      "step": 1418
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.004363636363637e-06,
+      "loss": 3.8025,
+      "step": 1419
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.003636363636364e-06,
+      "loss": 3.7503,
+      "step": 1420
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.002909090909091e-06,
+      "loss": 3.9678,
+      "step": 1421
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.0021818181818185e-06,
+      "loss": 3.7892,
+      "step": 1422
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.001454545454545e-06,
+      "loss": 3.7112,
+      "step": 1423
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.000727272727273e-06,
+      "loss": 3.8135,
+      "step": 1424
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.8551,
+      "step": 1425
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.9992727272727275e-06,
+      "loss": 3.8309,
+      "step": 1426
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.998545454545455e-06,
+      "loss": 3.8579,
+      "step": 1427
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.997818181818182e-06,
+      "loss": 3.8326,
+      "step": 1428
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.99709090909091e-06,
+      "loss": 3.7553,
+      "step": 1429
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.9963636363636365e-06,
+      "loss": 3.8573,
+      "step": 1430
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.995636363636363e-06,
+      "loss": 3.9523,
+      "step": 1431
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.994909090909091e-06,
+      "loss": 3.788,
+      "step": 1432
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.994181818181819e-06,
+      "loss": 3.7729,
+      "step": 1433
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.9934545454545456e-06,
+      "loss": 3.9479,
+      "step": 1434
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.992727272727273e-06,
+      "loss": 3.9181,
+      "step": 1435
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.992e-06,
+      "loss": 3.7658,
+      "step": 1436
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.991272727272728e-06,
+      "loss": 3.8347,
+      "step": 1437
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.990545454545455e-06,
+      "loss": 3.8627,
+      "step": 1438
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.989818181818182e-06,
+      "loss": 3.785,
+      "step": 1439
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.98909090909091e-06,
+      "loss": 3.8882,
+      "step": 1440
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.988363636363637e-06,
+      "loss": 3.7881,
+      "step": 1441
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.9876363636363645e-06,
+      "loss": 3.6019,
+      "step": 1442
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.986909090909091e-06,
+      "loss": 3.8312,
+      "step": 1443
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.986181818181818e-06,
+      "loss": 3.6812,
+      "step": 1444
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.985454545454546e-06,
+      "loss": 3.8177,
+      "step": 1445
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.984727272727273e-06,
+      "loss": 3.8308,
+      "step": 1446
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.984e-06,
+      "loss": 3.8007,
+      "step": 1447
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.983272727272728e-06,
+      "loss": 3.8658,
+      "step": 1448
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.982545454545455e-06,
+      "loss": 3.8029,
+      "step": 1449
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.9818181818181825e-06,
+      "loss": 3.8808,
+      "step": 1450
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.981090909090909e-06,
+      "loss": 3.8743,
+      "step": 1451
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.980363636363637e-06,
+      "loss": 3.8559,
+      "step": 1452
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.979636363636364e-06,
+      "loss": 3.8788,
+      "step": 1453
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.978909090909091e-06,
+      "loss": 3.8688,
+      "step": 1454
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.978181818181818e-06,
+      "loss": 3.856,
+      "step": 1455
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.977454545454546e-06,
+      "loss": 3.8636,
+      "step": 1456
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.976727272727273e-06,
+      "loss": 3.7816,
+      "step": 1457
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.9760000000000006e-06,
+      "loss": 3.7323,
+      "step": 1458
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.975272727272727e-06,
+      "loss": 3.715,
+      "step": 1459
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.974545454545455e-06,
+      "loss": 3.8508,
+      "step": 1460
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.973818181818182e-06,
+      "loss": 3.8529,
+      "step": 1461
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.97309090909091e-06,
+      "loss": 3.8737,
+      "step": 1462
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.972363636363636e-06,
+      "loss": 3.8102,
+      "step": 1463
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.971636363636364e-06,
+      "loss": 3.7645,
+      "step": 1464
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.970909090909092e-06,
+      "loss": 3.8164,
+      "step": 1465
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.970181818181819e-06,
+      "loss": 3.85,
+      "step": 1466
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.9694545454545454e-06,
+      "loss": 3.7983,
+      "step": 1467
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.968727272727273e-06,
+      "loss": 3.7505,
+      "step": 1468
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.968e-06,
+      "loss": 3.8544,
+      "step": 1469
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.967272727272728e-06,
+      "loss": 3.7679,
+      "step": 1470
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.966545454545455e-06,
+      "loss": 3.7335,
+      "step": 1471
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.965818181818182e-06,
+      "loss": 3.7786,
+      "step": 1472
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.96509090909091e-06,
+      "loss": 3.7359,
+      "step": 1473
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.964363636363637e-06,
+      "loss": 3.7272,
+      "step": 1474
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.963636363636364e-06,
+      "loss": 3.9035,
+      "step": 1475
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.962909090909091e-06,
+      "loss": 3.8441,
+      "step": 1476
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.962181818181818e-06,
+      "loss": 3.9215,
+      "step": 1477
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.961454545454546e-06,
+      "loss": 3.8738,
+      "step": 1478
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.960727272727273e-06,
+      "loss": 3.8207,
+      "step": 1479
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.96e-06,
+      "loss": 3.7626,
+      "step": 1480
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.959272727272728e-06,
+      "loss": 3.6799,
+      "step": 1481
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.958545454545455e-06,
+      "loss": 3.7805,
+      "step": 1482
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.957818181818182e-06,
+      "loss": 3.784,
+      "step": 1483
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.957090909090909e-06,
+      "loss": 3.7597,
+      "step": 1484
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.956363636363637e-06,
+      "loss": 3.6881,
+      "step": 1485
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.955636363636364e-06,
+      "loss": 3.8488,
+      "step": 1486
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.954909090909091e-06,
+      "loss": 3.872,
+      "step": 1487
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.954181818181818e-06,
+      "loss": 3.796,
+      "step": 1488
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.953454545454546e-06,
+      "loss": 3.8167,
+      "step": 1489
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.952727272727273e-06,
+      "loss": 3.8719,
+      "step": 1490
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.9520000000000004e-06,
+      "loss": 3.6982,
+      "step": 1491
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.951272727272727e-06,
+      "loss": 3.7405,
+      "step": 1492
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.950545454545455e-06,
+      "loss": 3.8775,
+      "step": 1493
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.949818181818182e-06,
+      "loss": 3.6052,
+      "step": 1494
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.9490909090909095e-06,
+      "loss": 3.9574,
+      "step": 1495
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.948363636363637e-06,
+      "loss": 3.8416,
+      "step": 1496
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.947636363636364e-06,
+      "loss": 3.7362,
+      "step": 1497
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.946909090909092e-06,
+      "loss": 3.7041,
+      "step": 1498
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.9461818181818185e-06,
+      "loss": 3.8839,
+      "step": 1499
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.945454545454545e-06,
+      "loss": 3.8459,
+      "step": 1500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.944727272727273e-06,
+      "loss": 3.8364,
+      "step": 1501
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.944e-06,
+      "loss": 3.7752,
+      "step": 1502
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.9432727272727275e-06,
+      "loss": 3.8015,
+      "step": 1503
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.942545454545455e-06,
+      "loss": 3.8439,
+      "step": 1504
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.941818181818182e-06,
+      "loss": 3.8426,
+      "step": 1505
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.94109090909091e-06,
+      "loss": 3.6527,
+      "step": 1506
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.9403636363636366e-06,
+      "loss": 3.8205,
+      "step": 1507
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.939636363636364e-06,
+      "loss": 3.7864,
+      "step": 1508
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.938909090909091e-06,
+      "loss": 3.9143,
+      "step": 1509
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.938181818181819e-06,
+      "loss": 3.7792,
+      "step": 1510
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.937454545454546e-06,
+      "loss": 3.7708,
+      "step": 1511
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.936727272727273e-06,
+      "loss": 3.9455,
+      "step": 1512
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.936e-06,
+      "loss": 3.6294,
+      "step": 1513
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.935272727272728e-06,
+      "loss": 3.7419,
+      "step": 1514
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.934545454545455e-06,
+      "loss": 3.7253,
+      "step": 1515
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.933818181818182e-06,
+      "loss": 3.8862,
+      "step": 1516
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.933090909090909e-06,
+      "loss": 3.7374,
+      "step": 1517
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.932363636363637e-06,
+      "loss": 3.8794,
+      "step": 1518
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.9316363636363645e-06,
+      "loss": 3.7751,
+      "step": 1519
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.930909090909091e-06,
+      "loss": 3.8728,
+      "step": 1520
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.930181818181818e-06,
+      "loss": 3.8087,
+      "step": 1521
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.929454545454546e-06,
+      "loss": 3.705,
+      "step": 1522
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.928727272727273e-06,
+      "loss": 3.9175,
+      "step": 1523
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.928e-06,
+      "loss": 3.8371,
+      "step": 1524
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.927272727272727e-06,
+      "loss": 3.8443,
+      "step": 1525
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.926545454545455e-06,
+      "loss": 3.7371,
+      "step": 1526
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.9258181818181825e-06,
+      "loss": 3.9013,
+      "step": 1527
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.925090909090909e-06,
+      "loss": 3.8199,
+      "step": 1528
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.924363636363637e-06,
+      "loss": 3.802,
+      "step": 1529
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.923636363636364e-06,
+      "loss": 3.8059,
+      "step": 1530
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.922909090909091e-06,
+      "loss": 3.8321,
+      "step": 1531
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.922181818181818e-06,
+      "loss": 3.7718,
+      "step": 1532
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.921454545454545e-06,
+      "loss": 3.8209,
+      "step": 1533
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.920727272727273e-06,
+      "loss": 3.8591,
+      "step": 1534
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 3.6995,
+      "step": 1535
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.919272727272727e-06,
+      "loss": 3.7404,
+      "step": 1536
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.918545454545455e-06,
+      "loss": 3.7019,
+      "step": 1537
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.917818181818182e-06,
+      "loss": 3.7861,
+      "step": 1538
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.91709090909091e-06,
+      "loss": 3.6743,
+      "step": 1539
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.9163636363636364e-06,
+      "loss": 3.8073,
+      "step": 1540
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.915636363636363e-06,
+      "loss": 3.6931,
+      "step": 1541
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.914909090909092e-06,
+      "loss": 3.6763,
+      "step": 1542
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.914181818181819e-06,
+      "loss": 3.7658,
+      "step": 1543
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.9134545454545455e-06,
+      "loss": 3.7211,
+      "step": 1544
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.912727272727273e-06,
+      "loss": 3.9732,
+      "step": 1545
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.912e-06,
+      "loss": 3.6549,
+      "step": 1546
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.911272727272728e-06,
+      "loss": 3.7516,
+      "step": 1547
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.9105454545454545e-06,
+      "loss": 3.704,
+      "step": 1548
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.909818181818182e-06,
+      "loss": 3.8328,
+      "step": 1549
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.90909090909091e-06,
+      "loss": 3.7888,
+      "step": 1550
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.908363636363637e-06,
+      "loss": 3.846,
+      "step": 1551
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.907636363636364e-06,
+      "loss": 3.6095,
+      "step": 1552
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.906909090909091e-06,
+      "loss": 3.8314,
+      "step": 1553
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.906181818181818e-06,
+      "loss": 3.7698,
+      "step": 1554
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.905454545454546e-06,
+      "loss": 3.7381,
+      "step": 1555
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.9047272727272725e-06,
+      "loss": 3.7936,
+      "step": 1556
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.904e-06,
+      "loss": 3.8626,
+      "step": 1557
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.903272727272728e-06,
+      "loss": 3.8858,
+      "step": 1558
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.902545454545455e-06,
+      "loss": 3.6551,
+      "step": 1559
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.901818181818182e-06,
+      "loss": 3.6966,
+      "step": 1560
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.901090909090909e-06,
+      "loss": 3.7744,
+      "step": 1561
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.900363636363637e-06,
+      "loss": 3.9618,
+      "step": 1562
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.899636363636364e-06,
+      "loss": 3.8007,
+      "step": 1563
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.898909090909091e-06,
+      "loss": 3.841,
+      "step": 1564
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.898181818181819e-06,
+      "loss": 3.956,
+      "step": 1565
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.897454545454546e-06,
+      "loss": 3.7366,
+      "step": 1566
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.896727272727273e-06,
+      "loss": 3.8906,
+      "step": 1567
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.8960000000000005e-06,
+      "loss": 3.7343,
+      "step": 1568
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.895272727272727e-06,
+      "loss": 3.9309,
+      "step": 1569
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.894545454545455e-06,
+      "loss": 3.7579,
+      "step": 1570
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.893818181818182e-06,
+      "loss": 3.7617,
+      "step": 1571
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.8930909090909095e-06,
+      "loss": 3.7559,
+      "step": 1572
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.892363636363637e-06,
+      "loss": 3.8303,
+      "step": 1573
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.891636363636364e-06,
+      "loss": 3.7045,
+      "step": 1574
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.890909090909092e-06,
+      "loss": 3.7421,
+      "step": 1575
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.8901818181818185e-06,
+      "loss": 3.8482,
+      "step": 1576
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.889454545454545e-06,
+      "loss": 3.8427,
+      "step": 1577
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.888727272727273e-06,
+      "loss": 3.5941,
+      "step": 1578
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.888e-06,
+      "loss": 3.6549,
+      "step": 1579
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.8872727272727276e-06,
+      "loss": 3.7988,
+      "step": 1580
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.886545454545455e-06,
+      "loss": 3.7492,
+      "step": 1581
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.885818181818182e-06,
+      "loss": 3.7732,
+      "step": 1582
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.88509090909091e-06,
+      "loss": 3.6684,
+      "step": 1583
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.884363636363637e-06,
+      "loss": 3.8542,
+      "step": 1584
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.883636363636364e-06,
+      "loss": 3.8604,
+      "step": 1585
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.882909090909091e-06,
+      "loss": 3.6975,
+      "step": 1586
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.882181818181818e-06,
+      "loss": 3.8162,
+      "step": 1587
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.881454545454546e-06,
+      "loss": 3.8579,
+      "step": 1588
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.880727272727273e-06,
+      "loss": 3.7855,
+      "step": 1589
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.88e-06,
+      "loss": 3.7083,
+      "step": 1590
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.879272727272728e-06,
+      "loss": 3.7489,
+      "step": 1591
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.878545454545455e-06,
+      "loss": 3.8481,
+      "step": 1592
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.877818181818182e-06,
+      "loss": 3.7905,
+      "step": 1593
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.877090909090909e-06,
+      "loss": 3.8395,
+      "step": 1594
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.876363636363637e-06,
+      "loss": 3.7786,
+      "step": 1595
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.8756363636363645e-06,
+      "loss": 3.8084,
+      "step": 1596
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.874909090909091e-06,
+      "loss": 3.7826,
+      "step": 1597
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.874181818181819e-06,
+      "loss": 3.8484,
+      "step": 1598
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.873454545454546e-06,
+      "loss": 3.7437,
+      "step": 1599
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.872727272727273e-06,
+      "loss": 3.876,
+      "step": 1600
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.872e-06,
+      "loss": 3.7913,
+      "step": 1601
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.871272727272727e-06,
+      "loss": 3.7631,
+      "step": 1602
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.870545454545455e-06,
+      "loss": 3.6632,
+      "step": 1603
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.8698181818181826e-06,
+      "loss": 3.7325,
+      "step": 1604
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.869090909090909e-06,
+      "loss": 3.833,
+      "step": 1605
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.868363636363637e-06,
+      "loss": 3.8095,
+      "step": 1606
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.867636363636364e-06,
+      "loss": 3.8466,
+      "step": 1607
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.866909090909092e-06,
+      "loss": 3.7233,
+      "step": 1608
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.866181818181818e-06,
+      "loss": 3.8647,
+      "step": 1609
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.865454545454545e-06,
+      "loss": 3.8611,
+      "step": 1610
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.864727272727273e-06,
+      "loss": 3.8201,
+      "step": 1611
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.864000000000001e-06,
+      "loss": 3.7908,
+      "step": 1612
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.8632727272727274e-06,
+      "loss": 3.7476,
+      "step": 1613
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.862545454545455e-06,
+      "loss": 3.7757,
+      "step": 1614
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.861818181818182e-06,
+      "loss": 3.7222,
+      "step": 1615
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.86109090909091e-06,
+      "loss": 4.031,
+      "step": 1616
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.8603636363636365e-06,
+      "loss": 3.7599,
+      "step": 1617
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.859636363636364e-06,
+      "loss": 3.7309,
+      "step": 1618
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.858909090909092e-06,
+      "loss": 3.8043,
+      "step": 1619
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.858181818181819e-06,
+      "loss": 3.8961,
+      "step": 1620
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.8574545454545455e-06,
+      "loss": 3.7135,
+      "step": 1621
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.856727272727273e-06,
+      "loss": 3.7439,
+      "step": 1622
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.856e-06,
+      "loss": 3.6914,
+      "step": 1623
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.855272727272728e-06,
+      "loss": 3.6955,
+      "step": 1624
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.8545454545454545e-06,
+      "loss": 3.9264,
+      "step": 1625
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.853818181818182e-06,
+      "loss": 3.7631,
+      "step": 1626
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.85309090909091e-06,
+      "loss": 3.6958,
+      "step": 1627
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.852363636363637e-06,
+      "loss": 3.8053,
+      "step": 1628
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.851636363636364e-06,
+      "loss": 3.8275,
+      "step": 1629
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.850909090909091e-06,
+      "loss": 3.8382,
+      "step": 1630
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.850181818181818e-06,
+      "loss": 3.8574,
+      "step": 1631
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.849454545454546e-06,
+      "loss": 3.8358,
+      "step": 1632
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.848727272727273e-06,
+      "loss": 3.6566,
+      "step": 1633
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.848e-06,
+      "loss": 3.7698,
+      "step": 1634
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.847272727272728e-06,
+      "loss": 3.6251,
+      "step": 1635
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.846545454545455e-06,
+      "loss": 3.6785,
+      "step": 1636
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.8458181818181825e-06,
+      "loss": 3.7153,
+      "step": 1637
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.845090909090909e-06,
+      "loss": 3.6493,
+      "step": 1638
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.844363636363637e-06,
+      "loss": 3.7234,
+      "step": 1639
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.843636363636364e-06,
+      "loss": 3.7571,
+      "step": 1640
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.842909090909091e-06,
+      "loss": 3.7139,
+      "step": 1641
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.842181818181818e-06,
+      "loss": 3.6992,
+      "step": 1642
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.841454545454546e-06,
+      "loss": 3.748,
+      "step": 1643
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.840727272727273e-06,
+      "loss": 3.8039,
+      "step": 1644
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 3.834,
+      "step": 1645
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.839272727272727e-06,
+      "loss": 3.8681,
+      "step": 1646
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.838545454545455e-06,
+      "loss": 3.83,
+      "step": 1647
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.837818181818182e-06,
+      "loss": 3.7207,
+      "step": 1648
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.8370909090909095e-06,
+      "loss": 3.886,
+      "step": 1649
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.836363636363636e-06,
+      "loss": 3.677,
+      "step": 1650
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.835636363636364e-06,
+      "loss": 3.6675,
+      "step": 1651
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.834909090909092e-06,
+      "loss": 3.8198,
+      "step": 1652
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.8341818181818186e-06,
+      "loss": 3.6098,
+      "step": 1653
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.833454545454545e-06,
+      "loss": 3.741,
+      "step": 1654
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.832727272727273e-06,
+      "loss": 3.7613,
+      "step": 1655
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.832e-06,
+      "loss": 3.7991,
+      "step": 1656
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.831272727272728e-06,
+      "loss": 3.6874,
+      "step": 1657
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.830545454545455e-06,
+      "loss": 3.8348,
+      "step": 1658
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.829818181818182e-06,
+      "loss": 3.6058,
+      "step": 1659
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.82909090909091e-06,
+      "loss": 3.8653,
+      "step": 1660
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.828363636363637e-06,
+      "loss": 3.8235,
+      "step": 1661
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.827636363636364e-06,
+      "loss": 3.6185,
+      "step": 1662
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.0918918918918919,
+      "eval_loss": 3.7590231895446777,
+      "eval_runtime": 143.111,
+      "eval_samples_per_second": 2.585,
+      "eval_steps_per_second": 0.65,
+      "step": 1662
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.826909090909091e-06,
+      "loss": 3.7678,
+      "step": 1663
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.826181818181818e-06,
+      "loss": 3.8191,
+      "step": 1664
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.825454545454546e-06,
+      "loss": 3.671,
+      "step": 1665
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.824727272727273e-06,
+      "loss": 3.8745,
+      "step": 1666
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.824e-06,
+      "loss": 3.732,
+      "step": 1667
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.823272727272728e-06,
+      "loss": 3.8302,
+      "step": 1668
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.822545454545455e-06,
+      "loss": 3.7501,
+      "step": 1669
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.821818181818182e-06,
+      "loss": 3.8751,
+      "step": 1670
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.821090909090909e-06,
+      "loss": 3.6486,
+      "step": 1671
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.820363636363637e-06,
+      "loss": 3.7628,
+      "step": 1672
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.819636363636364e-06,
+      "loss": 3.7907,
+      "step": 1673
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.818909090909091e-06,
+      "loss": 3.7697,
+      "step": 1674
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.818181818181819e-06,
+      "loss": 3.6624,
+      "step": 1675
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.817454545454546e-06,
+      "loss": 3.7181,
+      "step": 1676
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.816727272727273e-06,
+      "loss": 3.7166,
+      "step": 1677
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.816e-06,
+      "loss": 3.7598,
+      "step": 1678
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.815272727272727e-06,
+      "loss": 3.8113,
+      "step": 1679
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.8145454545454545e-06,
+      "loss": 3.7942,
+      "step": 1680
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.8138181818181817e-06,
+      "loss": 3.7038,
+      "step": 1681
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.8130909090909094e-06,
+      "loss": 3.8192,
+      "step": 1682
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.8123636363636367e-06,
+      "loss": 3.7747,
+      "step": 1683
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.811636363636364e-06,
+      "loss": 3.7922,
+      "step": 1684
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.810909090909091e-06,
+      "loss": 3.676,
+      "step": 1685
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.8101818181818184e-06,
+      "loss": 3.7199,
+      "step": 1686
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.8094545454545457e-06,
+      "loss": 3.6877,
+      "step": 1687
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.808727272727273e-06,
+      "loss": 3.6888,
+      "step": 1688
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.8080000000000006e-06,
+      "loss": 3.7797,
+      "step": 1689
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.807272727272728e-06,
+      "loss": 3.9042,
+      "step": 1690
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.806545454545455e-06,
+      "loss": 3.8788,
+      "step": 1691
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.805818181818182e-06,
+      "loss": 3.8661,
+      "step": 1692
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.8050909090909092e-06,
+      "loss": 3.8058,
+      "step": 1693
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.8043636363636365e-06,
+      "loss": 3.7875,
+      "step": 1694
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.8036363636363638e-06,
+      "loss": 3.8995,
+      "step": 1695
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.802909090909091e-06,
+      "loss": 3.7972,
+      "step": 1696
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.8021818181818187e-06,
+      "loss": 3.7758,
+      "step": 1697
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.801454545454546e-06,
+      "loss": 3.7709,
+      "step": 1698
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.800727272727273e-06,
+      "loss": 3.6637,
+      "step": 1699
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 3.8765,
+      "step": 1700
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.7992727272727277e-06,
+      "loss": 3.7489,
+      "step": 1701
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.7985454545454546e-06,
+      "loss": 3.6835,
+      "step": 1702
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.797818181818182e-06,
+      "loss": 3.7537,
+      "step": 1703
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.797090909090909e-06,
+      "loss": 3.7428,
+      "step": 1704
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.7963636363636367e-06,
+      "loss": 3.8327,
+      "step": 1705
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.795636363636364e-06,
+      "loss": 3.7988,
+      "step": 1706
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.7949090909090913e-06,
+      "loss": 3.8584,
+      "step": 1707
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.7941818181818185e-06,
+      "loss": 3.6367,
+      "step": 1708
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.7934545454545458e-06,
+      "loss": 3.8353,
+      "step": 1709
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.792727272727273e-06,
+      "loss": 3.8406,
+      "step": 1710
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.7920000000000003e-06,
+      "loss": 3.8534,
+      "step": 1711
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.791272727272727e-06,
+      "loss": 3.714,
+      "step": 1712
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.7905454545454552e-06,
+      "loss": 3.6629,
+      "step": 1713
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.7898181818181825e-06,
+      "loss": 3.79,
+      "step": 1714
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.7890909090909093e-06,
+      "loss": 3.7298,
+      "step": 1715
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.7883636363636366e-06,
+      "loss": 3.8389,
+      "step": 1716
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.787636363636364e-06,
+      "loss": 3.6825,
+      "step": 1717
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.786909090909091e-06,
+      "loss": 3.7666,
+      "step": 1718
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.7861818181818183e-06,
+      "loss": 3.8867,
+      "step": 1719
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.7854545454545456e-06,
+      "loss": 3.7116,
+      "step": 1720
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.7847272727272733e-06,
+      "loss": 3.8001,
+      "step": 1721
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.7840000000000005e-06,
+      "loss": 3.7951,
+      "step": 1722
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.783272727272728e-06,
+      "loss": 3.8306,
+      "step": 1723
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.782545454545455e-06,
+      "loss": 3.796,
+      "step": 1724
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.781818181818182e-06,
+      "loss": 3.7739,
+      "step": 1725
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.781090909090909e-06,
+      "loss": 3.8753,
+      "step": 1726
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.7803636363636364e-06,
+      "loss": 3.8015,
+      "step": 1727
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.779636363636364e-06,
+      "loss": 3.8312,
+      "step": 1728
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.7789090909090913e-06,
+      "loss": 3.6974,
+      "step": 1729
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.7781818181818186e-06,
+      "loss": 3.8891,
+      "step": 1730
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.777454545454546e-06,
+      "loss": 3.7925,
+      "step": 1731
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.776727272727273e-06,
+      "loss": 3.7479,
+      "step": 1732
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.7760000000000004e-06,
+      "loss": 3.8235,
+      "step": 1733
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.7752727272727276e-06,
+      "loss": 3.745,
+      "step": 1734
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.7745454545454544e-06,
+      "loss": 3.6679,
+      "step": 1735
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.7738181818181825e-06,
+      "loss": 3.8192,
+      "step": 1736
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.7730909090909094e-06,
+      "loss": 3.7943,
+      "step": 1737
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.7723636363636366e-06,
+      "loss": 3.7213,
+      "step": 1738
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.771636363636364e-06,
+      "loss": 3.7928,
+      "step": 1739
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.770909090909091e-06,
+      "loss": 3.8005,
+      "step": 1740
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.7701818181818184e-06,
+      "loss": 3.7913,
+      "step": 1741
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.7694545454545457e-06,
+      "loss": 3.7058,
+      "step": 1742
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.768727272727273e-06,
+      "loss": 3.6987,
+      "step": 1743
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.7680000000000006e-06,
+      "loss": 3.7314,
+      "step": 1744
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.767272727272728e-06,
+      "loss": 3.7787,
+      "step": 1745
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.766545454545455e-06,
+      "loss": 3.8911,
+      "step": 1746
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.765818181818182e-06,
+      "loss": 3.7446,
+      "step": 1747
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.765090909090909e-06,
+      "loss": 3.8137,
+      "step": 1748
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.7643636363636365e-06,
+      "loss": 3.7153,
+      "step": 1749
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.7636363636363637e-06,
+      "loss": 3.8658,
+      "step": 1750
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.762909090909091e-06,
+      "loss": 3.7987,
+      "step": 1751
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.7621818181818187e-06,
+      "loss": 3.7062,
+      "step": 1752
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.761454545454546e-06,
+      "loss": 3.633,
+      "step": 1753
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.760727272727273e-06,
+      "loss": 3.7632,
+      "step": 1754
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 3.7145,
+      "step": 1755
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.7592727272727277e-06,
+      "loss": 3.7638,
+      "step": 1756
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.7585454545454545e-06,
+      "loss": 3.7993,
+      "step": 1757
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.7578181818181818e-06,
+      "loss": 3.7297,
+      "step": 1758
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.757090909090909e-06,
+      "loss": 3.7505,
+      "step": 1759
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.7563636363636367e-06,
+      "loss": 3.8419,
+      "step": 1760
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.755636363636364e-06,
+      "loss": 3.7747,
+      "step": 1761
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.7549090909090912e-06,
+      "loss": 3.7513,
+      "step": 1762
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.7541818181818185e-06,
+      "loss": 3.8351,
+      "step": 1763
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.7534545454545457e-06,
+      "loss": 3.6752,
+      "step": 1764
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.752727272727273e-06,
+      "loss": 3.6753,
+      "step": 1765
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.7520000000000002e-06,
+      "loss": 3.8933,
+      "step": 1766
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.751272727272728e-06,
+      "loss": 3.6348,
+      "step": 1767
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.750545454545455e-06,
+      "loss": 3.7785,
+      "step": 1768
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.7498181818181824e-06,
+      "loss": 3.7421,
+      "step": 1769
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.7490909090909093e-06,
+      "loss": 3.7024,
+      "step": 1770
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.7483636363636365e-06,
+      "loss": 3.8295,
+      "step": 1771
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.7476363636363638e-06,
+      "loss": 3.7945,
+      "step": 1772
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.746909090909091e-06,
+      "loss": 3.5753,
+      "step": 1773
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.7461818181818183e-06,
+      "loss": 3.8578,
+      "step": 1774
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.745454545454546e-06,
+      "loss": 3.7427,
+      "step": 1775
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.7447272727272732e-06,
+      "loss": 3.8012,
+      "step": 1776
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.7440000000000005e-06,
+      "loss": 3.7254,
+      "step": 1777
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.7432727272727277e-06,
+      "loss": 3.75,
+      "step": 1778
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.742545454545455e-06,
+      "loss": 3.6606,
+      "step": 1779
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.741818181818182e-06,
+      "loss": 3.7805,
+      "step": 1780
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.741090909090909e-06,
+      "loss": 3.9268,
+      "step": 1781
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.7403636363636364e-06,
+      "loss": 3.6454,
+      "step": 1782
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.739636363636364e-06,
+      "loss": 3.6613,
+      "step": 1783
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.7389090909090913e-06,
+      "loss": 3.7226,
+      "step": 1784
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.7381818181818185e-06,
+      "loss": 3.8334,
+      "step": 1785
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.737454545454546e-06,
+      "loss": 3.792,
+      "step": 1786
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.736727272727273e-06,
+      "loss": 3.837,
+      "step": 1787
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.7360000000000003e-06,
+      "loss": 3.8547,
+      "step": 1788
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.7352727272727276e-06,
+      "loss": 3.6605,
+      "step": 1789
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.7345454545454544e-06,
+      "loss": 3.7173,
+      "step": 1790
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.7338181818181825e-06,
+      "loss": 3.7884,
+      "step": 1791
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.7330909090909093e-06,
+      "loss": 3.616,
+      "step": 1792
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.7323636363636366e-06,
+      "loss": 3.8275,
+      "step": 1793
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.731636363636364e-06,
+      "loss": 3.7309,
+      "step": 1794
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.730909090909091e-06,
+      "loss": 3.8765,
+      "step": 1795
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.7301818181818184e-06,
+      "loss": 3.8897,
+      "step": 1796
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.7294545454545456e-06,
+      "loss": 3.6732,
+      "step": 1797
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.7287272727272733e-06,
+      "loss": 3.7775,
+      "step": 1798
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.7280000000000006e-06,
+      "loss": 3.7591,
+      "step": 1799
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.727272727272728e-06,
+      "loss": 3.6945,
+      "step": 1800
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.726545454545455e-06,
+      "loss": 3.7178,
+      "step": 1801
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.725818181818182e-06,
+      "loss": 3.7451,
+      "step": 1802
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.725090909090909e-06,
+      "loss": 3.8255,
+      "step": 1803
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.7243636363636364e-06,
+      "loss": 3.765,
+      "step": 1804
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.7236363636363637e-06,
+      "loss": 3.7754,
+      "step": 1805
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.7229090909090914e-06,
+      "loss": 3.757,
+      "step": 1806
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.7221818181818186e-06,
+      "loss": 3.8567,
+      "step": 1807
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.721454545454546e-06,
+      "loss": 3.6565,
+      "step": 1808
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.720727272727273e-06,
+      "loss": 3.7965,
+      "step": 1809
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 3.7908,
+      "step": 1810
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.7192727272727276e-06,
+      "loss": 3.7512,
+      "step": 1811
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.7185454545454545e-06,
+      "loss": 3.7349,
+      "step": 1812
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.7178181818181817e-06,
+      "loss": 3.883,
+      "step": 1813
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.71709090909091e-06,
+      "loss": 3.7983,
+      "step": 1814
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.7163636363636367e-06,
+      "loss": 3.683,
+      "step": 1815
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.715636363636364e-06,
+      "loss": 3.8797,
+      "step": 1816
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.714909090909091e-06,
+      "loss": 3.7627,
+      "step": 1817
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.7141818181818184e-06,
+      "loss": 3.7594,
+      "step": 1818
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.7134545454545457e-06,
+      "loss": 3.6267,
+      "step": 1819
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.712727272727273e-06,
+      "loss": 3.6754,
+      "step": 1820
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.712e-06,
+      "loss": 3.9435,
+      "step": 1821
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.711272727272728e-06,
+      "loss": 3.729,
+      "step": 1822
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.710545454545455e-06,
+      "loss": 3.683,
+      "step": 1823
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.7098181818181824e-06,
+      "loss": 3.8529,
+      "step": 1824
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.7090909090909092e-06,
+      "loss": 3.6543,
+      "step": 1825
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.7083636363636365e-06,
+      "loss": 3.7282,
+      "step": 1826
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.7076363636363637e-06,
+      "loss": 3.7038,
+      "step": 1827
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.706909090909091e-06,
+      "loss": 3.758,
+      "step": 1828
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.7061818181818183e-06,
+      "loss": 3.8035,
+      "step": 1829
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.705454545454546e-06,
+      "loss": 3.5989,
+      "step": 1830
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.704727272727273e-06,
+      "loss": 3.6283,
+      "step": 1831
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.7040000000000005e-06,
+      "loss": 3.7359,
+      "step": 1832
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.7032727272727277e-06,
+      "loss": 3.6762,
+      "step": 1833
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.702545454545455e-06,
+      "loss": 3.8718,
+      "step": 1834
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.701818181818182e-06,
+      "loss": 3.6872,
+      "step": 1835
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.701090909090909e-06,
+      "loss": 3.7932,
+      "step": 1836
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.7003636363636367e-06,
+      "loss": 3.6741,
+      "step": 1837
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.699636363636364e-06,
+      "loss": 3.7534,
+      "step": 1838
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.6989090909090912e-06,
+      "loss": 3.7359,
+      "step": 1839
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.6981818181818185e-06,
+      "loss": 3.7867,
+      "step": 1840
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.6974545454545458e-06,
+      "loss": 3.8821,
+      "step": 1841
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.696727272727273e-06,
+      "loss": 3.6366,
+      "step": 1842
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.6960000000000003e-06,
+      "loss": 3.723,
+      "step": 1843
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.6952727272727275e-06,
+      "loss": 3.7171,
+      "step": 1844
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.694545454545455e-06,
+      "loss": 3.6386,
+      "step": 1845
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.6938181818181825e-06,
+      "loss": 3.86,
+      "step": 1846
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.6930909090909093e-06,
+      "loss": 3.783,
+      "step": 1847
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.6923636363636366e-06,
+      "loss": 3.8403,
+      "step": 1848
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.691636363636364e-06,
+      "loss": 3.826,
+      "step": 1849
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.690909090909091e-06,
+      "loss": 3.7015,
+      "step": 1850
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.6901818181818183e-06,
+      "loss": 3.7343,
+      "step": 1851
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.6894545454545456e-06,
+      "loss": 3.7894,
+      "step": 1852
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.6887272727272733e-06,
+      "loss": 3.7084,
+      "step": 1853
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.6880000000000005e-06,
+      "loss": 3.7457,
+      "step": 1854
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.6872727272727278e-06,
+      "loss": 3.8046,
+      "step": 1855
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.686545454545455e-06,
+      "loss": 3.7606,
+      "step": 1856
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.685818181818182e-06,
+      "loss": 3.737,
+      "step": 1857
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.685090909090909e-06,
+      "loss": 3.6642,
+      "step": 1858
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.6843636363636364e-06,
+      "loss": 3.8296,
+      "step": 1859
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.6836363636363636e-06,
+      "loss": 3.78,
+      "step": 1860
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.6829090909090913e-06,
+      "loss": 3.6646,
+      "step": 1861
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.6821818181818186e-06,
+      "loss": 3.6721,
+      "step": 1862
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.681454545454546e-06,
+      "loss": 3.6228,
+      "step": 1863
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.680727272727273e-06,
+      "loss": 3.8498,
+      "step": 1864
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 3.7637,
+      "step": 1865
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.6792727272727276e-06,
+      "loss": 3.8001,
+      "step": 1866
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.6785454545454544e-06,
+      "loss": 3.8892,
+      "step": 1867
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.6778181818181817e-06,
+      "loss": 3.7697,
+      "step": 1868
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.67709090909091e-06,
+      "loss": 3.7709,
+      "step": 1869
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.6763636363636366e-06,
+      "loss": 3.8277,
+      "step": 1870
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.675636363636364e-06,
+      "loss": 3.7056,
+      "step": 1871
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.674909090909091e-06,
+      "loss": 3.8098,
+      "step": 1872
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.6741818181818184e-06,
+      "loss": 3.5832,
+      "step": 1873
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.6734545454545457e-06,
+      "loss": 3.8194,
+      "step": 1874
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.672727272727273e-06,
+      "loss": 3.57,
+      "step": 1875
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.6720000000000006e-06,
+      "loss": 3.8622,
+      "step": 1876
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.671272727272728e-06,
+      "loss": 3.7015,
+      "step": 1877
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.670545454545455e-06,
+      "loss": 3.7979,
+      "step": 1878
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.6698181818181824e-06,
+      "loss": 3.8305,
+      "step": 1879
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.669090909090909e-06,
+      "loss": 3.7393,
+      "step": 1880
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.6683636363636364e-06,
+      "loss": 3.5917,
+      "step": 1881
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.6676363636363637e-06,
+      "loss": 3.732,
+      "step": 1882
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.666909090909091e-06,
+      "loss": 3.5987,
+      "step": 1883
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.6661818181818186e-06,
+      "loss": 3.8811,
+      "step": 1884
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.665454545454546e-06,
+      "loss": 3.8484,
+      "step": 1885
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.664727272727273e-06,
+      "loss": 3.8114,
+      "step": 1886
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.6640000000000004e-06,
+      "loss": 3.8575,
+      "step": 1887
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.6632727272727277e-06,
+      "loss": 3.6955,
+      "step": 1888
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.662545454545455e-06,
+      "loss": 3.5746,
+      "step": 1889
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.6618181818181818e-06,
+      "loss": 3.7987,
+      "step": 1890
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.661090909090909e-06,
+      "loss": 3.7936,
+      "step": 1891
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.6603636363636367e-06,
+      "loss": 3.7971,
+      "step": 1892
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.659636363636364e-06,
+      "loss": 3.561,
+      "step": 1893
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.658909090909091e-06,
+      "loss": 3.7203,
+      "step": 1894
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.6581818181818185e-06,
+      "loss": 3.7144,
+      "step": 1895
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.6574545454545457e-06,
+      "loss": 3.7289,
+      "step": 1896
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.656727272727273e-06,
+      "loss": 3.7929,
+      "step": 1897
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.6560000000000002e-06,
+      "loss": 3.654,
+      "step": 1898
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.6552727272727275e-06,
+      "loss": 3.8333,
+      "step": 1899
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.654545454545455e-06,
+      "loss": 3.8613,
+      "step": 1900
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.6538181818181824e-06,
+      "loss": 3.7978,
+      "step": 1901
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.6530909090909093e-06,
+      "loss": 3.8106,
+      "step": 1902
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.6523636363636365e-06,
+      "loss": 3.9465,
+      "step": 1903
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.6516363636363638e-06,
+      "loss": 3.8576,
+      "step": 1904
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.650909090909091e-06,
+      "loss": 3.653,
+      "step": 1905
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.6501818181818183e-06,
+      "loss": 3.6894,
+      "step": 1906
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.649454545454546e-06,
+      "loss": 3.774,
+      "step": 1907
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.6487272727272732e-06,
+      "loss": 3.7415,
+      "step": 1908
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.6480000000000005e-06,
+      "loss": 3.7389,
+      "step": 1909
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.6472727272727277e-06,
+      "loss": 3.8053,
+      "step": 1910
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.646545454545455e-06,
+      "loss": 3.7988,
+      "step": 1911
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.645818181818182e-06,
+      "loss": 3.6918,
+      "step": 1912
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.645090909090909e-06,
+      "loss": 3.7693,
+      "step": 1913
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.6443636363636363e-06,
+      "loss": 3.6569,
+      "step": 1914
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.643636363636364e-06,
+      "loss": 3.8641,
+      "step": 1915
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.6429090909090913e-06,
+      "loss": 3.5218,
+      "step": 1916
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.6421818181818185e-06,
+      "loss": 3.6221,
+      "step": 1917
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.6414545454545458e-06,
+      "loss": 3.6483,
+      "step": 1918
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.640727272727273e-06,
+      "loss": 3.8629,
+      "step": 1919
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 3.7789,
+      "step": 1920
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.6392727272727276e-06,
+      "loss": 3.8317,
+      "step": 1921
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.638545454545455e-06,
+      "loss": 3.553,
+      "step": 1922
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.6378181818181825e-06,
+      "loss": 3.6989,
+      "step": 1923
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.6370909090909098e-06,
+      "loss": 3.7671,
+      "step": 1924
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 3.7224,
+      "step": 1925
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.635636363636364e-06,
+      "loss": 3.8105,
+      "step": 1926
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.634909090909091e-06,
+      "loss": 3.6896,
+      "step": 1927
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.6341818181818184e-06,
+      "loss": 3.8297,
+      "step": 1928
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.6334545454545456e-06,
+      "loss": 3.7186,
+      "step": 1929
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.632727272727273e-06,
+      "loss": 3.8079,
+      "step": 1930
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.6320000000000005e-06,
+      "loss": 3.7899,
+      "step": 1931
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.631272727272728e-06,
+      "loss": 3.6346,
+      "step": 1932
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.630545454545455e-06,
+      "loss": 3.8547,
+      "step": 1933
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.6298181818181823e-06,
+      "loss": 3.8532,
+      "step": 1934
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.629090909090909e-06,
+      "loss": 3.8701,
+      "step": 1935
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.6283636363636364e-06,
+      "loss": 3.6198,
+      "step": 1936
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.6276363636363637e-06,
+      "loss": 3.8679,
+      "step": 1937
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.626909090909091e-06,
+      "loss": 3.8545,
+      "step": 1938
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.6261818181818186e-06,
+      "loss": 3.8412,
+      "step": 1939
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.10810810810810811,
+      "eval_loss": 3.7326009273529053,
+      "eval_runtime": 144.011,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 0.646,
+      "step": 1939
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.625454545454546e-06,
+      "loss": 3.7392,
+      "step": 1940
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.624727272727273e-06,
+      "loss": 3.8086,
+      "step": 1941
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.6240000000000004e-06,
+      "loss": 3.7438,
+      "step": 1942
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.6232727272727276e-06,
+      "loss": 3.6714,
+      "step": 1943
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.622545454545455e-06,
+      "loss": 3.6929,
+      "step": 1944
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.6218181818181817e-06,
+      "loss": 3.7084,
+      "step": 1945
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.62109090909091e-06,
+      "loss": 3.8159,
+      "step": 1946
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.6203636363636367e-06,
+      "loss": 3.8495,
+      "step": 1947
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.619636363636364e-06,
+      "loss": 3.7632,
+      "step": 1948
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.618909090909091e-06,
+      "loss": 3.7583,
+      "step": 1949
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.6181818181818184e-06,
+      "loss": 3.6583,
+      "step": 1950
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.6174545454545457e-06,
+      "loss": 3.7697,
+      "step": 1951
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.616727272727273e-06,
+      "loss": 3.6458,
+      "step": 1952
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.616e-06,
+      "loss": 3.7771,
+      "step": 1953
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.615272727272728e-06,
+      "loss": 3.683,
+      "step": 1954
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.614545454545455e-06,
+      "loss": 3.7596,
+      "step": 1955
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.6138181818181824e-06,
+      "loss": 3.8439,
+      "step": 1956
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.6130909090909092e-06,
+      "loss": 3.8541,
+      "step": 1957
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.6123636363636365e-06,
+      "loss": 3.7604,
+      "step": 1958
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.6116363636363637e-06,
+      "loss": 3.9048,
+      "step": 1959
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.610909090909091e-06,
+      "loss": 3.7063,
+      "step": 1960
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.6101818181818182e-06,
+      "loss": 3.6582,
+      "step": 1961
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.609454545454546e-06,
+      "loss": 3.8501,
+      "step": 1962
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.608727272727273e-06,
+      "loss": 3.8491,
+      "step": 1963
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.6080000000000004e-06,
+      "loss": 3.7719,
+      "step": 1964
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.6072727272727277e-06,
+      "loss": 3.6693,
+      "step": 1965
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.606545454545455e-06,
+      "loss": 3.659,
+      "step": 1966
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.6058181818181818e-06,
+      "loss": 3.6973,
+      "step": 1967
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.605090909090909e-06,
+      "loss": 3.6377,
+      "step": 1968
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.6043636363636363e-06,
+      "loss": 3.6418,
+      "step": 1969
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.603636363636364e-06,
+      "loss": 3.8824,
+      "step": 1970
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.6029090909090912e-06,
+      "loss": 3.7032,
+      "step": 1971
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.6021818181818185e-06,
+      "loss": 3.7487,
+      "step": 1972
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.6014545454545457e-06,
+      "loss": 3.8653,
+      "step": 1973
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.600727272727273e-06,
+      "loss": 3.8275,
+      "step": 1974
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 3.749,
+      "step": 1975
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.5992727272727275e-06,
+      "loss": 3.7196,
+      "step": 1976
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.5985454545454548e-06,
+      "loss": 3.7933,
+      "step": 1977
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.5978181818181825e-06,
+      "loss": 3.7727,
+      "step": 1978
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.5970909090909097e-06,
+      "loss": 3.726,
+      "step": 1979
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.5963636363636365e-06,
+      "loss": 3.649,
+      "step": 1980
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.595636363636364e-06,
+      "loss": 3.7976,
+      "step": 1981
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.594909090909091e-06,
+      "loss": 3.5842,
+      "step": 1982
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.5941818181818183e-06,
+      "loss": 3.6653,
+      "step": 1983
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.5934545454545456e-06,
+      "loss": 3.6556,
+      "step": 1984
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.5927272727272733e-06,
+      "loss": 3.7332,
+      "step": 1985
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.5920000000000005e-06,
+      "loss": 3.8391,
+      "step": 1986
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.5912727272727278e-06,
+      "loss": 3.7375,
+      "step": 1987
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.590545454545455e-06,
+      "loss": 3.7842,
+      "step": 1988
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.5898181818181823e-06,
+      "loss": 3.7066,
+      "step": 1989
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.589090909090909e-06,
+      "loss": 3.7381,
+      "step": 1990
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.5883636363636364e-06,
+      "loss": 3.5969,
+      "step": 1991
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.5876363636363636e-06,
+      "loss": 3.624,
+      "step": 1992
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.5869090909090913e-06,
+      "loss": 3.6102,
+      "step": 1993
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.5861818181818186e-06,
+      "loss": 3.6375,
+      "step": 1994
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.585454545454546e-06,
+      "loss": 3.9282,
+      "step": 1995
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.584727272727273e-06,
+      "loss": 3.8047,
+      "step": 1996
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.5840000000000003e-06,
+      "loss": 3.7719,
+      "step": 1997
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.5832727272727276e-06,
+      "loss": 3.5818,
+      "step": 1998
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.582545454545455e-06,
+      "loss": 3.6219,
+      "step": 1999
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.5818181818181817e-06,
+      "loss": 3.7834,
+      "step": 2000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.5810909090909098e-06,
+      "loss": 3.8017,
+      "step": 2001
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.5803636363636366e-06,
+      "loss": 3.7464,
+      "step": 2002
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.579636363636364e-06,
+      "loss": 3.6174,
+      "step": 2003
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.578909090909091e-06,
+      "loss": 3.7991,
+      "step": 2004
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.5781818181818184e-06,
+      "loss": 3.6651,
+      "step": 2005
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.5774545454545456e-06,
+      "loss": 3.6545,
+      "step": 2006
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.576727272727273e-06,
+      "loss": 3.819,
+      "step": 2007
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.576e-06,
+      "loss": 3.8692,
+      "step": 2008
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.575272727272728e-06,
+      "loss": 3.8063,
+      "step": 2009
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.574545454545455e-06,
+      "loss": 3.7879,
+      "step": 2010
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.5738181818181823e-06,
+      "loss": 3.5509,
+      "step": 2011
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.573090909090909e-06,
+      "loss": 3.7757,
+      "step": 2012
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.5723636363636364e-06,
+      "loss": 3.8407,
+      "step": 2013
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.5716363636363637e-06,
+      "loss": 3.825,
+      "step": 2014
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.570909090909091e-06,
+      "loss": 3.7645,
+      "step": 2015
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.5701818181818186e-06,
+      "loss": 3.8209,
+      "step": 2016
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.569454545454546e-06,
+      "loss": 3.8275,
+      "step": 2017
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.568727272727273e-06,
+      "loss": 3.6514,
+      "step": 2018
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.5680000000000004e-06,
+      "loss": 3.6681,
+      "step": 2019
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.5672727272727277e-06,
+      "loss": 3.8574,
+      "step": 2020
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.566545454545455e-06,
+      "loss": 3.6706,
+      "step": 2021
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.565818181818182e-06,
+      "loss": 3.8904,
+      "step": 2022
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.565090909090909e-06,
+      "loss": 3.8857,
+      "step": 2023
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.564363636363637e-06,
+      "loss": 3.7985,
+      "step": 2024
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.563636363636364e-06,
+      "loss": 3.6389,
+      "step": 2025
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.562909090909091e-06,
+      "loss": 3.8081,
+      "step": 2026
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.5621818181818185e-06,
+      "loss": 3.721,
+      "step": 2027
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.5614545454545457e-06,
+      "loss": 3.7087,
+      "step": 2028
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.560727272727273e-06,
+      "loss": 3.6557,
+      "step": 2029
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 3.6317,
+      "step": 2030
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.5592727272727275e-06,
+      "loss": 3.7639,
+      "step": 2031
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.558545454545455e-06,
+      "loss": 3.7771,
+      "step": 2032
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.5578181818181824e-06,
+      "loss": 3.6947,
+      "step": 2033
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.5570909090909097e-06,
+      "loss": 3.8463,
+      "step": 2034
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.5563636363636365e-06,
+      "loss": 3.7798,
+      "step": 2035
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.5556363636363638e-06,
+      "loss": 3.6264,
+      "step": 2036
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.554909090909091e-06,
+      "loss": 3.6556,
+      "step": 2037
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.5541818181818183e-06,
+      "loss": 3.5502,
+      "step": 2038
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.5534545454545455e-06,
+      "loss": 3.6677,
+      "step": 2039
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.552727272727273e-06,
+      "loss": 3.5992,
+      "step": 2040
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.5520000000000005e-06,
+      "loss": 3.7624,
+      "step": 2041
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.5512727272727277e-06,
+      "loss": 3.8724,
+      "step": 2042
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.550545454545455e-06,
+      "loss": 3.8066,
+      "step": 2043
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.5498181818181822e-06,
+      "loss": 3.8462,
+      "step": 2044
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.549090909090909e-06,
+      "loss": 3.7211,
+      "step": 2045
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.5483636363636363e-06,
+      "loss": 3.7353,
+      "step": 2046
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.5476363636363636e-06,
+      "loss": 3.8584,
+      "step": 2047
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.5469090909090913e-06,
+      "loss": 3.7525,
+      "step": 2048
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.5461818181818185e-06,
+      "loss": 3.6846,
+      "step": 2049
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.5454545454545458e-06,
+      "loss": 3.7707,
+      "step": 2050
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.544727272727273e-06,
+      "loss": 3.9253,
+      "step": 2051
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.5440000000000003e-06,
+      "loss": 3.7136,
+      "step": 2052
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.5432727272727275e-06,
+      "loss": 3.6183,
+      "step": 2053
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.542545454545455e-06,
+      "loss": 3.6854,
+      "step": 2054
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.5418181818181825e-06,
+      "loss": 3.7942,
+      "step": 2055
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.5410909090909097e-06,
+      "loss": 3.8675,
+      "step": 2056
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.5403636363636366e-06,
+      "loss": 3.7718,
+      "step": 2057
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.539636363636364e-06,
+      "loss": 3.8042,
+      "step": 2058
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.538909090909091e-06,
+      "loss": 3.7775,
+      "step": 2059
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.5381818181818183e-06,
+      "loss": 3.7866,
+      "step": 2060
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.5374545454545456e-06,
+      "loss": 3.8646,
+      "step": 2061
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.536727272727273e-06,
+      "loss": 3.8975,
+      "step": 2062
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.5360000000000005e-06,
+      "loss": 3.83,
+      "step": 2063
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.535272727272728e-06,
+      "loss": 3.9052,
+      "step": 2064
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.534545454545455e-06,
+      "loss": 3.7255,
+      "step": 2065
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.5338181818181823e-06,
+      "loss": 3.8336,
+      "step": 2066
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.533090909090909e-06,
+      "loss": 3.7507,
+      "step": 2067
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.5323636363636364e-06,
+      "loss": 3.6565,
+      "step": 2068
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.5316363636363637e-06,
+      "loss": 3.6243,
+      "step": 2069
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.530909090909091e-06,
+      "loss": 3.6,
+      "step": 2070
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.5301818181818186e-06,
+      "loss": 3.835,
+      "step": 2071
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.529454545454546e-06,
+      "loss": 3.6688,
+      "step": 2072
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.528727272727273e-06,
+      "loss": 3.6426,
+      "step": 2073
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.5280000000000004e-06,
+      "loss": 3.7026,
+      "step": 2074
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.5272727272727276e-06,
+      "loss": 3.6227,
+      "step": 2075
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.526545454545455e-06,
+      "loss": 3.6946,
+      "step": 2076
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.525818181818182e-06,
+      "loss": 3.6443,
+      "step": 2077
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.525090909090909e-06,
+      "loss": 3.8036,
+      "step": 2078
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.524363636363637e-06,
+      "loss": 3.5937,
+      "step": 2079
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.523636363636364e-06,
+      "loss": 3.7797,
+      "step": 2080
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.522909090909091e-06,
+      "loss": 3.741,
+      "step": 2081
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.5221818181818184e-06,
+      "loss": 3.8721,
+      "step": 2082
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.5214545454545457e-06,
+      "loss": 3.6246,
+      "step": 2083
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.520727272727273e-06,
+      "loss": 3.6289,
+      "step": 2084
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.52e-06,
+      "loss": 3.7188,
+      "step": 2085
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.5192727272727274e-06,
+      "loss": 3.8132,
+      "step": 2086
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.518545454545455e-06,
+      "loss": 3.6016,
+      "step": 2087
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.5178181818181824e-06,
+      "loss": 3.6448,
+      "step": 2088
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.5170909090909096e-06,
+      "loss": 3.7012,
+      "step": 2089
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.5163636363636365e-06,
+      "loss": 3.8063,
+      "step": 2090
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.5156363636363637e-06,
+      "loss": 3.6651,
+      "step": 2091
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.514909090909091e-06,
+      "loss": 3.8914,
+      "step": 2092
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.5141818181818182e-06,
+      "loss": 3.7689,
+      "step": 2093
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.513454545454546e-06,
+      "loss": 3.8369,
+      "step": 2094
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.512727272727273e-06,
+      "loss": 3.6823,
+      "step": 2095
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.5120000000000004e-06,
+      "loss": 3.7111,
+      "step": 2096
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.5112727272727277e-06,
+      "loss": 3.6662,
+      "step": 2097
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.510545454545455e-06,
+      "loss": 3.6494,
+      "step": 2098
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.509818181818182e-06,
+      "loss": 3.6321,
+      "step": 2099
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.509090909090909e-06,
+      "loss": 3.7609,
+      "step": 2100
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.5083636363636363e-06,
+      "loss": 3.7308,
+      "step": 2101
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.507636363636364e-06,
+      "loss": 3.724,
+      "step": 2102
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.5069090909090912e-06,
+      "loss": 3.6688,
+      "step": 2103
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.5061818181818185e-06,
+      "loss": 3.6186,
+      "step": 2104
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.5054545454545457e-06,
+      "loss": 3.6242,
+      "step": 2105
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.504727272727273e-06,
+      "loss": 3.8369,
+      "step": 2106
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.5040000000000002e-06,
+      "loss": 3.8027,
+      "step": 2107
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.5032727272727275e-06,
+      "loss": 3.8462,
+      "step": 2108
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.5025454545454548e-06,
+      "loss": 3.799,
+      "step": 2109
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.5018181818181824e-06,
+      "loss": 3.7276,
+      "step": 2110
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.5010909090909097e-06,
+      "loss": 3.7279,
+      "step": 2111
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.5003636363636365e-06,
+      "loss": 3.7182,
+      "step": 2112
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.4996363636363638e-06,
+      "loss": 3.795,
+      "step": 2113
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.498909090909091e-06,
+      "loss": 3.7562,
+      "step": 2114
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.4981818181818183e-06,
+      "loss": 3.6848,
+      "step": 2115
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.4974545454545456e-06,
+      "loss": 3.6339,
+      "step": 2116
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.496727272727273e-06,
+      "loss": 3.5806,
+      "step": 2117
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.4960000000000005e-06,
+      "loss": 3.7459,
+      "step": 2118
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.4952727272727278e-06,
+      "loss": 3.7873,
+      "step": 2119
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.494545454545455e-06,
+      "loss": 3.7561,
+      "step": 2120
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4938181818181823e-06,
+      "loss": 3.6844,
+      "step": 2121
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4930909090909095e-06,
+      "loss": 3.7565,
+      "step": 2122
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4923636363636364e-06,
+      "loss": 3.6759,
+      "step": 2123
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.4916363636363636e-06,
+      "loss": 3.5669,
+      "step": 2124
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.4909090909090913e-06,
+      "loss": 3.6725,
+      "step": 2125
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.4901818181818185e-06,
+      "loss": 3.6854,
+      "step": 2126
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.489454545454546e-06,
+      "loss": 3.5979,
+      "step": 2127
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.488727272727273e-06,
+      "loss": 3.7274,
+      "step": 2128
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.4880000000000003e-06,
+      "loss": 3.7284,
+      "step": 2129
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.4872727272727276e-06,
+      "loss": 3.7258,
+      "step": 2130
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.486545454545455e-06,
+      "loss": 3.6689,
+      "step": 2131
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.485818181818182e-06,
+      "loss": 3.8194,
+      "step": 2132
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.4850909090909098e-06,
+      "loss": 3.6615,
+      "step": 2133
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.484363636363637e-06,
+      "loss": 3.5802,
+      "step": 2134
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.483636363636364e-06,
+      "loss": 3.6214,
+      "step": 2135
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.482909090909091e-06,
+      "loss": 3.7043,
+      "step": 2136
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.4821818181818184e-06,
+      "loss": 3.5572,
+      "step": 2137
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.4814545454545456e-06,
+      "loss": 3.81,
+      "step": 2138
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.480727272727273e-06,
+      "loss": 3.6975,
+      "step": 2139
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.48e-06,
+      "loss": 3.7012,
+      "step": 2140
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.479272727272728e-06,
+      "loss": 3.6468,
+      "step": 2141
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.478545454545455e-06,
+      "loss": 3.6925,
+      "step": 2142
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.4778181818181823e-06,
+      "loss": 3.6485,
+      "step": 2143
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.4770909090909096e-06,
+      "loss": 3.7474,
+      "step": 2144
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.4763636363636364e-06,
+      "loss": 3.8049,
+      "step": 2145
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.4756363636363637e-06,
+      "loss": 3.8122,
+      "step": 2146
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.474909090909091e-06,
+      "loss": 3.6597,
+      "step": 2147
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.474181818181818e-06,
+      "loss": 3.8012,
+      "step": 2148
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.473454545454546e-06,
+      "loss": 3.5068,
+      "step": 2149
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.472727272727273e-06,
+      "loss": 3.7402,
+      "step": 2150
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.4720000000000004e-06,
+      "loss": 3.8288,
+      "step": 2151
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.4712727272727276e-06,
+      "loss": 3.6819,
+      "step": 2152
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.470545454545455e-06,
+      "loss": 3.8367,
+      "step": 2153
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.469818181818182e-06,
+      "loss": 3.7473,
+      "step": 2154
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.469090909090909e-06,
+      "loss": 3.7765,
+      "step": 2155
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.4683636363636362e-06,
+      "loss": 3.8348,
+      "step": 2156
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.467636363636364e-06,
+      "loss": 3.9965,
+      "step": 2157
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.466909090909091e-06,
+      "loss": 3.5564,
+      "step": 2158
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.4661818181818184e-06,
+      "loss": 3.7962,
+      "step": 2159
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.4654545454545457e-06,
+      "loss": 3.6577,
+      "step": 2160
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.464727272727273e-06,
+      "loss": 3.824,
+      "step": 2161
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.464e-06,
+      "loss": 3.6487,
+      "step": 2162
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.4632727272727275e-06,
+      "loss": 3.7405,
+      "step": 2163
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.462545454545455e-06,
+      "loss": 3.5275,
+      "step": 2164
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.4618181818181824e-06,
+      "loss": 3.7014,
+      "step": 2165
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.4610909090909097e-06,
+      "loss": 3.6035,
+      "step": 2166
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.460363636363637e-06,
+      "loss": 3.7468,
+      "step": 2167
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.4596363636363637e-06,
+      "loss": 3.5891,
+      "step": 2168
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.458909090909091e-06,
+      "loss": 3.7707,
+      "step": 2169
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.4581818181818183e-06,
+      "loss": 3.7718,
+      "step": 2170
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.4574545454545455e-06,
+      "loss": 3.5864,
+      "step": 2171
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.456727272727273e-06,
+      "loss": 3.77,
+      "step": 2172
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.4560000000000005e-06,
+      "loss": 3.7372,
+      "step": 2173
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.4552727272727277e-06,
+      "loss": 3.6752,
+      "step": 2174
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.454545454545455e-06,
+      "loss": 3.7222,
+      "step": 2175
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.4538181818181822e-06,
+      "loss": 3.8281,
+      "step": 2176
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.4530909090909095e-06,
+      "loss": 3.6184,
+      "step": 2177
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.4523636363636363e-06,
+      "loss": 3.8646,
+      "step": 2178
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.4516363636363636e-06,
+      "loss": 3.602,
+      "step": 2179
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.4509090909090912e-06,
+      "loss": 3.7154,
+      "step": 2180
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.4501818181818185e-06,
+      "loss": 3.8077,
+      "step": 2181
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.4494545454545458e-06,
+      "loss": 3.8434,
+      "step": 2182
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.448727272727273e-06,
+      "loss": 3.4998,
+      "step": 2183
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.4480000000000003e-06,
+      "loss": 3.8239,
+      "step": 2184
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.4472727272727275e-06,
+      "loss": 3.7798,
+      "step": 2185
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.446545454545455e-06,
+      "loss": 3.601,
+      "step": 2186
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.445818181818182e-06,
+      "loss": 3.6518,
+      "step": 2187
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.4450909090909097e-06,
+      "loss": 3.6949,
+      "step": 2188
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.444363636363637e-06,
+      "loss": 3.8006,
+      "step": 2189
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.443636363636364e-06,
+      "loss": 3.8558,
+      "step": 2190
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.442909090909091e-06,
+      "loss": 3.7347,
+      "step": 2191
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.4421818181818183e-06,
+      "loss": 3.6791,
+      "step": 2192
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.4414545454545456e-06,
+      "loss": 3.7504,
+      "step": 2193
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.440727272727273e-06,
+      "loss": 3.5424,
+      "step": 2194
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.44e-06,
+      "loss": 3.8802,
+      "step": 2195
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.4392727272727278e-06,
+      "loss": 3.7134,
+      "step": 2196
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.438545454545455e-06,
+      "loss": 3.8307,
+      "step": 2197
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.4378181818181823e-06,
+      "loss": 3.7102,
+      "step": 2198
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.4370909090909095e-06,
+      "loss": 3.5967,
+      "step": 2199
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.4363636363636364e-06,
+      "loss": 3.7824,
+      "step": 2200
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.4356363636363636e-06,
+      "loss": 3.7522,
+      "step": 2201
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.434909090909091e-06,
+      "loss": 3.5887,
+      "step": 2202
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.4341818181818186e-06,
+      "loss": 3.7204,
+      "step": 2203
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.433454545454546e-06,
+      "loss": 3.6259,
+      "step": 2204
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.432727272727273e-06,
+      "loss": 3.8984,
+      "step": 2205
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.4320000000000003e-06,
+      "loss": 3.6854,
+      "step": 2206
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.4312727272727276e-06,
+      "loss": 3.5857,
+      "step": 2207
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.430545454545455e-06,
+      "loss": 3.7179,
+      "step": 2208
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.429818181818182e-06,
+      "loss": 3.67,
+      "step": 2209
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.429090909090909e-06,
+      "loss": 3.5897,
+      "step": 2210
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.428363636363637e-06,
+      "loss": 3.7265,
+      "step": 2211
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.427636363636364e-06,
+      "loss": 3.7372,
+      "step": 2212
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.426909090909091e-06,
+      "loss": 3.5444,
+      "step": 2213
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.4261818181818184e-06,
+      "loss": 3.5807,
+      "step": 2214
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.4254545454545457e-06,
+      "loss": 3.643,
+      "step": 2215
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.424727272727273e-06,
+      "loss": 3.6914,
+      "step": 2216
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.12162162162162163,
+      "eval_loss": 3.6862800121307373,
+      "eval_runtime": 142.8422,
+      "eval_samples_per_second": 2.59,
+      "eval_steps_per_second": 0.651,
+      "step": 2216
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.424e-06,
+      "loss": 3.6273,
+      "step": 2217
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.4232727272727274e-06,
+      "loss": 3.7575,
+      "step": 2218
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.422545454545455e-06,
+      "loss": 3.5874,
+      "step": 2219
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.4218181818181824e-06,
+      "loss": 3.5968,
+      "step": 2220
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.4210909090909096e-06,
+      "loss": 3.6468,
+      "step": 2221
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.420363636363637e-06,
+      "loss": 3.67,
+      "step": 2222
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.4196363636363637e-06,
+      "loss": 3.7215,
+      "step": 2223
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.418909090909091e-06,
+      "loss": 3.6532,
+      "step": 2224
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.4181818181818182e-06,
+      "loss": 3.8293,
+      "step": 2225
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.4174545454545455e-06,
+      "loss": 3.6828,
+      "step": 2226
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.416727272727273e-06,
+      "loss": 3.6636,
+      "step": 2227
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.4160000000000004e-06,
+      "loss": 3.7249,
+      "step": 2228
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.4152727272727277e-06,
+      "loss": 3.8532,
+      "step": 2229
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.414545454545455e-06,
+      "loss": 3.5725,
+      "step": 2230
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.413818181818182e-06,
+      "loss": 3.7043,
+      "step": 2231
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.4130909090909094e-06,
+      "loss": 3.5361,
+      "step": 2232
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.4123636363636363e-06,
+      "loss": 3.6609,
+      "step": 2233
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.4116363636363644e-06,
+      "loss": 3.5874,
+      "step": 2234
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.410909090909091e-06,
+      "loss": 3.5518,
+      "step": 2235
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.4101818181818185e-06,
+      "loss": 3.6885,
+      "step": 2236
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.4094545454545457e-06,
+      "loss": 3.8017,
+      "step": 2237
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.408727272727273e-06,
+      "loss": 3.4736,
+      "step": 2238
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.4080000000000002e-06,
+      "loss": 3.4692,
+      "step": 2239
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.4072727272727275e-06,
+      "loss": 3.5503,
+      "step": 2240
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.4065454545454547e-06,
+      "loss": 3.7268,
+      "step": 2241
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.4058181818181824e-06,
+      "loss": 3.9222,
+      "step": 2242
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.4050909090909097e-06,
+      "loss": 3.512,
+      "step": 2243
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.404363636363637e-06,
+      "loss": 3.6641,
+      "step": 2244
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.4036363636363638e-06,
+      "loss": 3.6901,
+      "step": 2245
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.402909090909091e-06,
+      "loss": 3.6191,
+      "step": 2246
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.4021818181818183e-06,
+      "loss": 3.5433,
+      "step": 2247
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.4014545454545455e-06,
+      "loss": 3.8109,
+      "step": 2248
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.400727272727273e-06,
+      "loss": 3.6631,
+      "step": 2249
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 3.7968,
+      "step": 2250
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.3992727272727277e-06,
+      "loss": 3.7565,
+      "step": 2251
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.398545454545455e-06,
+      "loss": 3.7103,
+      "step": 2252
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.3978181818181823e-06,
+      "loss": 3.5513,
+      "step": 2253
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.3970909090909095e-06,
+      "loss": 3.5964,
+      "step": 2254
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.3963636363636363e-06,
+      "loss": 3.563,
+      "step": 2255
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.3956363636363636e-06,
+      "loss": 3.683,
+      "step": 2256
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.394909090909091e-06,
+      "loss": 3.7219,
+      "step": 2257
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.3941818181818185e-06,
+      "loss": 3.6016,
+      "step": 2258
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.393454545454546e-06,
+      "loss": 3.8856,
+      "step": 2259
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.392727272727273e-06,
+      "loss": 3.7343,
+      "step": 2260
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.3920000000000003e-06,
+      "loss": 3.8528,
+      "step": 2261
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.3912727272727276e-06,
+      "loss": 3.6212,
+      "step": 2262
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.390545454545455e-06,
+      "loss": 3.7921,
+      "step": 2263
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.389818181818182e-06,
+      "loss": 3.7346,
+      "step": 2264
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.389090909090909e-06,
+      "loss": 3.7222,
+      "step": 2265
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.388363636363637e-06,
+      "loss": 3.6977,
+      "step": 2266
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.3876363636363643e-06,
+      "loss": 3.5766,
+      "step": 2267
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.386909090909091e-06,
+      "loss": 3.5687,
+      "step": 2268
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.3861818181818184e-06,
+      "loss": 3.663,
+      "step": 2269
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.3854545454545456e-06,
+      "loss": 3.8192,
+      "step": 2270
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.384727272727273e-06,
+      "loss": 3.8862,
+      "step": 2271
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.384e-06,
+      "loss": 3.8424,
+      "step": 2272
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.383272727272728e-06,
+      "loss": 3.5562,
+      "step": 2273
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.382545454545455e-06,
+      "loss": 3.7642,
+      "step": 2274
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.3818181818181823e-06,
+      "loss": 3.872,
+      "step": 2275
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.3810909090909096e-06,
+      "loss": 3.6395,
+      "step": 2276
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.380363636363637e-06,
+      "loss": 3.6272,
+      "step": 2277
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.3796363636363637e-06,
+      "loss": 3.55,
+      "step": 2278
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.378909090909091e-06,
+      "loss": 3.7077,
+      "step": 2279
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.378181818181818e-06,
+      "loss": 3.6008,
+      "step": 2280
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.377454545454546e-06,
+      "loss": 3.8535,
+      "step": 2281
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.376727272727273e-06,
+      "loss": 3.7194,
+      "step": 2282
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.3760000000000004e-06,
+      "loss": 3.8056,
+      "step": 2283
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.3752727272727276e-06,
+      "loss": 3.6137,
+      "step": 2284
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.374545454545455e-06,
+      "loss": 3.6043,
+      "step": 2285
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.373818181818182e-06,
+      "loss": 3.6649,
+      "step": 2286
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.3730909090909094e-06,
+      "loss": 3.6672,
+      "step": 2287
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.3723636363636362e-06,
+      "loss": 3.8549,
+      "step": 2288
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.3716363636363643e-06,
+      "loss": 3.688,
+      "step": 2289
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.370909090909091e-06,
+      "loss": 3.6753,
+      "step": 2290
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.3701818181818184e-06,
+      "loss": 3.7796,
+      "step": 2291
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.3694545454545457e-06,
+      "loss": 3.6853,
+      "step": 2292
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.368727272727273e-06,
+      "loss": 3.7515,
+      "step": 2293
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.368e-06,
+      "loss": 3.6046,
+      "step": 2294
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.3672727272727275e-06,
+      "loss": 3.6577,
+      "step": 2295
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.3665454545454547e-06,
+      "loss": 3.6227,
+      "step": 2296
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.3658181818181824e-06,
+      "loss": 3.5712,
+      "step": 2297
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.3650909090909096e-06,
+      "loss": 3.83,
+      "step": 2298
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.364363636363637e-06,
+      "loss": 3.8489,
+      "step": 2299
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.3636363636363637e-06,
+      "loss": 3.6126,
+      "step": 2300
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.362909090909091e-06,
+      "loss": 3.6945,
+      "step": 2301
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.3621818181818182e-06,
+      "loss": 3.6626,
+      "step": 2302
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.3614545454545455e-06,
+      "loss": 3.6598,
+      "step": 2303
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.3607272727272728e-06,
+      "loss": 3.7248,
+      "step": 2304
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 3.7497,
+      "step": 2305
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.3592727272727277e-06,
+      "loss": 3.6787,
+      "step": 2306
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.358545454545455e-06,
+      "loss": 3.8288,
+      "step": 2307
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.357818181818182e-06,
+      "loss": 3.695,
+      "step": 2308
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.3570909090909095e-06,
+      "loss": 3.5579,
+      "step": 2309
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.3563636363636363e-06,
+      "loss": 3.6412,
+      "step": 2310
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.3556363636363636e-06,
+      "loss": 3.513,
+      "step": 2311
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.3549090909090912e-06,
+      "loss": 3.5828,
+      "step": 2312
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.3541818181818185e-06,
+      "loss": 3.7185,
+      "step": 2313
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.3534545454545458e-06,
+      "loss": 3.6412,
+      "step": 2314
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.352727272727273e-06,
+      "loss": 3.7591,
+      "step": 2315
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.3520000000000003e-06,
+      "loss": 3.8685,
+      "step": 2316
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.3512727272727275e-06,
+      "loss": 3.7537,
+      "step": 2317
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.3505454545454548e-06,
+      "loss": 3.579,
+      "step": 2318
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.349818181818182e-06,
+      "loss": 3.7407,
+      "step": 2319
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.3490909090909097e-06,
+      "loss": 3.7021,
+      "step": 2320
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.348363636363637e-06,
+      "loss": 3.8353,
+      "step": 2321
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.3476363636363642e-06,
+      "loss": 3.6066,
+      "step": 2322
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.346909090909091e-06,
+      "loss": 3.6125,
+      "step": 2323
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.3461818181818183e-06,
+      "loss": 3.6783,
+      "step": 2324
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.3454545454545456e-06,
+      "loss": 3.5054,
+      "step": 2325
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.344727272727273e-06,
+      "loss": 3.7961,
+      "step": 2326
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.344e-06,
+      "loss": 3.5692,
+      "step": 2327
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.3432727272727278e-06,
+      "loss": 3.5569,
+      "step": 2328
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.342545454545455e-06,
+      "loss": 3.7375,
+      "step": 2329
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.3418181818181823e-06,
+      "loss": 3.9403,
+      "step": 2330
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.3410909090909095e-06,
+      "loss": 3.7985,
+      "step": 2331
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.340363636363637e-06,
+      "loss": 3.6657,
+      "step": 2332
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.3396363636363636e-06,
+      "loss": 3.7141,
+      "step": 2333
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.338909090909091e-06,
+      "loss": 3.7133,
+      "step": 2334
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.338181818181818e-06,
+      "loss": 3.7469,
+      "step": 2335
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.337454545454546e-06,
+      "loss": 3.6069,
+      "step": 2336
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.336727272727273e-06,
+      "loss": 3.5157,
+      "step": 2337
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.3360000000000003e-06,
+      "loss": 3.9055,
+      "step": 2338
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.3352727272727276e-06,
+      "loss": 3.7414,
+      "step": 2339
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.334545454545455e-06,
+      "loss": 3.659,
+      "step": 2340
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.333818181818182e-06,
+      "loss": 3.7067,
+      "step": 2341
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.3330909090909094e-06,
+      "loss": 3.7856,
+      "step": 2342
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.332363636363637e-06,
+      "loss": 3.6486,
+      "step": 2343
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.3316363636363643e-06,
+      "loss": 3.8589,
+      "step": 2344
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.330909090909091e-06,
+      "loss": 3.6051,
+      "step": 2345
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.3301818181818184e-06,
+      "loss": 3.811,
+      "step": 2346
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.3294545454545456e-06,
+      "loss": 3.8319,
+      "step": 2347
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.328727272727273e-06,
+      "loss": 3.6632,
+      "step": 2348
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.328e-06,
+      "loss": 3.7334,
+      "step": 2349
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.3272727272727274e-06,
+      "loss": 3.6795,
+      "step": 2350
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.326545454545455e-06,
+      "loss": 3.6964,
+      "step": 2351
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.3258181818181823e-06,
+      "loss": 3.7309,
+      "step": 2352
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.3250909090909096e-06,
+      "loss": 3.7439,
+      "step": 2353
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.324363636363637e-06,
+      "loss": 3.6285,
+      "step": 2354
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.3236363636363637e-06,
+      "loss": 3.7435,
+      "step": 2355
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.322909090909091e-06,
+      "loss": 3.8403,
+      "step": 2356
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.322181818181818e-06,
+      "loss": 3.8491,
+      "step": 2357
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.3214545454545455e-06,
+      "loss": 3.5483,
+      "step": 2358
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.320727272727273e-06,
+      "loss": 3.6314,
+      "step": 2359
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 3.6775,
+      "step": 2360
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.3192727272727277e-06,
+      "loss": 3.7373,
+      "step": 2361
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.318545454545455e-06,
+      "loss": 3.7288,
+      "step": 2362
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.317818181818182e-06,
+      "loss": 3.5963,
+      "step": 2363
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.3170909090909094e-06,
+      "loss": 3.6258,
+      "step": 2364
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.3163636363636363e-06,
+      "loss": 3.6937,
+      "step": 2365
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.3156363636363635e-06,
+      "loss": 3.6676,
+      "step": 2366
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.3149090909090916e-06,
+      "loss": 3.596,
+      "step": 2367
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.3141818181818185e-06,
+      "loss": 3.82,
+      "step": 2368
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.3134545454545457e-06,
+      "loss": 3.7912,
+      "step": 2369
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.312727272727273e-06,
+      "loss": 3.7044,
+      "step": 2370
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.3120000000000002e-06,
+      "loss": 3.9306,
+      "step": 2371
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.3112727272727275e-06,
+      "loss": 3.7805,
+      "step": 2372
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.3105454545454547e-06,
+      "loss": 3.7928,
+      "step": 2373
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.309818181818182e-06,
+      "loss": 3.6229,
+      "step": 2374
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.3090909090909097e-06,
+      "loss": 3.8397,
+      "step": 2375
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.308363636363637e-06,
+      "loss": 3.5027,
+      "step": 2376
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.307636363636364e-06,
+      "loss": 3.7552,
+      "step": 2377
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.306909090909091e-06,
+      "loss": 3.5208,
+      "step": 2378
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.3061818181818183e-06,
+      "loss": 3.7203,
+      "step": 2379
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.3054545454545455e-06,
+      "loss": 3.6223,
+      "step": 2380
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.304727272727273e-06,
+      "loss": 3.8405,
+      "step": 2381
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.3040000000000005e-06,
+      "loss": 3.6266,
+      "step": 2382
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.3032727272727277e-06,
+      "loss": 3.6721,
+      "step": 2383
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.302545454545455e-06,
+      "loss": 3.6243,
+      "step": 2384
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.3018181818181822e-06,
+      "loss": 3.4311,
+      "step": 2385
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.3010909090909095e-06,
+      "loss": 3.8252,
+      "step": 2386
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.3003636363636368e-06,
+      "loss": 3.6829,
+      "step": 2387
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.2996363636363636e-06,
+      "loss": 3.567,
+      "step": 2388
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.298909090909091e-06,
+      "loss": 3.7176,
+      "step": 2389
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.2981818181818185e-06,
+      "loss": 3.8425,
+      "step": 2390
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.2974545454545458e-06,
+      "loss": 3.7026,
+      "step": 2391
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.296727272727273e-06,
+      "loss": 3.6751,
+      "step": 2392
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.2960000000000003e-06,
+      "loss": 3.6795,
+      "step": 2393
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.2952727272727275e-06,
+      "loss": 3.7621,
+      "step": 2394
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.294545454545455e-06,
+      "loss": 3.7134,
+      "step": 2395
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.293818181818182e-06,
+      "loss": 3.7228,
+      "step": 2396
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.2930909090909093e-06,
+      "loss": 3.7552,
+      "step": 2397
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.292363636363637e-06,
+      "loss": 3.8049,
+      "step": 2398
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.2916363636363643e-06,
+      "loss": 3.7118,
+      "step": 2399
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.290909090909091e-06,
+      "loss": 3.6747,
+      "step": 2400
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.2901818181818183e-06,
+      "loss": 3.7417,
+      "step": 2401
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.2894545454545456e-06,
+      "loss": 3.6048,
+      "step": 2402
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.288727272727273e-06,
+      "loss": 3.7003,
+      "step": 2403
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.288e-06,
+      "loss": 3.5488,
+      "step": 2404
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.2872727272727274e-06,
+      "loss": 3.6874,
+      "step": 2405
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.286545454545455e-06,
+      "loss": 3.8512,
+      "step": 2406
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.2858181818181823e-06,
+      "loss": 3.9857,
+      "step": 2407
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.2850909090909096e-06,
+      "loss": 3.5405,
+      "step": 2408
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.284363636363637e-06,
+      "loss": 3.5602,
+      "step": 2409
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.2836363636363637e-06,
+      "loss": 3.6891,
+      "step": 2410
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.282909090909091e-06,
+      "loss": 3.683,
+      "step": 2411
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.282181818181818e-06,
+      "loss": 3.6067,
+      "step": 2412
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.2814545454545454e-06,
+      "loss": 3.661,
+      "step": 2413
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.280727272727273e-06,
+      "loss": 3.7555,
+      "step": 2414
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 3.7069,
+      "step": 2415
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.2792727272727276e-06,
+      "loss": 3.7269,
+      "step": 2416
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.278545454545455e-06,
+      "loss": 3.8288,
+      "step": 2417
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.277818181818182e-06,
+      "loss": 3.6514,
+      "step": 2418
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.2770909090909094e-06,
+      "loss": 3.6277,
+      "step": 2419
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.2763636363636362e-06,
+      "loss": 3.6307,
+      "step": 2420
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.2756363636363643e-06,
+      "loss": 4.045,
+      "step": 2421
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.2749090909090916e-06,
+      "loss": 3.6467,
+      "step": 2422
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.2741818181818184e-06,
+      "loss": 3.739,
+      "step": 2423
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.2734545454545457e-06,
+      "loss": 3.6809,
+      "step": 2424
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.272727272727273e-06,
+      "loss": 3.7242,
+      "step": 2425
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.272e-06,
+      "loss": 3.7024,
+      "step": 2426
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.2712727272727274e-06,
+      "loss": 3.7126,
+      "step": 2427
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.2705454545454547e-06,
+      "loss": 3.7858,
+      "step": 2428
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.2698181818181824e-06,
+      "loss": 3.6297,
+      "step": 2429
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.2690909090909096e-06,
+      "loss": 3.832,
+      "step": 2430
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.268363636363637e-06,
+      "loss": 3.4995,
+      "step": 2431
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.267636363636364e-06,
+      "loss": 3.6423,
+      "step": 2432
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.266909090909091e-06,
+      "loss": 3.6117,
+      "step": 2433
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.2661818181818182e-06,
+      "loss": 3.7941,
+      "step": 2434
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.2654545454545455e-06,
+      "loss": 3.8083,
+      "step": 2435
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.2647272727272727e-06,
+      "loss": 3.6127,
+      "step": 2436
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.2640000000000004e-06,
+      "loss": 3.7588,
+      "step": 2437
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.2632727272727277e-06,
+      "loss": 3.6757,
+      "step": 2438
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.262545454545455e-06,
+      "loss": 3.766,
+      "step": 2439
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.261818181818182e-06,
+      "loss": 3.9261,
+      "step": 2440
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.2610909090909095e-06,
+      "loss": 3.6546,
+      "step": 2441
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.2603636363636367e-06,
+      "loss": 3.7303,
+      "step": 2442
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.2596363636363635e-06,
+      "loss": 3.5772,
+      "step": 2443
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.258909090909091e-06,
+      "loss": 3.6017,
+      "step": 2444
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.2581818181818185e-06,
+      "loss": 3.6299,
+      "step": 2445
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.2574545454545457e-06,
+      "loss": 3.869,
+      "step": 2446
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.256727272727273e-06,
+      "loss": 3.7199,
+      "step": 2447
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.2560000000000003e-06,
+      "loss": 3.5936,
+      "step": 2448
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.2552727272727275e-06,
+      "loss": 3.8643,
+      "step": 2449
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.2545454545454548e-06,
+      "loss": 3.571,
+      "step": 2450
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.253818181818182e-06,
+      "loss": 3.7584,
+      "step": 2451
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.2530909090909097e-06,
+      "loss": 3.6422,
+      "step": 2452
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.252363636363637e-06,
+      "loss": 3.6078,
+      "step": 2453
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.2516363636363642e-06,
+      "loss": 3.9067,
+      "step": 2454
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.250909090909091e-06,
+      "loss": 3.6835,
+      "step": 2455
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.2501818181818183e-06,
+      "loss": 3.9195,
+      "step": 2456
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.2494545454545456e-06,
+      "loss": 3.5601,
+      "step": 2457
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.248727272727273e-06,
+      "loss": 3.8337,
+      "step": 2458
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.248e-06,
+      "loss": 3.62,
+      "step": 2459
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.2472727272727278e-06,
+      "loss": 3.6061,
+      "step": 2460
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.246545454545455e-06,
+      "loss": 3.679,
+      "step": 2461
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.2458181818181823e-06,
+      "loss": 3.6331,
+      "step": 2462
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.2450909090909095e-06,
+      "loss": 3.5539,
+      "step": 2463
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.2443636363636368e-06,
+      "loss": 3.8261,
+      "step": 2464
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.2436363636363636e-06,
+      "loss": 3.8752,
+      "step": 2465
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.242909090909091e-06,
+      "loss": 3.5892,
+      "step": 2466
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.242181818181818e-06,
+      "loss": 3.612,
+      "step": 2467
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.241454545454546e-06,
+      "loss": 3.7913,
+      "step": 2468
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.240727272727273e-06,
+      "loss": 3.7264,
+      "step": 2469
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 3.6785,
+      "step": 2470
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.2392727272727276e-06,
+      "loss": 3.6349,
+      "step": 2471
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.238545454545455e-06,
+      "loss": 3.7605,
+      "step": 2472
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.237818181818182e-06,
+      "loss": 3.6108,
+      "step": 2473
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.2370909090909093e-06,
+      "loss": 3.8214,
+      "step": 2474
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.236363636363636e-06,
+      "loss": 3.8446,
+      "step": 2475
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.2356363636363643e-06,
+      "loss": 3.6888,
+      "step": 2476
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.2349090909090915e-06,
+      "loss": 3.754,
+      "step": 2477
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.2341818181818184e-06,
+      "loss": 3.645,
+      "step": 2478
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.2334545454545456e-06,
+      "loss": 3.6024,
+      "step": 2479
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.232727272727273e-06,
+      "loss": 3.8636,
+      "step": 2480
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.232e-06,
+      "loss": 3.7215,
+      "step": 2481
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.2312727272727274e-06,
+      "loss": 3.7719,
+      "step": 2482
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.2305454545454547e-06,
+      "loss": 3.6109,
+      "step": 2483
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.2298181818181823e-06,
+      "loss": 3.6192,
+      "step": 2484
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.2290909090909096e-06,
+      "loss": 3.7946,
+      "step": 2485
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.228363636363637e-06,
+      "loss": 3.7159,
+      "step": 2486
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.227636363636364e-06,
+      "loss": 3.4434,
+      "step": 2487
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.226909090909091e-06,
+      "loss": 3.622,
+      "step": 2488
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.226181818181818e-06,
+      "loss": 3.8824,
+      "step": 2489
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.2254545454545455e-06,
+      "loss": 3.7658,
+      "step": 2490
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.224727272727273e-06,
+      "loss": 3.661,
+      "step": 2491
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2240000000000004e-06,
+      "loss": 3.6602,
+      "step": 2492
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2232727272727276e-06,
+      "loss": 3.5162,
+      "step": 2493
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.10270270270270271,
+      "eval_loss": 3.652031660079956,
+      "eval_runtime": 143.5263,
+      "eval_samples_per_second": 2.578,
+      "eval_steps_per_second": 0.648,
+      "step": 2493
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.222545454545455e-06,
+      "loss": 3.8361,
+      "step": 2494
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.221818181818182e-06,
+      "loss": 3.655,
+      "step": 2495
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.2210909090909094e-06,
+      "loss": 3.6673,
+      "step": 2496
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.2203636363636367e-06,
+      "loss": 3.6259,
+      "step": 2497
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.2196363636363635e-06,
+      "loss": 3.9185,
+      "step": 2498
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.2189090909090916e-06,
+      "loss": 3.6254,
+      "step": 2499
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.2181818181818184e-06,
+      "loss": 3.548,
+      "step": 2500
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.2174545454545457e-06,
+      "loss": 3.5795,
+      "step": 2501
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.216727272727273e-06,
+      "loss": 3.9024,
+      "step": 2502
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.216e-06,
+      "loss": 3.6569,
+      "step": 2503
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.2152727272727275e-06,
+      "loss": 3.8268,
+      "step": 2504
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.2145454545454547e-06,
+      "loss": 3.5793,
+      "step": 2505
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.213818181818182e-06,
+      "loss": 3.565,
+      "step": 2506
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.2130909090909097e-06,
+      "loss": 3.8462,
+      "step": 2507
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.212363636363637e-06,
+      "loss": 3.5428,
+      "step": 2508
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.211636363636364e-06,
+      "loss": 3.8577,
+      "step": 2509
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.210909090909091e-06,
+      "loss": 3.8504,
+      "step": 2510
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.2101818181818183e-06,
+      "loss": 3.5735,
+      "step": 2511
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.2094545454545455e-06,
+      "loss": 3.5634,
+      "step": 2512
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.2087272727272728e-06,
+      "loss": 4.0019,
+      "step": 2513
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.208e-06,
+      "loss": 3.697,
+      "step": 2514
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.2072727272727277e-06,
+      "loss": 3.6075,
+      "step": 2515
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.206545454545455e-06,
+      "loss": 3.718,
+      "step": 2516
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.2058181818181822e-06,
+      "loss": 3.6526,
+      "step": 2517
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.2050909090909095e-06,
+      "loss": 3.5746,
+      "step": 2518
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.2043636363636367e-06,
+      "loss": 3.6535,
+      "step": 2519
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.2036363636363636e-06,
+      "loss": 3.9458,
+      "step": 2520
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.202909090909091e-06,
+      "loss": 3.9518,
+      "step": 2521
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.202181818181818e-06,
+      "loss": 3.7085,
+      "step": 2522
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.2014545454545458e-06,
+      "loss": 3.6497,
+      "step": 2523
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.200727272727273e-06,
+      "loss": 3.5035,
+      "step": 2524
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 3.6346,
+      "step": 2525
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.1992727272727275e-06,
+      "loss": 3.884,
+      "step": 2526
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.198545454545455e-06,
+      "loss": 3.6322,
+      "step": 2527
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.197818181818182e-06,
+      "loss": 3.7028,
+      "step": 2528
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.1970909090909093e-06,
+      "loss": 3.5909,
+      "step": 2529
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.196363636363637e-06,
+      "loss": 3.7813,
+      "step": 2530
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.1956363636363642e-06,
+      "loss": 3.7084,
+      "step": 2531
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.1949090909090915e-06,
+      "loss": 3.8073,
+      "step": 2532
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.1941818181818183e-06,
+      "loss": 3.6439,
+      "step": 2533
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.1934545454545456e-06,
+      "loss": 3.5961,
+      "step": 2534
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.192727272727273e-06,
+      "loss": 3.7325,
+      "step": 2535
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.192e-06,
+      "loss": 3.5916,
+      "step": 2536
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.1912727272727274e-06,
+      "loss": 3.8376,
+      "step": 2537
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.190545454545455e-06,
+      "loss": 3.6749,
+      "step": 2538
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.1898181818181823e-06,
+      "loss": 3.5947,
+      "step": 2539
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.1890909090909096e-06,
+      "loss": 3.7303,
+      "step": 2540
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.188363636363637e-06,
+      "loss": 3.7313,
+      "step": 2541
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.187636363636364e-06,
+      "loss": 3.5809,
+      "step": 2542
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.186909090909091e-06,
+      "loss": 3.4833,
+      "step": 2543
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.186181818181818e-06,
+      "loss": 3.5436,
+      "step": 2544
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.1854545454545454e-06,
+      "loss": 3.6393,
+      "step": 2545
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.184727272727273e-06,
+      "loss": 3.6575,
+      "step": 2546
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.1840000000000003e-06,
+      "loss": 3.7289,
+      "step": 2547
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.1832727272727276e-06,
+      "loss": 3.6458,
+      "step": 2548
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.182545454545455e-06,
+      "loss": 3.8517,
+      "step": 2549
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.181818181818182e-06,
+      "loss": 3.5781,
+      "step": 2550
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.1810909090909094e-06,
+      "loss": 3.6303,
+      "step": 2551
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.1803636363636366e-06,
+      "loss": 3.6441,
+      "step": 2552
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.1796363636363635e-06,
+      "loss": 3.7021,
+      "step": 2553
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.1789090909090916e-06,
+      "loss": 3.9893,
+      "step": 2554
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.1781818181818184e-06,
+      "loss": 3.6471,
+      "step": 2555
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.1774545454545457e-06,
+      "loss": 3.6318,
+      "step": 2556
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.176727272727273e-06,
+      "loss": 3.5877,
+      "step": 2557
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.176e-06,
+      "loss": 3.7318,
+      "step": 2558
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.1752727272727274e-06,
+      "loss": 3.8145,
+      "step": 2559
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.1745454545454547e-06,
+      "loss": 3.6005,
+      "step": 2560
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.1738181818181824e-06,
+      "loss": 3.6603,
+      "step": 2561
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.1730909090909096e-06,
+      "loss": 3.7411,
+      "step": 2562
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.172363636363637e-06,
+      "loss": 3.6002,
+      "step": 2563
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.171636363636364e-06,
+      "loss": 3.6008,
+      "step": 2564
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.170909090909091e-06,
+      "loss": 3.588,
+      "step": 2565
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.1701818181818182e-06,
+      "loss": 3.6572,
+      "step": 2566
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.1694545454545455e-06,
+      "loss": 3.6535,
+      "step": 2567
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.1687272727272727e-06,
+      "loss": 3.7257,
+      "step": 2568
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.1680000000000004e-06,
+      "loss": 3.8447,
+      "step": 2569
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.1672727272727277e-06,
+      "loss": 3.6316,
+      "step": 2570
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.166545454545455e-06,
+      "loss": 3.5959,
+      "step": 2571
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.165818181818182e-06,
+      "loss": 3.6481,
+      "step": 2572
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.1650909090909094e-06,
+      "loss": 3.6288,
+      "step": 2573
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.1643636363636367e-06,
+      "loss": 3.44,
+      "step": 2574
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.1636363636363635e-06,
+      "loss": 3.6025,
+      "step": 2575
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.162909090909091e-06,
+      "loss": 3.5896,
+      "step": 2576
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.162181818181819e-06,
+      "loss": 3.7705,
+      "step": 2577
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.1614545454545457e-06,
+      "loss": 3.6616,
+      "step": 2578
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.160727272727273e-06,
+      "loss": 3.6426,
+      "step": 2579
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 3.6856,
+      "step": 2580
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.1592727272727275e-06,
+      "loss": 3.7386,
+      "step": 2581
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.1585454545454548e-06,
+      "loss": 3.8261,
+      "step": 2582
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.157818181818182e-06,
+      "loss": 3.7146,
+      "step": 2583
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.1570909090909093e-06,
+      "loss": 3.4161,
+      "step": 2584
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.156363636363637e-06,
+      "loss": 3.7045,
+      "step": 2585
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.155636363636364e-06,
+      "loss": 3.6778,
+      "step": 2586
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.1549090909090915e-06,
+      "loss": 3.7624,
+      "step": 2587
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.1541818181818183e-06,
+      "loss": 3.7621,
+      "step": 2588
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.1534545454545455e-06,
+      "loss": 3.8137,
+      "step": 2589
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.152727272727273e-06,
+      "loss": 3.6443,
+      "step": 2590
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.152e-06,
+      "loss": 3.6811,
+      "step": 2591
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.1512727272727273e-06,
+      "loss": 3.626,
+      "step": 2592
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.150545454545455e-06,
+      "loss": 3.7635,
+      "step": 2593
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.1498181818181823e-06,
+      "loss": 3.5529,
+      "step": 2594
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.1490909090909095e-06,
+      "loss": 3.6582,
+      "step": 2595
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.1483636363636368e-06,
+      "loss": 3.5921,
+      "step": 2596
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.147636363636364e-06,
+      "loss": 3.5136,
+      "step": 2597
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.146909090909091e-06,
+      "loss": 3.4965,
+      "step": 2598
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.146181818181818e-06,
+      "loss": 3.4837,
+      "step": 2599
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.145454545454546e-06,
+      "loss": 3.7368,
+      "step": 2600
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.144727272727273e-06,
+      "loss": 3.686,
+      "step": 2601
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.1440000000000003e-06,
+      "loss": 3.7054,
+      "step": 2602
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.1432727272727276e-06,
+      "loss": 3.7557,
+      "step": 2603
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.142545454545455e-06,
+      "loss": 3.5878,
+      "step": 2604
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.141818181818182e-06,
+      "loss": 3.5707,
+      "step": 2605
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.1410909090909093e-06,
+      "loss": 3.4679,
+      "step": 2606
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.1403636363636366e-06,
+      "loss": 3.7096,
+      "step": 2607
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.1396363636363643e-06,
+      "loss": 3.5699,
+      "step": 2608
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.1389090909090915e-06,
+      "loss": 3.8521,
+      "step": 2609
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.1381818181818184e-06,
+      "loss": 3.6045,
+      "step": 2610
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.1374545454545456e-06,
+      "loss": 3.4734,
+      "step": 2611
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.136727272727273e-06,
+      "loss": 3.7425,
+      "step": 2612
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.136e-06,
+      "loss": 3.4445,
+      "step": 2613
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.1352727272727274e-06,
+      "loss": 3.8485,
+      "step": 2614
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.1345454545454546e-06,
+      "loss": 3.5716,
+      "step": 2615
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.1338181818181823e-06,
+      "loss": 3.7152,
+      "step": 2616
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.1330909090909096e-06,
+      "loss": 3.6231,
+      "step": 2617
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.132363636363637e-06,
+      "loss": 3.7554,
+      "step": 2618
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.131636363636364e-06,
+      "loss": 3.662,
+      "step": 2619
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.130909090909091e-06,
+      "loss": 3.5258,
+      "step": 2620
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.130181818181818e-06,
+      "loss": 3.7855,
+      "step": 2621
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.1294545454545454e-06,
+      "loss": 3.7534,
+      "step": 2622
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.1287272727272727e-06,
+      "loss": 3.7445,
+      "step": 2623
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.1280000000000004e-06,
+      "loss": 3.537,
+      "step": 2624
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.1272727272727276e-06,
+      "loss": 3.5535,
+      "step": 2625
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.126545454545455e-06,
+      "loss": 3.5043,
+      "step": 2626
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.125818181818182e-06,
+      "loss": 3.6264,
+      "step": 2627
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.1250909090909094e-06,
+      "loss": 3.5474,
+      "step": 2628
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.1243636363636367e-06,
+      "loss": 3.9054,
+      "step": 2629
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.1236363636363635e-06,
+      "loss": 3.7626,
+      "step": 2630
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.1229090909090907e-06,
+      "loss": 3.5938,
+      "step": 2631
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.122181818181819e-06,
+      "loss": 3.6608,
+      "step": 2632
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.1214545454545457e-06,
+      "loss": 3.6942,
+      "step": 2633
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.120727272727273e-06,
+      "loss": 3.6488,
+      "step": 2634
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.12e-06,
+      "loss": 3.8279,
+      "step": 2635
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.1192727272727275e-06,
+      "loss": 3.5721,
+      "step": 2636
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.1185454545454547e-06,
+      "loss": 3.6578,
+      "step": 2637
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.117818181818182e-06,
+      "loss": 3.6099,
+      "step": 2638
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.1170909090909096e-06,
+      "loss": 3.5489,
+      "step": 2639
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.116363636363637e-06,
+      "loss": 3.6372,
+      "step": 2640
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.115636363636364e-06,
+      "loss": 3.6135,
+      "step": 2641
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.1149090909090914e-06,
+      "loss": 3.6491,
+      "step": 2642
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.1141818181818183e-06,
+      "loss": 3.7698,
+      "step": 2643
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.1134545454545455e-06,
+      "loss": 3.7581,
+      "step": 2644
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.1127272727272728e-06,
+      "loss": 3.6213,
+      "step": 2645
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.112e-06,
+      "loss": 3.573,
+      "step": 2646
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.1112727272727277e-06,
+      "loss": 3.7392,
+      "step": 2647
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.110545454545455e-06,
+      "loss": 3.8074,
+      "step": 2648
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.1098181818181822e-06,
+      "loss": 3.5709,
+      "step": 2649
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.1090909090909095e-06,
+      "loss": 3.8434,
+      "step": 2650
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.1083636363636367e-06,
+      "loss": 3.6127,
+      "step": 2651
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.107636363636364e-06,
+      "loss": 3.6872,
+      "step": 2652
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.106909090909091e-06,
+      "loss": 3.6449,
+      "step": 2653
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.106181818181818e-06,
+      "loss": 3.7065,
+      "step": 2654
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.1054545454545458e-06,
+      "loss": 3.7175,
+      "step": 2655
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.104727272727273e-06,
+      "loss": 3.4666,
+      "step": 2656
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.1040000000000003e-06,
+      "loss": 3.7215,
+      "step": 2657
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.1032727272727275e-06,
+      "loss": 3.4707,
+      "step": 2658
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.1025454545454548e-06,
+      "loss": 3.7347,
+      "step": 2659
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.101818181818182e-06,
+      "loss": 3.5699,
+      "step": 2660
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.1010909090909093e-06,
+      "loss": 3.6217,
+      "step": 2661
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.1003636363636365e-06,
+      "loss": 3.7221,
+      "step": 2662
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.0996363636363642e-06,
+      "loss": 3.722,
+      "step": 2663
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.0989090909090915e-06,
+      "loss": 3.6013,
+      "step": 2664
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.0981818181818183e-06,
+      "loss": 3.7058,
+      "step": 2665
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.0974545454545456e-06,
+      "loss": 3.6093,
+      "step": 2666
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.096727272727273e-06,
+      "loss": 3.5509,
+      "step": 2667
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.096e-06,
+      "loss": 3.4423,
+      "step": 2668
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.0952727272727273e-06,
+      "loss": 3.915,
+      "step": 2669
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.094545454545455e-06,
+      "loss": 3.6174,
+      "step": 2670
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.0938181818181823e-06,
+      "loss": 3.6322,
+      "step": 2671
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.0930909090909095e-06,
+      "loss": 3.5862,
+      "step": 2672
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.092363636363637e-06,
+      "loss": 3.7014,
+      "step": 2673
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.091636363636364e-06,
+      "loss": 3.453,
+      "step": 2674
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.090909090909091e-06,
+      "loss": 3.6614,
+      "step": 2675
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.090181818181818e-06,
+      "loss": 3.5716,
+      "step": 2676
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.0894545454545454e-06,
+      "loss": 3.8152,
+      "step": 2677
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.088727272727273e-06,
+      "loss": 3.7139,
+      "step": 2678
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.0880000000000003e-06,
+      "loss": 3.6787,
+      "step": 2679
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.0872727272727276e-06,
+      "loss": 3.617,
+      "step": 2680
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.086545454545455e-06,
+      "loss": 3.5668,
+      "step": 2681
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.085818181818182e-06,
+      "loss": 3.4854,
+      "step": 2682
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.0850909090909094e-06,
+      "loss": 3.5931,
+      "step": 2683
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.0843636363636366e-06,
+      "loss": 3.6258,
+      "step": 2684
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.0836363636363635e-06,
+      "loss": 3.7608,
+      "step": 2685
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.0829090909090916e-06,
+      "loss": 3.461,
+      "step": 2686
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.082181818181819e-06,
+      "loss": 3.7392,
+      "step": 2687
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.0814545454545456e-06,
+      "loss": 3.479,
+      "step": 2688
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.080727272727273e-06,
+      "loss": 3.5312,
+      "step": 2689
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.08e-06,
+      "loss": 3.5225,
+      "step": 2690
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.0792727272727274e-06,
+      "loss": 3.4722,
+      "step": 2691
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.0785454545454547e-06,
+      "loss": 3.6285,
+      "step": 2692
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.077818181818182e-06,
+      "loss": 3.6293,
+      "step": 2693
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.0770909090909096e-06,
+      "loss": 3.6874,
+      "step": 2694
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.076363636363637e-06,
+      "loss": 3.5556,
+      "step": 2695
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.075636363636364e-06,
+      "loss": 3.694,
+      "step": 2696
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.0749090909090914e-06,
+      "loss": 3.6492,
+      "step": 2697
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.074181818181818e-06,
+      "loss": 3.6479,
+      "step": 2698
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.0734545454545455e-06,
+      "loss": 3.6227,
+      "step": 2699
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.0727272727272727e-06,
+      "loss": 3.5269,
+      "step": 2700
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.072e-06,
+      "loss": 3.6495,
+      "step": 2701
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.0712727272727277e-06,
+      "loss": 3.6614,
+      "step": 2702
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.070545454545455e-06,
+      "loss": 3.661,
+      "step": 2703
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.069818181818182e-06,
+      "loss": 3.7891,
+      "step": 2704
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.0690909090909094e-06,
+      "loss": 3.7041,
+      "step": 2705
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.0683636363636367e-06,
+      "loss": 3.7739,
+      "step": 2706
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.067636363636364e-06,
+      "loss": 3.4874,
+      "step": 2707
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.0669090909090908e-06,
+      "loss": 3.5734,
+      "step": 2708
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.066181818181819e-06,
+      "loss": 3.7312,
+      "step": 2709
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.0654545454545457e-06,
+      "loss": 3.6793,
+      "step": 2710
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.064727272727273e-06,
+      "loss": 3.7123,
+      "step": 2711
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.0640000000000002e-06,
+      "loss": 3.7493,
+      "step": 2712
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.0632727272727275e-06,
+      "loss": 3.6115,
+      "step": 2713
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.0625454545454547e-06,
+      "loss": 3.6334,
+      "step": 2714
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.061818181818182e-06,
+      "loss": 3.5625,
+      "step": 2715
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.0610909090909093e-06,
+      "loss": 3.4637,
+      "step": 2716
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.060363636363637e-06,
+      "loss": 3.7673,
+      "step": 2717
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.059636363636364e-06,
+      "loss": 3.6413,
+      "step": 2718
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.0589090909090914e-06,
+      "loss": 3.8051,
+      "step": 2719
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.0581818181818183e-06,
+      "loss": 3.5408,
+      "step": 2720
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.0574545454545455e-06,
+      "loss": 3.4808,
+      "step": 2721
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.056727272727273e-06,
+      "loss": 3.7353,
+      "step": 2722
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.056e-06,
+      "loss": 3.5382,
+      "step": 2723
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.0552727272727273e-06,
+      "loss": 3.694,
+      "step": 2724
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.054545454545455e-06,
+      "loss": 3.5885,
+      "step": 2725
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.0538181818181822e-06,
+      "loss": 3.8218,
+      "step": 2726
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.0530909090909095e-06,
+      "loss": 3.5352,
+      "step": 2727
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.0523636363636368e-06,
+      "loss": 3.7688,
+      "step": 2728
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.051636363636364e-06,
+      "loss": 3.6255,
+      "step": 2729
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.050909090909091e-06,
+      "loss": 3.6582,
+      "step": 2730
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.050181818181818e-06,
+      "loss": 3.5731,
+      "step": 2731
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.0494545454545454e-06,
+      "loss": 3.7102,
+      "step": 2732
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.048727272727273e-06,
+      "loss": 3.5607,
+      "step": 2733
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.0480000000000003e-06,
+      "loss": 3.8444,
+      "step": 2734
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.0472727272727276e-06,
+      "loss": 3.5499,
+      "step": 2735
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.046545454545455e-06,
+      "loss": 3.4993,
+      "step": 2736
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.045818181818182e-06,
+      "loss": 3.7853,
+      "step": 2737
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.0450909090909093e-06,
+      "loss": 3.5163,
+      "step": 2738
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.0443636363636366e-06,
+      "loss": 3.707,
+      "step": 2739
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.0436363636363634e-06,
+      "loss": 3.4832,
+      "step": 2740
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.0429090909090915e-06,
+      "loss": 3.6901,
+      "step": 2741
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.0421818181818188e-06,
+      "loss": 3.5268,
+      "step": 2742
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.0414545454545456e-06,
+      "loss": 3.6065,
+      "step": 2743
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.040727272727273e-06,
+      "loss": 3.7915,
+      "step": 2744
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.04e-06,
+      "loss": 3.5731,
+      "step": 2745
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.0392727272727274e-06,
+      "loss": 3.5826,
+      "step": 2746
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.0385454545454546e-06,
+      "loss": 3.8505,
+      "step": 2747
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.0378181818181823e-06,
+      "loss": 3.8481,
+      "step": 2748
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.0370909090909096e-06,
+      "loss": 3.6693,
+      "step": 2749
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.036363636363637e-06,
+      "loss": 3.4608,
+      "step": 2750
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.035636363636364e-06,
+      "loss": 3.6979,
+      "step": 2751
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.0349090909090913e-06,
+      "loss": 3.4816,
+      "step": 2752
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.034181818181818e-06,
+      "loss": 3.7466,
+      "step": 2753
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.0334545454545454e-06,
+      "loss": 3.6711,
+      "step": 2754
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.0327272727272727e-06,
+      "loss": 3.8554,
+      "step": 2755
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.0320000000000004e-06,
+      "loss": 3.7154,
+      "step": 2756
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.0312727272727276e-06,
+      "loss": 3.5555,
+      "step": 2757
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.030545454545455e-06,
+      "loss": 3.68,
+      "step": 2758
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.029818181818182e-06,
+      "loss": 3.7914,
+      "step": 2759
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.0290909090909094e-06,
+      "loss": 3.6606,
+      "step": 2760
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.0283636363636366e-06,
+      "loss": 3.5525,
+      "step": 2761
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.027636363636364e-06,
+      "loss": 3.5781,
+      "step": 2762
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.0269090909090907e-06,
+      "loss": 3.6453,
+      "step": 2763
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.026181818181819e-06,
+      "loss": 3.7414,
+      "step": 2764
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.0254545454545457e-06,
+      "loss": 3.846,
+      "step": 2765
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.024727272727273e-06,
+      "loss": 3.6809,
+      "step": 2766
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.024e-06,
+      "loss": 3.5083,
+      "step": 2767
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.0232727272727274e-06,
+      "loss": 3.6094,
+      "step": 2768
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.0225454545454547e-06,
+      "loss": 3.5255,
+      "step": 2769
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.021818181818182e-06,
+      "loss": 3.575,
+      "step": 2770
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.11081081081081082,
+      "eval_loss": 3.627187728881836,
+      "eval_runtime": 143.5726,
+      "eval_samples_per_second": 2.577,
+      "eval_steps_per_second": 0.648,
+      "step": 2770
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.021090909090909e-06,
+      "loss": 3.5671,
+      "step": 2771
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.020363636363637e-06,
+      "loss": 3.6823,
+      "step": 2772
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.019636363636364e-06,
+      "loss": 3.5836,
+      "step": 2773
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.0189090909090914e-06,
+      "loss": 3.5118,
+      "step": 2774
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.0181818181818182e-06,
+      "loss": 3.679,
+      "step": 2775
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.0174545454545455e-06,
+      "loss": 3.5263,
+      "step": 2776
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.0167272727272728e-06,
+      "loss": 3.7367,
+      "step": 2777
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.016e-06,
+      "loss": 3.5658,
+      "step": 2778
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.0152727272727277e-06,
+      "loss": 3.5282,
+      "step": 2779
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.014545454545455e-06,
+      "loss": 3.7534,
+      "step": 2780
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.013818181818182e-06,
+      "loss": 3.6797,
+      "step": 2781
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.0130909090909095e-06,
+      "loss": 3.4615,
+      "step": 2782
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.0123636363636367e-06,
+      "loss": 3.7037,
+      "step": 2783
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.011636363636364e-06,
+      "loss": 3.6396,
+      "step": 2784
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.010909090909091e-06,
+      "loss": 3.5341,
+      "step": 2785
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.010181818181818e-06,
+      "loss": 3.5647,
+      "step": 2786
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.009454545454546e-06,
+      "loss": 3.5095,
+      "step": 2787
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.008727272727273e-06,
+      "loss": 3.6153,
+      "step": 2788
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.0080000000000003e-06,
+      "loss": 3.7015,
+      "step": 2789
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.0072727272727275e-06,
+      "loss": 3.6733,
+      "step": 2790
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.0065454545454548e-06,
+      "loss": 3.6276,
+      "step": 2791
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.005818181818182e-06,
+      "loss": 3.6867,
+      "step": 2792
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.0050909090909093e-06,
+      "loss": 3.6831,
+      "step": 2793
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.0043636363636365e-06,
+      "loss": 3.663,
+      "step": 2794
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.0036363636363642e-06,
+      "loss": 3.7813,
+      "step": 2795
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.0029090909090915e-06,
+      "loss": 3.5144,
+      "step": 2796
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.0021818181818187e-06,
+      "loss": 3.5641,
+      "step": 2797
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.0014545454545456e-06,
+      "loss": 3.5372,
+      "step": 2798
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.000727272727273e-06,
+      "loss": 3.5807,
+      "step": 2799
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3e-06,
+      "loss": 3.7813,
+      "step": 2800
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.9992727272727273e-06,
+      "loss": 3.7763,
+      "step": 2801
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.9985454545454546e-06,
+      "loss": 3.5892,
+      "step": 2802
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.9978181818181823e-06,
+      "loss": 3.5141,
+      "step": 2803
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.9970909090909095e-06,
+      "loss": 3.5462,
+      "step": 2804
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.9963636363636368e-06,
+      "loss": 3.4598,
+      "step": 2805
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.995636363636364e-06,
+      "loss": 3.6687,
+      "step": 2806
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.9949090909090913e-06,
+      "loss": 3.5503,
+      "step": 2807
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.994181818181818e-06,
+      "loss": 3.4521,
+      "step": 2808
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.9934545454545454e-06,
+      "loss": 3.4781,
+      "step": 2809
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.9927272727272726e-06,
+      "loss": 3.5826,
+      "step": 2810
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.9920000000000003e-06,
+      "loss": 3.8054,
+      "step": 2811
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.9912727272727276e-06,
+      "loss": 3.5207,
+      "step": 2812
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.990545454545455e-06,
+      "loss": 3.594,
+      "step": 2813
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.989818181818182e-06,
+      "loss": 3.5837,
+      "step": 2814
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.9890909090909093e-06,
+      "loss": 3.5153,
+      "step": 2815
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.9883636363636366e-06,
+      "loss": 3.59,
+      "step": 2816
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.987636363636364e-06,
+      "loss": 3.5121,
+      "step": 2817
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.9869090909090915e-06,
+      "loss": 3.5643,
+      "step": 2818
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.986181818181819e-06,
+      "loss": 3.7369,
+      "step": 2819
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.9854545454545456e-06,
+      "loss": 3.5248,
+      "step": 2820
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.984727272727273e-06,
+      "loss": 3.7174,
+      "step": 2821
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.984e-06,
+      "loss": 3.6093,
+      "step": 2822
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.9832727272727274e-06,
+      "loss": 3.4533,
+      "step": 2823
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.9825454545454547e-06,
+      "loss": 3.6357,
+      "step": 2824
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.981818181818182e-06,
+      "loss": 3.5456,
+      "step": 2825
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.9810909090909096e-06,
+      "loss": 3.7098,
+      "step": 2826
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.980363636363637e-06,
+      "loss": 3.6211,
+      "step": 2827
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.979636363636364e-06,
+      "loss": 3.6544,
+      "step": 2828
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.9789090909090914e-06,
+      "loss": 3.5249,
+      "step": 2829
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.978181818181818e-06,
+      "loss": 3.7207,
+      "step": 2830
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.9774545454545455e-06,
+      "loss": 3.5891,
+      "step": 2831
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.9767272727272727e-06,
+      "loss": 3.7317,
+      "step": 2832
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.976e-06,
+      "loss": 3.5632,
+      "step": 2833
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.9752727272727276e-06,
+      "loss": 3.6085,
+      "step": 2834
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.974545454545455e-06,
+      "loss": 4.1295,
+      "step": 2835
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.973818181818182e-06,
+      "loss": 3.5575,
+      "step": 2836
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.9730909090909094e-06,
+      "loss": 3.6491,
+      "step": 2837
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.9723636363636367e-06,
+      "loss": 3.5747,
+      "step": 2838
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.971636363636364e-06,
+      "loss": 3.62,
+      "step": 2839
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.9709090909090908e-06,
+      "loss": 3.5425,
+      "step": 2840
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.970181818181818e-06,
+      "loss": 3.6735,
+      "step": 2841
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.969454545454546e-06,
+      "loss": 3.7152,
+      "step": 2842
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.968727272727273e-06,
+      "loss": 3.6842,
+      "step": 2843
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.9680000000000002e-06,
+      "loss": 3.6827,
+      "step": 2844
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.9672727272727275e-06,
+      "loss": 3.4381,
+      "step": 2845
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.9665454545454547e-06,
+      "loss": 3.9054,
+      "step": 2846
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.965818181818182e-06,
+      "loss": 3.5419,
+      "step": 2847
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.9650909090909092e-06,
+      "loss": 3.7246,
+      "step": 2848
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.9643636363636365e-06,
+      "loss": 3.5557,
+      "step": 2849
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.963636363636364e-06,
+      "loss": 3.6013,
+      "step": 2850
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.9629090909090914e-06,
+      "loss": 3.5601,
+      "step": 2851
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.9621818181818187e-06,
+      "loss": 3.8548,
+      "step": 2852
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.9614545454545455e-06,
+      "loss": 3.5644,
+      "step": 2853
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.9607272727272728e-06,
+      "loss": 3.4227,
+      "step": 2854
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.96e-06,
+      "loss": 3.5464,
+      "step": 2855
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.9592727272727273e-06,
+      "loss": 3.7327,
+      "step": 2856
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.958545454545455e-06,
+      "loss": 3.6124,
+      "step": 2857
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.9578181818181822e-06,
+      "loss": 3.5764,
+      "step": 2858
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.9570909090909095e-06,
+      "loss": 3.7866,
+      "step": 2859
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.9563636363636367e-06,
+      "loss": 3.7732,
+      "step": 2860
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.955636363636364e-06,
+      "loss": 3.7085,
+      "step": 2861
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.9549090909090913e-06,
+      "loss": 3.4562,
+      "step": 2862
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.954181818181818e-06,
+      "loss": 3.7176,
+      "step": 2863
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.9534545454545453e-06,
+      "loss": 3.7704,
+      "step": 2864
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.952727272727273e-06,
+      "loss": 3.644,
+      "step": 2865
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.9520000000000003e-06,
+      "loss": 3.4577,
+      "step": 2866
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.9512727272727275e-06,
+      "loss": 3.8254,
+      "step": 2867
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.950545454545455e-06,
+      "loss": 3.6691,
+      "step": 2868
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.949818181818182e-06,
+      "loss": 3.6292,
+      "step": 2869
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.9490909090909093e-06,
+      "loss": 3.6568,
+      "step": 2870
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.9483636363636366e-06,
+      "loss": 3.6642,
+      "step": 2871
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.947636363636364e-06,
+      "loss": 3.8021,
+      "step": 2872
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.9469090909090915e-06,
+      "loss": 3.6875,
+      "step": 2873
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.9461818181818188e-06,
+      "loss": 3.3492,
+      "step": 2874
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.9454545454545456e-06,
+      "loss": 3.5706,
+      "step": 2875
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.944727272727273e-06,
+      "loss": 3.6476,
+      "step": 2876
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.944e-06,
+      "loss": 3.5811,
+      "step": 2877
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.9432727272727274e-06,
+      "loss": 3.4065,
+      "step": 2878
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.9425454545454546e-06,
+      "loss": 3.6823,
+      "step": 2879
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.941818181818182e-06,
+      "loss": 3.6613,
+      "step": 2880
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.9410909090909096e-06,
+      "loss": 3.817,
+      "step": 2881
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.940363636363637e-06,
+      "loss": 3.6042,
+      "step": 2882
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.939636363636364e-06,
+      "loss": 3.7866,
+      "step": 2883
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.9389090909090913e-06,
+      "loss": 3.5806,
+      "step": 2884
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.938181818181818e-06,
+      "loss": 3.8164,
+      "step": 2885
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.9374545454545454e-06,
+      "loss": 3.6613,
+      "step": 2886
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.9367272727272727e-06,
+      "loss": 3.6174,
+      "step": 2887
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.9360000000000003e-06,
+      "loss": 3.5323,
+      "step": 2888
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.9352727272727276e-06,
+      "loss": 3.6444,
+      "step": 2889
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.934545454545455e-06,
+      "loss": 3.6331,
+      "step": 2890
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.933818181818182e-06,
+      "loss": 3.5936,
+      "step": 2891
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.9330909090909094e-06,
+      "loss": 3.6554,
+      "step": 2892
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.9323636363636366e-06,
+      "loss": 3.8044,
+      "step": 2893
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.931636363636364e-06,
+      "loss": 3.606,
+      "step": 2894
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.9309090909090907e-06,
+      "loss": 3.4642,
+      "step": 2895
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.930181818181819e-06,
+      "loss": 3.6698,
+      "step": 2896
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.929454545454546e-06,
+      "loss": 3.5653,
+      "step": 2897
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.928727272727273e-06,
+      "loss": 3.6691,
+      "step": 2898
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.928e-06,
+      "loss": 3.6311,
+      "step": 2899
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.9272727272727274e-06,
+      "loss": 3.7976,
+      "step": 2900
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.9265454545454547e-06,
+      "loss": 3.8184,
+      "step": 2901
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.925818181818182e-06,
+      "loss": 3.5537,
+      "step": 2902
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.925090909090909e-06,
+      "loss": 3.5677,
+      "step": 2903
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.924363636363637e-06,
+      "loss": 3.6365,
+      "step": 2904
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.923636363636364e-06,
+      "loss": 3.671,
+      "step": 2905
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.9229090909090914e-06,
+      "loss": 3.7614,
+      "step": 2906
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.9221818181818186e-06,
+      "loss": 3.6804,
+      "step": 2907
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.9214545454545455e-06,
+      "loss": 3.7199,
+      "step": 2908
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.9207272727272727e-06,
+      "loss": 3.8525,
+      "step": 2909
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.92e-06,
+      "loss": 3.5475,
+      "step": 2910
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.9192727272727273e-06,
+      "loss": 3.8354,
+      "step": 2911
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.918545454545455e-06,
+      "loss": 3.6474,
+      "step": 2912
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.917818181818182e-06,
+      "loss": 3.8061,
+      "step": 2913
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.9170909090909094e-06,
+      "loss": 3.5536,
+      "step": 2914
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.9163636363636367e-06,
+      "loss": 3.6972,
+      "step": 2915
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.915636363636364e-06,
+      "loss": 3.6434,
+      "step": 2916
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.9149090909090912e-06,
+      "loss": 3.5194,
+      "step": 2917
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.914181818181818e-06,
+      "loss": 3.581,
+      "step": 2918
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.9134545454545453e-06,
+      "loss": 3.5242,
+      "step": 2919
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.912727272727273e-06,
+      "loss": 3.8688,
+      "step": 2920
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.9120000000000002e-06,
+      "loss": 3.4269,
+      "step": 2921
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.9112727272727275e-06,
+      "loss": 3.8046,
+      "step": 2922
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.9105454545454548e-06,
+      "loss": 3.7053,
+      "step": 2923
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.909818181818182e-06,
+      "loss": 3.8934,
+      "step": 2924
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.9090909090909093e-06,
+      "loss": 3.6716,
+      "step": 2925
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.9083636363636365e-06,
+      "loss": 3.8236,
+      "step": 2926
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.907636363636364e-06,
+      "loss": 3.4842,
+      "step": 2927
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.9069090909090915e-06,
+      "loss": 3.6489,
+      "step": 2928
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.9061818181818187e-06,
+      "loss": 3.7354,
+      "step": 2929
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.9054545454545456e-06,
+      "loss": 3.5995,
+      "step": 2930
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.904727272727273e-06,
+      "loss": 3.6887,
+      "step": 2931
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.904e-06,
+      "loss": 3.4516,
+      "step": 2932
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.9032727272727273e-06,
+      "loss": 3.7326,
+      "step": 2933
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.9025454545454546e-06,
+      "loss": 3.7652,
+      "step": 2934
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.9018181818181823e-06,
+      "loss": 3.6579,
+      "step": 2935
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.9010909090909095e-06,
+      "loss": 3.8758,
+      "step": 2936
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.9003636363636368e-06,
+      "loss": 3.5154,
+      "step": 2937
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.899636363636364e-06,
+      "loss": 3.7857,
+      "step": 2938
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.8989090909090913e-06,
+      "loss": 3.505,
+      "step": 2939
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.898181818181818e-06,
+      "loss": 3.4916,
+      "step": 2940
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.8974545454545454e-06,
+      "loss": 3.5267,
+      "step": 2941
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.8967272727272726e-06,
+      "loss": 3.7781,
+      "step": 2942
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.8960000000000003e-06,
+      "loss": 3.6072,
+      "step": 2943
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.8952727272727276e-06,
+      "loss": 3.6557,
+      "step": 2944
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.894545454545455e-06,
+      "loss": 3.5537,
+      "step": 2945
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.893818181818182e-06,
+      "loss": 3.5326,
+      "step": 2946
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.8930909090909093e-06,
+      "loss": 3.7247,
+      "step": 2947
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.8923636363636366e-06,
+      "loss": 3.4809,
+      "step": 2948
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.891636363636364e-06,
+      "loss": 3.7335,
+      "step": 2949
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.8909090909090907e-06,
+      "loss": 3.6208,
+      "step": 2950
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.8901818181818188e-06,
+      "loss": 3.4591,
+      "step": 2951
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.889454545454546e-06,
+      "loss": 3.6695,
+      "step": 2952
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.888727272727273e-06,
+      "loss": 3.6787,
+      "step": 2953
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.888e-06,
+      "loss": 3.6635,
+      "step": 2954
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.8872727272727274e-06,
+      "loss": 3.7023,
+      "step": 2955
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.8865454545454546e-06,
+      "loss": 3.7175,
+      "step": 2956
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.885818181818182e-06,
+      "loss": 3.7819,
+      "step": 2957
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.885090909090909e-06,
+      "loss": 3.6527,
+      "step": 2958
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.884363636363637e-06,
+      "loss": 3.8935,
+      "step": 2959
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.883636363636364e-06,
+      "loss": 3.7021,
+      "step": 2960
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.8829090909090914e-06,
+      "loss": 3.4554,
+      "step": 2961
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.8821818181818186e-06,
+      "loss": 3.4386,
+      "step": 2962
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.8814545454545454e-06,
+      "loss": 3.7753,
+      "step": 2963
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.8807272727272727e-06,
+      "loss": 3.4642,
+      "step": 2964
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.88e-06,
+      "loss": 3.6339,
+      "step": 2965
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.8792727272727276e-06,
+      "loss": 3.6908,
+      "step": 2966
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.878545454545455e-06,
+      "loss": 3.5435,
+      "step": 2967
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.877818181818182e-06,
+      "loss": 3.5656,
+      "step": 2968
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.8770909090909094e-06,
+      "loss": 3.6315,
+      "step": 2969
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.8763636363636367e-06,
+      "loss": 3.6271,
+      "step": 2970
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.875636363636364e-06,
+      "loss": 3.7758,
+      "step": 2971
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.874909090909091e-06,
+      "loss": 3.5434,
+      "step": 2972
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.874181818181818e-06,
+      "loss": 3.6143,
+      "step": 2973
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.873454545454546e-06,
+      "loss": 3.586,
+      "step": 2974
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.872727272727273e-06,
+      "loss": 3.6276,
+      "step": 2975
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.872e-06,
+      "loss": 3.6679,
+      "step": 2976
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.8712727272727275e-06,
+      "loss": 3.616,
+      "step": 2977
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.8705454545454547e-06,
+      "loss": 3.8301,
+      "step": 2978
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.869818181818182e-06,
+      "loss": 3.6409,
+      "step": 2979
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.8690909090909092e-06,
+      "loss": 3.7168,
+      "step": 2980
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.8683636363636365e-06,
+      "loss": 3.4642,
+      "step": 2981
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.867636363636364e-06,
+      "loss": 3.5919,
+      "step": 2982
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.8669090909090914e-06,
+      "loss": 3.48,
+      "step": 2983
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.8661818181818187e-06,
+      "loss": 3.6102,
+      "step": 2984
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.8654545454545455e-06,
+      "loss": 3.6852,
+      "step": 2985
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.8647272727272728e-06,
+      "loss": 3.4939,
+      "step": 2986
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.864e-06,
+      "loss": 3.7021,
+      "step": 2987
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.8632727272727273e-06,
+      "loss": 3.4459,
+      "step": 2988
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.8625454545454545e-06,
+      "loss": 3.8312,
+      "step": 2989
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.8618181818181822e-06,
+      "loss": 3.5913,
+      "step": 2990
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.8610909090909095e-06,
+      "loss": 3.7671,
+      "step": 2991
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.8603636363636367e-06,
+      "loss": 3.443,
+      "step": 2992
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.859636363636364e-06,
+      "loss": 3.5854,
+      "step": 2993
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.8589090909090912e-06,
+      "loss": 3.6113,
+      "step": 2994
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.858181818181818e-06,
+      "loss": 3.5913,
+      "step": 2995
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.8574545454545453e-06,
+      "loss": 3.7458,
+      "step": 2996
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.8567272727272734e-06,
+      "loss": 3.6856,
+      "step": 2997
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.8560000000000003e-06,
+      "loss": 3.6117,
+      "step": 2998
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.8552727272727275e-06,
+      "loss": 3.761,
+      "step": 2999
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.8545454545454548e-06,
+      "loss": 3.6662,
+      "step": 3000
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.853818181818182e-06,
+      "loss": 3.731,
+      "step": 3001
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.8530909090909093e-06,
+      "loss": 3.5375,
+      "step": 3002
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.8523636363636366e-06,
+      "loss": 3.6863,
+      "step": 3003
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.851636363636364e-06,
+      "loss": 3.6109,
+      "step": 3004
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.8509090909090915e-06,
+      "loss": 3.741,
+      "step": 3005
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.8501818181818187e-06,
+      "loss": 3.5373,
+      "step": 3006
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.849454545454546e-06,
+      "loss": 3.5011,
+      "step": 3007
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.848727272727273e-06,
+      "loss": 3.5595,
+      "step": 3008
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.848e-06,
+      "loss": 3.4339,
+      "step": 3009
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.8472727272727273e-06,
+      "loss": 3.3387,
+      "step": 3010
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.8465454545454546e-06,
+      "loss": 3.5196,
+      "step": 3011
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.845818181818182e-06,
+      "loss": 3.8656,
+      "step": 3012
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.8450909090909095e-06,
+      "loss": 3.6411,
+      "step": 3013
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.844363636363637e-06,
+      "loss": 3.4902,
+      "step": 3014
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.843636363636364e-06,
+      "loss": 3.5744,
+      "step": 3015
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.8429090909090913e-06,
+      "loss": 3.639,
+      "step": 3016
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.8421818181818186e-06,
+      "loss": 3.731,
+      "step": 3017
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.8414545454545454e-06,
+      "loss": 3.543,
+      "step": 3018
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.8407272727272727e-06,
+      "loss": 3.4504,
+      "step": 3019
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.84e-06,
+      "loss": 3.503,
+      "step": 3020
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.8392727272727276e-06,
+      "loss": 3.7372,
+      "step": 3021
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.838545454545455e-06,
+      "loss": 3.6896,
+      "step": 3022
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.837818181818182e-06,
+      "loss": 3.7846,
+      "step": 3023
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.8370909090909094e-06,
+      "loss": 3.8746,
+      "step": 3024
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.8363636363636366e-06,
+      "loss": 3.6765,
+      "step": 3025
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.835636363636364e-06,
+      "loss": 3.6334,
+      "step": 3026
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.834909090909091e-06,
+      "loss": 3.581,
+      "step": 3027
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.834181818181818e-06,
+      "loss": 3.5972,
+      "step": 3028
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.833454545454546e-06,
+      "loss": 3.5309,
+      "step": 3029
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.832727272727273e-06,
+      "loss": 3.9291,
+      "step": 3030
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.832e-06,
+      "loss": 3.3898,
+      "step": 3031
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.8312727272727274e-06,
+      "loss": 3.6128,
+      "step": 3032
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.8305454545454547e-06,
+      "loss": 3.8453,
+      "step": 3033
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.829818181818182e-06,
+      "loss": 3.7807,
+      "step": 3034
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.829090909090909e-06,
+      "loss": 3.5071,
+      "step": 3035
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.828363636363637e-06,
+      "loss": 3.6506,
+      "step": 3036
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.827636363636364e-06,
+      "loss": 3.7005,
+      "step": 3037
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.8269090909090914e-06,
+      "loss": 3.6353,
+      "step": 3038
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.8261818181818186e-06,
+      "loss": 3.5723,
+      "step": 3039
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.8254545454545455e-06,
+      "loss": 3.4626,
+      "step": 3040
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.8247272727272727e-06,
+      "loss": 3.4749,
+      "step": 3041
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.824e-06,
+      "loss": 3.5042,
+      "step": 3042
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.8232727272727272e-06,
+      "loss": 3.5694,
+      "step": 3043
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.822545454545455e-06,
+      "loss": 3.632,
+      "step": 3044
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.821818181818182e-06,
+      "loss": 3.6396,
+      "step": 3045
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.8210909090909094e-06,
+      "loss": 3.5375,
+      "step": 3046
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.8203636363636367e-06,
+      "loss": 3.4806,
+      "step": 3047
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.15945945945945947,
+      "eval_loss": 3.5812857151031494,
+      "eval_runtime": 144.2331,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 0.645,
+      "step": 3047
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.819636363636364e-06,
+      "loss": 3.6777,
+      "step": 3048
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.818909090909091e-06,
+      "loss": 3.8959,
+      "step": 3049
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.818181818181818e-06,
+      "loss": 3.8359,
+      "step": 3050
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.8174545454545453e-06,
+      "loss": 3.919,
+      "step": 3051
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.8167272727272734e-06,
+      "loss": 3.4125,
+      "step": 3052
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.8160000000000002e-06,
+      "loss": 3.8415,
+      "step": 3053
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.8152727272727275e-06,
+      "loss": 3.717,
+      "step": 3054
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.8145454545454547e-06,
+      "loss": 3.5924,
+      "step": 3055
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.813818181818182e-06,
+      "loss": 3.664,
+      "step": 3056
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.8130909090909093e-06,
+      "loss": 3.3707,
+      "step": 3057
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.8123636363636365e-06,
+      "loss": 3.5407,
+      "step": 3058
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.8116363636363638e-06,
+      "loss": 3.608,
+      "step": 3059
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.8109090909090914e-06,
+      "loss": 3.4917,
+      "step": 3060
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.8101818181818187e-06,
+      "loss": 3.6838,
+      "step": 3061
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.809454545454546e-06,
+      "loss": 3.7894,
+      "step": 3062
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.808727272727273e-06,
+      "loss": 3.507,
+      "step": 3063
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.808e-06,
+      "loss": 3.8222,
+      "step": 3064
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.8072727272727273e-06,
+      "loss": 3.7264,
+      "step": 3065
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.8065454545454546e-06,
+      "loss": 3.5522,
+      "step": 3066
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.805818181818182e-06,
+      "loss": 3.6481,
+      "step": 3067
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.8050909090909095e-06,
+      "loss": 3.6298,
+      "step": 3068
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.8043636363636368e-06,
+      "loss": 3.4279,
+      "step": 3069
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.803636363636364e-06,
+      "loss": 3.3204,
+      "step": 3070
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.8029090909090913e-06,
+      "loss": 3.4036,
+      "step": 3071
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.8021818181818185e-06,
+      "loss": 3.7541,
+      "step": 3072
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.8014545454545454e-06,
+      "loss": 3.5619,
+      "step": 3073
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.8007272727272726e-06,
+      "loss": 3.7626,
+      "step": 3074
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 3.7078,
+      "step": 3075
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.7992727272727276e-06,
+      "loss": 3.5341,
+      "step": 3076
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.798545454545455e-06,
+      "loss": 3.6674,
+      "step": 3077
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.797818181818182e-06,
+      "loss": 3.5595,
+      "step": 3078
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.7970909090909093e-06,
+      "loss": 3.6684,
+      "step": 3079
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.7963636363636366e-06,
+      "loss": 3.5851,
+      "step": 3080
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.795636363636364e-06,
+      "loss": 3.6151,
+      "step": 3081
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.794909090909091e-06,
+      "loss": 3.721,
+      "step": 3082
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.7941818181818188e-06,
+      "loss": 3.5298,
+      "step": 3083
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.793454545454546e-06,
+      "loss": 3.7016,
+      "step": 3084
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.792727272727273e-06,
+      "loss": 3.5055,
+      "step": 3085
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.792e-06,
+      "loss": 3.7416,
+      "step": 3086
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.7912727272727274e-06,
+      "loss": 3.43,
+      "step": 3087
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.7905454545454546e-06,
+      "loss": 3.6589,
+      "step": 3088
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.789818181818182e-06,
+      "loss": 3.7268,
+      "step": 3089
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.789090909090909e-06,
+      "loss": 3.4344,
+      "step": 3090
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.788363636363637e-06,
+      "loss": 3.6066,
+      "step": 3091
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.787636363636364e-06,
+      "loss": 3.395,
+      "step": 3092
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.7869090909090913e-06,
+      "loss": 3.6097,
+      "step": 3093
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.7861818181818186e-06,
+      "loss": 3.3842,
+      "step": 3094
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.7854545454545454e-06,
+      "loss": 3.4406,
+      "step": 3095
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.7847272727272727e-06,
+      "loss": 3.5245,
+      "step": 3096
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.784e-06,
+      "loss": 3.4254,
+      "step": 3097
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.783272727272727e-06,
+      "loss": 3.7958,
+      "step": 3098
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.782545454545455e-06,
+      "loss": 3.5651,
+      "step": 3099
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.781818181818182e-06,
+      "loss": 3.5724,
+      "step": 3100
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.7810909090909094e-06,
+      "loss": 3.8211,
+      "step": 3101
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.7803636363636366e-06,
+      "loss": 3.6586,
+      "step": 3102
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.779636363636364e-06,
+      "loss": 3.4822,
+      "step": 3103
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.778909090909091e-06,
+      "loss": 3.6889,
+      "step": 3104
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.778181818181818e-06,
+      "loss": 3.8092,
+      "step": 3105
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.777454545454546e-06,
+      "loss": 3.6179,
+      "step": 3106
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.7767272727272734e-06,
+      "loss": 3.3292,
+      "step": 3107
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.776e-06,
+      "loss": 3.5278,
+      "step": 3108
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.7752727272727274e-06,
+      "loss": 3.5687,
+      "step": 3109
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.7745454545454547e-06,
+      "loss": 3.5289,
+      "step": 3110
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.773818181818182e-06,
+      "loss": 3.5874,
+      "step": 3111
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.7730909090909092e-06,
+      "loss": 3.5105,
+      "step": 3112
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.7723636363636365e-06,
+      "loss": 3.5001,
+      "step": 3113
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.771636363636364e-06,
+      "loss": 3.684,
+      "step": 3114
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.7709090909090914e-06,
+      "loss": 3.9782,
+      "step": 3115
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.7701818181818187e-06,
+      "loss": 3.5252,
+      "step": 3116
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.769454545454546e-06,
+      "loss": 3.609,
+      "step": 3117
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.7687272727272728e-06,
+      "loss": 3.4944,
+      "step": 3118
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.768e-06,
+      "loss": 3.7415,
+      "step": 3119
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.7672727272727273e-06,
+      "loss": 3.5218,
+      "step": 3120
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.7665454545454545e-06,
+      "loss": 3.7511,
+      "step": 3121
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.765818181818182e-06,
+      "loss": 3.383,
+      "step": 3122
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.7650909090909095e-06,
+      "loss": 3.8094,
+      "step": 3123
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.7643636363636367e-06,
+      "loss": 3.5213,
+      "step": 3124
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.763636363636364e-06,
+      "loss": 3.5601,
+      "step": 3125
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.7629090909090912e-06,
+      "loss": 3.6954,
+      "step": 3126
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.7621818181818185e-06,
+      "loss": 3.5519,
+      "step": 3127
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.7614545454545453e-06,
+      "loss": 3.7325,
+      "step": 3128
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.7607272727272726e-06,
+      "loss": 3.6645,
+      "step": 3129
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 3.6362,
+      "step": 3130
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.7592727272727275e-06,
+      "loss": 3.5245,
+      "step": 3131
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.7585454545454548e-06,
+      "loss": 3.4731,
+      "step": 3132
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.757818181818182e-06,
+      "loss": 3.4874,
+      "step": 3133
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.7570909090909093e-06,
+      "loss": 3.534,
+      "step": 3134
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.7563636363636365e-06,
+      "loss": 3.672,
+      "step": 3135
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.755636363636364e-06,
+      "loss": 3.4693,
+      "step": 3136
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.754909090909091e-06,
+      "loss": 3.5953,
+      "step": 3137
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.7541818181818187e-06,
+      "loss": 3.7546,
+      "step": 3138
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.753454545454546e-06,
+      "loss": 3.7508,
+      "step": 3139
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.752727272727273e-06,
+      "loss": 3.5374,
+      "step": 3140
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.752e-06,
+      "loss": 3.6165,
+      "step": 3141
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.7512727272727273e-06,
+      "loss": 3.4757,
+      "step": 3142
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.7505454545454546e-06,
+      "loss": 3.6961,
+      "step": 3143
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.749818181818182e-06,
+      "loss": 3.6689,
+      "step": 3144
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.7490909090909095e-06,
+      "loss": 3.5731,
+      "step": 3145
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.7483636363636368e-06,
+      "loss": 3.3888,
+      "step": 3146
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.747636363636364e-06,
+      "loss": 3.6007,
+      "step": 3147
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.7469090909090913e-06,
+      "loss": 3.5967,
+      "step": 3148
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.7461818181818186e-06,
+      "loss": 3.5737,
+      "step": 3149
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.7454545454545454e-06,
+      "loss": 3.6586,
+      "step": 3150
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.7447272727272726e-06,
+      "loss": 3.4759,
+      "step": 3151
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.744e-06,
+      "loss": 3.4587,
+      "step": 3152
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.7432727272727276e-06,
+      "loss": 3.4808,
+      "step": 3153
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.742545454545455e-06,
+      "loss": 3.5344,
+      "step": 3154
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.741818181818182e-06,
+      "loss": 3.6218,
+      "step": 3155
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.7410909090909094e-06,
+      "loss": 3.6047,
+      "step": 3156
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.7403636363636366e-06,
+      "loss": 3.5618,
+      "step": 3157
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.739636363636364e-06,
+      "loss": 3.5918,
+      "step": 3158
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.738909090909091e-06,
+      "loss": 4.0049,
+      "step": 3159
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.738181818181818e-06,
+      "loss": 3.822,
+      "step": 3160
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.737454545454546e-06,
+      "loss": 3.6645,
+      "step": 3161
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.7367272727272733e-06,
+      "loss": 3.7666,
+      "step": 3162
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.736e-06,
+      "loss": 3.5503,
+      "step": 3163
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.7352727272727274e-06,
+      "loss": 3.624,
+      "step": 3164
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.7345454545454547e-06,
+      "loss": 3.5401,
+      "step": 3165
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.733818181818182e-06,
+      "loss": 3.7115,
+      "step": 3166
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.733090909090909e-06,
+      "loss": 3.5259,
+      "step": 3167
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.7323636363636364e-06,
+      "loss": 3.7466,
+      "step": 3168
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.731636363636364e-06,
+      "loss": 3.6575,
+      "step": 3169
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.7309090909090914e-06,
+      "loss": 3.6179,
+      "step": 3170
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.7301818181818186e-06,
+      "loss": 3.5892,
+      "step": 3171
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.729454545454546e-06,
+      "loss": 3.6911,
+      "step": 3172
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.7287272727272727e-06,
+      "loss": 3.8425,
+      "step": 3173
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.728e-06,
+      "loss": 3.7588,
+      "step": 3174
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 3.3412,
+      "step": 3175
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.7265454545454545e-06,
+      "loss": 3.6781,
+      "step": 3176
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.725818181818182e-06,
+      "loss": 3.8179,
+      "step": 3177
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.7250909090909094e-06,
+      "loss": 3.7445,
+      "step": 3178
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.7243636363636367e-06,
+      "loss": 3.5515,
+      "step": 3179
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.723636363636364e-06,
+      "loss": 3.5424,
+      "step": 3180
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.722909090909091e-06,
+      "loss": 3.5285,
+      "step": 3181
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.7221818181818184e-06,
+      "loss": 3.4286,
+      "step": 3182
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.7214545454545453e-06,
+      "loss": 3.6657,
+      "step": 3183
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.7207272727272734e-06,
+      "loss": 3.508,
+      "step": 3184
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 3.6226,
+      "step": 3185
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.7192727272727275e-06,
+      "loss": 3.6292,
+      "step": 3186
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.7185454545454547e-06,
+      "loss": 3.7003,
+      "step": 3187
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.717818181818182e-06,
+      "loss": 3.6466,
+      "step": 3188
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.7170909090909092e-06,
+      "loss": 3.8892,
+      "step": 3189
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.7163636363636365e-06,
+      "loss": 3.5462,
+      "step": 3190
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.7156363636363638e-06,
+      "loss": 3.5259,
+      "step": 3191
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.7149090909090914e-06,
+      "loss": 3.8765,
+      "step": 3192
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.7141818181818187e-06,
+      "loss": 3.5929,
+      "step": 3193
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.713454545454546e-06,
+      "loss": 3.8215,
+      "step": 3194
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.7127272727272728e-06,
+      "loss": 3.792,
+      "step": 3195
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.712e-06,
+      "loss": 3.5846,
+      "step": 3196
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.7112727272727273e-06,
+      "loss": 3.4677,
+      "step": 3197
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.7105454545454546e-06,
+      "loss": 3.4415,
+      "step": 3198
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.709818181818182e-06,
+      "loss": 3.6515,
+      "step": 3199
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.7090909090909095e-06,
+      "loss": 3.5892,
+      "step": 3200
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.7083636363636367e-06,
+      "loss": 3.6682,
+      "step": 3201
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.707636363636364e-06,
+      "loss": 3.473,
+      "step": 3202
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.7069090909090913e-06,
+      "loss": 3.5726,
+      "step": 3203
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.7061818181818185e-06,
+      "loss": 3.5619,
+      "step": 3204
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.7054545454545453e-06,
+      "loss": 3.6941,
+      "step": 3205
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.7047272727272726e-06,
+      "loss": 3.5882,
+      "step": 3206
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.704e-06,
+      "loss": 3.725,
+      "step": 3207
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.7032727272727275e-06,
+      "loss": 3.7,
+      "step": 3208
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.702545454545455e-06,
+      "loss": 3.4939,
+      "step": 3209
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.701818181818182e-06,
+      "loss": 3.4174,
+      "step": 3210
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.7010909090909093e-06,
+      "loss": 3.568,
+      "step": 3211
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.7003636363636366e-06,
+      "loss": 3.4821,
+      "step": 3212
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.699636363636364e-06,
+      "loss": 3.5239,
+      "step": 3213
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.698909090909091e-06,
+      "loss": 3.5907,
+      "step": 3214
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.6981818181818188e-06,
+      "loss": 3.5559,
+      "step": 3215
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.697454545454546e-06,
+      "loss": 3.5513,
+      "step": 3216
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.6967272727272733e-06,
+      "loss": 3.4609,
+      "step": 3217
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.696e-06,
+      "loss": 3.3858,
+      "step": 3218
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.6952727272727274e-06,
+      "loss": 3.5323,
+      "step": 3219
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.6945454545454546e-06,
+      "loss": 3.507,
+      "step": 3220
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.693818181818182e-06,
+      "loss": 3.4688,
+      "step": 3221
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.693090909090909e-06,
+      "loss": 3.6635,
+      "step": 3222
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.692363636363637e-06,
+      "loss": 3.5485,
+      "step": 3223
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.691636363636364e-06,
+      "loss": 3.624,
+      "step": 3224
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.6909090909090913e-06,
+      "loss": 3.5877,
+      "step": 3225
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.6901818181818186e-06,
+      "loss": 3.5004,
+      "step": 3226
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.689454545454546e-06,
+      "loss": 3.492,
+      "step": 3227
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.6887272727272727e-06,
+      "loss": 3.6253,
+      "step": 3228
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.688e-06,
+      "loss": 3.5196,
+      "step": 3229
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.687272727272727e-06,
+      "loss": 3.4527,
+      "step": 3230
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.686545454545455e-06,
+      "loss": 3.6539,
+      "step": 3231
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.685818181818182e-06,
+      "loss": 3.4697,
+      "step": 3232
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.6850909090909094e-06,
+      "loss": 3.6854,
+      "step": 3233
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.6843636363636366e-06,
+      "loss": 3.5121,
+      "step": 3234
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.683636363636364e-06,
+      "loss": 3.6156,
+      "step": 3235
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.682909090909091e-06,
+      "loss": 3.8368,
+      "step": 3236
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.6821818181818184e-06,
+      "loss": 3.6572,
+      "step": 3237
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.6814545454545452e-06,
+      "loss": 3.6289,
+      "step": 3238
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.6807272727272733e-06,
+      "loss": 3.8487,
+      "step": 3239
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.68e-06,
+      "loss": 3.3862,
+      "step": 3240
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.6792727272727274e-06,
+      "loss": 3.7314,
+      "step": 3241
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.6785454545454547e-06,
+      "loss": 3.6191,
+      "step": 3242
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.677818181818182e-06,
+      "loss": 3.4392,
+      "step": 3243
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.677090909090909e-06,
+      "loss": 3.7098,
+      "step": 3244
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.6763636363636365e-06,
+      "loss": 3.5437,
+      "step": 3245
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.6756363636363637e-06,
+      "loss": 3.5898,
+      "step": 3246
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.6749090909090914e-06,
+      "loss": 3.7352,
+      "step": 3247
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.6741818181818187e-06,
+      "loss": 3.5098,
+      "step": 3248
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.673454545454546e-06,
+      "loss": 3.34,
+      "step": 3249
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.6727272727272727e-06,
+      "loss": 3.3742,
+      "step": 3250
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.672e-06,
+      "loss": 3.8008,
+      "step": 3251
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.6712727272727273e-06,
+      "loss": 3.4353,
+      "step": 3252
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.6705454545454545e-06,
+      "loss": 3.666,
+      "step": 3253
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.669818181818182e-06,
+      "loss": 3.95,
+      "step": 3254
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.6690909090909094e-06,
+      "loss": 3.5534,
+      "step": 3255
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.6683636363636367e-06,
+      "loss": 3.6715,
+      "step": 3256
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.667636363636364e-06,
+      "loss": 3.5086,
+      "step": 3257
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.6669090909090912e-06,
+      "loss": 3.486,
+      "step": 3258
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.6661818181818185e-06,
+      "loss": 3.7581,
+      "step": 3259
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.6654545454545453e-06,
+      "loss": 3.661,
+      "step": 3260
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.6647272727272726e-06,
+      "loss": 3.364,
+      "step": 3261
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.6640000000000007e-06,
+      "loss": 3.5187,
+      "step": 3262
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.6632727272727275e-06,
+      "loss": 3.5466,
+      "step": 3263
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.6625454545454548e-06,
+      "loss": 3.5614,
+      "step": 3264
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.661818181818182e-06,
+      "loss": 3.5643,
+      "step": 3265
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.6610909090909093e-06,
+      "loss": 3.8094,
+      "step": 3266
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.6603636363636365e-06,
+      "loss": 3.4755,
+      "step": 3267
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.6596363636363638e-06,
+      "loss": 3.6815,
+      "step": 3268
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.658909090909091e-06,
+      "loss": 3.586,
+      "step": 3269
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.6581818181818187e-06,
+      "loss": 3.4476,
+      "step": 3270
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.657454545454546e-06,
+      "loss": 3.5301,
+      "step": 3271
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.6567272727272732e-06,
+      "loss": 3.7403,
+      "step": 3272
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.656e-06,
+      "loss": 3.5638,
+      "step": 3273
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.6552727272727273e-06,
+      "loss": 3.7684,
+      "step": 3274
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.6545454545454546e-06,
+      "loss": 3.4453,
+      "step": 3275
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.653818181818182e-06,
+      "loss": 3.5769,
+      "step": 3276
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.653090909090909e-06,
+      "loss": 3.8912,
+      "step": 3277
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.6523636363636368e-06,
+      "loss": 3.5276,
+      "step": 3278
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.651636363636364e-06,
+      "loss": 3.7964,
+      "step": 3279
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.6509090909090913e-06,
+      "loss": 3.4009,
+      "step": 3280
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.6501818181818185e-06,
+      "loss": 3.625,
+      "step": 3281
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.649454545454546e-06,
+      "loss": 3.7129,
+      "step": 3282
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.6487272727272726e-06,
+      "loss": 3.5707,
+      "step": 3283
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.648e-06,
+      "loss": 3.5001,
+      "step": 3284
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.647272727272727e-06,
+      "loss": 3.5648,
+      "step": 3285
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.646545454545455e-06,
+      "loss": 3.7139,
+      "step": 3286
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.645818181818182e-06,
+      "loss": 3.3723,
+      "step": 3287
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.6450909090909093e-06,
+      "loss": 3.7695,
+      "step": 3288
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.6443636363636366e-06,
+      "loss": 3.5248,
+      "step": 3289
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.643636363636364e-06,
+      "loss": 3.4633,
+      "step": 3290
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.642909090909091e-06,
+      "loss": 3.5549,
+      "step": 3291
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.6421818181818184e-06,
+      "loss": 3.4482,
+      "step": 3292
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.641454545454546e-06,
+      "loss": 3.637,
+      "step": 3293
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.6407272727272733e-06,
+      "loss": 3.6146,
+      "step": 3294
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.64e-06,
+      "loss": 3.7551,
+      "step": 3295
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.6392727272727274e-06,
+      "loss": 3.5325,
+      "step": 3296
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.6385454545454546e-06,
+      "loss": 3.5237,
+      "step": 3297
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.637818181818182e-06,
+      "loss": 3.6949,
+      "step": 3298
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.637090909090909e-06,
+      "loss": 3.4553,
+      "step": 3299
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.6363636363636364e-06,
+      "loss": 3.6074,
+      "step": 3300
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.635636363636364e-06,
+      "loss": 3.3591,
+      "step": 3301
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.6349090909090914e-06,
+      "loss": 3.6226,
+      "step": 3302
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.6341818181818186e-06,
+      "loss": 3.578,
+      "step": 3303
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.633454545454546e-06,
+      "loss": 3.4869,
+      "step": 3304
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.6327272727272727e-06,
+      "loss": 3.6729,
+      "step": 3305
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.632e-06,
+      "loss": 3.6676,
+      "step": 3306
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.6312727272727272e-06,
+      "loss": 3.3395,
+      "step": 3307
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.6305454545454545e-06,
+      "loss": 3.5196,
+      "step": 3308
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.629818181818182e-06,
+      "loss": 3.4144,
+      "step": 3309
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.6290909090909094e-06,
+      "loss": 3.6254,
+      "step": 3310
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.6283636363636367e-06,
+      "loss": 3.7582,
+      "step": 3311
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.627636363636364e-06,
+      "loss": 3.9147,
+      "step": 3312
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.626909090909091e-06,
+      "loss": 3.6733,
+      "step": 3313
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.6261818181818184e-06,
+      "loss": 3.4248,
+      "step": 3314
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.6254545454545453e-06,
+      "loss": 3.4891,
+      "step": 3315
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.6247272727272725e-06,
+      "loss": 3.7118,
+      "step": 3316
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.6240000000000006e-06,
+      "loss": 3.5712,
+      "step": 3317
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.6232727272727275e-06,
+      "loss": 3.5945,
+      "step": 3318
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.6225454545454547e-06,
+      "loss": 3.3732,
+      "step": 3319
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.621818181818182e-06,
+      "loss": 3.5652,
+      "step": 3320
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.6210909090909092e-06,
+      "loss": 3.4811,
+      "step": 3321
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.6203636363636365e-06,
+      "loss": 3.695,
+      "step": 3322
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2.6196363636363637e-06,
+      "loss": 3.6122,
+      "step": 3323
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2.6189090909090914e-06,
+      "loss": 3.6199,
+      "step": 3324
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.16216216216216217,
+      "eval_loss": 3.5709187984466553,
+      "eval_runtime": 143.6526,
+      "eval_samples_per_second": 2.576,
+      "eval_steps_per_second": 0.647,
+      "step": 3324
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2.6181818181818187e-06,
+      "loss": 3.6687,
+      "step": 3325
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2.617454545454546e-06,
+      "loss": 3.6513,
+      "step": 3326
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2.616727272727273e-06,
+      "loss": 3.5614,
+      "step": 3327
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2.616e-06,
+      "loss": 3.3967,
+      "step": 3328
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.6152727272727273e-06,
+      "loss": 3.5914,
+      "step": 3329
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.6145454545454545e-06,
+      "loss": 3.3913,
+      "step": 3330
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.613818181818182e-06,
+      "loss": 3.6028,
+      "step": 3331
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.6130909090909095e-06,
+      "loss": 3.3838,
+      "step": 3332
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.6123636363636367e-06,
+      "loss": 3.5151,
+      "step": 3333
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.611636363636364e-06,
+      "loss": 3.5456,
+      "step": 3334
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.6109090909090912e-06,
+      "loss": 3.4216,
+      "step": 3335
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.6101818181818185e-06,
+      "loss": 3.5101,
+      "step": 3336
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.6094545454545458e-06,
+      "loss": 3.887,
+      "step": 3337
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.6087272727272726e-06,
+      "loss": 3.9182,
+      "step": 3338
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.608e-06,
+      "loss": 3.3385,
+      "step": 3339
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.6072727272727275e-06,
+      "loss": 3.6183,
+      "step": 3340
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.6065454545454548e-06,
+      "loss": 3.7045,
+      "step": 3341
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.605818181818182e-06,
+      "loss": 3.5415,
+      "step": 3342
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.6050909090909093e-06,
+      "loss": 3.6784,
+      "step": 3343
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.6043636363636366e-06,
+      "loss": 3.4749,
+      "step": 3344
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.603636363636364e-06,
+      "loss": 3.6393,
+      "step": 3345
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.602909090909091e-06,
+      "loss": 3.8456,
+      "step": 3346
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.6021818181818183e-06,
+      "loss": 3.4238,
+      "step": 3347
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.601454545454546e-06,
+      "loss": 3.4339,
+      "step": 3348
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.6007272727272733e-06,
+      "loss": 3.7485,
+      "step": 3349
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.6e-06,
+      "loss": 3.4831,
+      "step": 3350
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.5992727272727274e-06,
+      "loss": 3.6148,
+      "step": 3351
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.5985454545454546e-06,
+      "loss": 3.7116,
+      "step": 3352
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.597818181818182e-06,
+      "loss": 3.5563,
+      "step": 3353
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.597090909090909e-06,
+      "loss": 3.5963,
+      "step": 3354
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.5963636363636364e-06,
+      "loss": 3.6463,
+      "step": 3355
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.595636363636364e-06,
+      "loss": 3.6049,
+      "step": 3356
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.5949090909090913e-06,
+      "loss": 3.6395,
+      "step": 3357
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.5941818181818186e-06,
+      "loss": 3.3389,
+      "step": 3358
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.593454545454546e-06,
+      "loss": 3.5266,
+      "step": 3359
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.5927272727272727e-06,
+      "loss": 3.4551,
+      "step": 3360
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.592e-06,
+      "loss": 3.9274,
+      "step": 3361
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.591272727272727e-06,
+      "loss": 3.626,
+      "step": 3362
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.590545454545455e-06,
+      "loss": 3.761,
+      "step": 3363
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.589818181818182e-06,
+      "loss": 3.471,
+      "step": 3364
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.5890909090909094e-06,
+      "loss": 3.5168,
+      "step": 3365
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.5883636363636366e-06,
+      "loss": 3.5765,
+      "step": 3366
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.587636363636364e-06,
+      "loss": 3.5846,
+      "step": 3367
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.586909090909091e-06,
+      "loss": 3.5233,
+      "step": 3368
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.5861818181818184e-06,
+      "loss": 3.4736,
+      "step": 3369
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.5854545454545456e-06,
+      "loss": 3.5959,
+      "step": 3370
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.5847272727272733e-06,
+      "loss": 3.6953,
+      "step": 3371
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.5840000000000006e-06,
+      "loss": 3.4129,
+      "step": 3372
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.5832727272727274e-06,
+      "loss": 3.647,
+      "step": 3373
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.5825454545454547e-06,
+      "loss": 3.4197,
+      "step": 3374
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.581818181818182e-06,
+      "loss": 3.3225,
+      "step": 3375
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.581090909090909e-06,
+      "loss": 3.7648,
+      "step": 3376
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.5803636363636364e-06,
+      "loss": 3.5954,
+      "step": 3377
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.5796363636363637e-06,
+      "loss": 3.3889,
+      "step": 3378
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.5789090909090914e-06,
+      "loss": 3.4261,
+      "step": 3379
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.5781818181818186e-06,
+      "loss": 3.3945,
+      "step": 3380
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.577454545454546e-06,
+      "loss": 3.4418,
+      "step": 3381
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.576727272727273e-06,
+      "loss": 3.4584,
+      "step": 3382
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.576e-06,
+      "loss": 3.4686,
+      "step": 3383
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.5752727272727272e-06,
+      "loss": 3.5309,
+      "step": 3384
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.5745454545454545e-06,
+      "loss": 3.4569,
+      "step": 3385
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.5738181818181818e-06,
+      "loss": 3.4688,
+      "step": 3386
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.5730909090909094e-06,
+      "loss": 3.6214,
+      "step": 3387
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.5723636363636367e-06,
+      "loss": 3.5262,
+      "step": 3388
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.571636363636364e-06,
+      "loss": 3.4522,
+      "step": 3389
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.570909090909091e-06,
+      "loss": 3.7767,
+      "step": 3390
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.5701818181818185e-06,
+      "loss": 3.5461,
+      "step": 3391
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.5694545454545457e-06,
+      "loss": 3.5497,
+      "step": 3392
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.5687272727272726e-06,
+      "loss": 3.3176,
+      "step": 3393
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.568e-06,
+      "loss": 3.5824,
+      "step": 3394
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.5672727272727275e-06,
+      "loss": 3.6843,
+      "step": 3395
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.5665454545454547e-06,
+      "loss": 3.4855,
+      "step": 3396
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.565818181818182e-06,
+      "loss": 3.5474,
+      "step": 3397
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.5650909090909093e-06,
+      "loss": 3.6545,
+      "step": 3398
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.5643636363636365e-06,
+      "loss": 3.3802,
+      "step": 3399
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.5636363636363638e-06,
+      "loss": 3.5765,
+      "step": 3400
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.562909090909091e-06,
+      "loss": 3.4804,
+      "step": 3401
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.5621818181818187e-06,
+      "loss": 3.7196,
+      "step": 3402
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.561454545454546e-06,
+      "loss": 3.5061,
+      "step": 3403
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.5607272727272732e-06,
+      "loss": 3.6146,
+      "step": 3404
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.56e-06,
+      "loss": 3.531,
+      "step": 3405
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.5592727272727273e-06,
+      "loss": 3.707,
+      "step": 3406
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.5585454545454546e-06,
+      "loss": 3.589,
+      "step": 3407
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.557818181818182e-06,
+      "loss": 3.5476,
+      "step": 3408
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.557090909090909e-06,
+      "loss": 3.3969,
+      "step": 3409
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.5563636363636368e-06,
+      "loss": 3.4906,
+      "step": 3410
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.555636363636364e-06,
+      "loss": 3.5071,
+      "step": 3411
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.5549090909090913e-06,
+      "loss": 3.4036,
+      "step": 3412
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.5541818181818185e-06,
+      "loss": 3.6483,
+      "step": 3413
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.5534545454545458e-06,
+      "loss": 3.3747,
+      "step": 3414
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.552727272727273e-06,
+      "loss": 3.4789,
+      "step": 3415
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.552e-06,
+      "loss": 3.4388,
+      "step": 3416
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.551272727272727e-06,
+      "loss": 3.5002,
+      "step": 3417
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.550545454545455e-06,
+      "loss": 3.622,
+      "step": 3418
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.549818181818182e-06,
+      "loss": 3.5739,
+      "step": 3419
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.5490909090909093e-06,
+      "loss": 3.6346,
+      "step": 3420
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.5483636363636366e-06,
+      "loss": 3.9001,
+      "step": 3421
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.547636363636364e-06,
+      "loss": 3.5514,
+      "step": 3422
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.546909090909091e-06,
+      "loss": 3.5601,
+      "step": 3423
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.5461818181818184e-06,
+      "loss": 3.4835,
+      "step": 3424
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.5454545454545456e-06,
+      "loss": 3.5189,
+      "step": 3425
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.5447272727272733e-06,
+      "loss": 3.8124,
+      "step": 3426
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.5440000000000005e-06,
+      "loss": 3.6898,
+      "step": 3427
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.5432727272727274e-06,
+      "loss": 3.5897,
+      "step": 3428
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.5425454545454546e-06,
+      "loss": 3.4224,
+      "step": 3429
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.541818181818182e-06,
+      "loss": 3.6297,
+      "step": 3430
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.541090909090909e-06,
+      "loss": 3.5469,
+      "step": 3431
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.5403636363636364e-06,
+      "loss": 3.6364,
+      "step": 3432
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.539636363636364e-06,
+      "loss": 3.6949,
+      "step": 3433
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.5389090909090913e-06,
+      "loss": 3.5662,
+      "step": 3434
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.5381818181818186e-06,
+      "loss": 3.6413,
+      "step": 3435
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.537454545454546e-06,
+      "loss": 3.3848,
+      "step": 3436
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.536727272727273e-06,
+      "loss": 3.7132,
+      "step": 3437
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.536e-06,
+      "loss": 3.674,
+      "step": 3438
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.535272727272727e-06,
+      "loss": 3.443,
+      "step": 3439
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.5345454545454545e-06,
+      "loss": 3.2387,
+      "step": 3440
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.533818181818182e-06,
+      "loss": 3.6545,
+      "step": 3441
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.5330909090909094e-06,
+      "loss": 3.5826,
+      "step": 3442
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.5323636363636367e-06,
+      "loss": 3.5667,
+      "step": 3443
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.531636363636364e-06,
+      "loss": 3.6181,
+      "step": 3444
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.530909090909091e-06,
+      "loss": 3.7134,
+      "step": 3445
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.5301818181818184e-06,
+      "loss": 3.5666,
+      "step": 3446
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.5294545454545457e-06,
+      "loss": 3.3216,
+      "step": 3447
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.5287272727272725e-06,
+      "loss": 3.4603,
+      "step": 3448
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.5280000000000006e-06,
+      "loss": 3.8482,
+      "step": 3449
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.5272727272727274e-06,
+      "loss": 3.7839,
+      "step": 3450
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.5265454545454547e-06,
+      "loss": 3.7921,
+      "step": 3451
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.525818181818182e-06,
+      "loss": 3.6836,
+      "step": 3452
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.5250909090909092e-06,
+      "loss": 3.2373,
+      "step": 3453
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.5243636363636365e-06,
+      "loss": 3.7699,
+      "step": 3454
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.5236363636363637e-06,
+      "loss": 3.4464,
+      "step": 3455
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.522909090909091e-06,
+      "loss": 3.6623,
+      "step": 3456
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.5221818181818187e-06,
+      "loss": 3.495,
+      "step": 3457
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.521454545454546e-06,
+      "loss": 3.5138,
+      "step": 3458
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.520727272727273e-06,
+      "loss": 3.7375,
+      "step": 3459
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.52e-06,
+      "loss": 3.416,
+      "step": 3460
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.5192727272727273e-06,
+      "loss": 3.8493,
+      "step": 3461
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.5185454545454545e-06,
+      "loss": 3.7601,
+      "step": 3462
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.5178181818181818e-06,
+      "loss": 3.4146,
+      "step": 3463
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.517090909090909e-06,
+      "loss": 3.5544,
+      "step": 3464
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.5163636363636367e-06,
+      "loss": 3.3524,
+      "step": 3465
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.515636363636364e-06,
+      "loss": 3.4527,
+      "step": 3466
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.5149090909090912e-06,
+      "loss": 3.6749,
+      "step": 3467
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.5141818181818185e-06,
+      "loss": 3.6064,
+      "step": 3468
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.5134545454545457e-06,
+      "loss": 3.5328,
+      "step": 3469
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.512727272727273e-06,
+      "loss": 3.6218,
+      "step": 3470
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.512e-06,
+      "loss": 3.6379,
+      "step": 3471
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.511272727272728e-06,
+      "loss": 3.8409,
+      "step": 3472
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.5105454545454548e-06,
+      "loss": 3.4927,
+      "step": 3473
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.509818181818182e-06,
+      "loss": 3.4456,
+      "step": 3474
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.5090909090909093e-06,
+      "loss": 3.5489,
+      "step": 3475
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.5083636363636365e-06,
+      "loss": 3.3287,
+      "step": 3476
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.507636363636364e-06,
+      "loss": 3.5221,
+      "step": 3477
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.506909090909091e-06,
+      "loss": 3.7197,
+      "step": 3478
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.5061818181818183e-06,
+      "loss": 3.4081,
+      "step": 3479
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.505454545454546e-06,
+      "loss": 3.442,
+      "step": 3480
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.5047272727272732e-06,
+      "loss": 3.4591,
+      "step": 3481
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.5040000000000005e-06,
+      "loss": 3.4561,
+      "step": 3482
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.5032727272727273e-06,
+      "loss": 3.655,
+      "step": 3483
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.5025454545454546e-06,
+      "loss": 3.4668,
+      "step": 3484
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.501818181818182e-06,
+      "loss": 3.6321,
+      "step": 3485
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.501090909090909e-06,
+      "loss": 3.6584,
+      "step": 3486
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.5003636363636364e-06,
+      "loss": 3.4138,
+      "step": 3487
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.4996363636363636e-06,
+      "loss": 3.4605,
+      "step": 3488
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.498909090909091e-06,
+      "loss": 3.641,
+      "step": 3489
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.4981818181818186e-06,
+      "loss": 3.6025,
+      "step": 3490
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.497454545454546e-06,
+      "loss": 3.6065,
+      "step": 3491
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.496727272727273e-06,
+      "loss": 3.8388,
+      "step": 3492
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.496e-06,
+      "loss": 3.5796,
+      "step": 3493
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.4952727272727276e-06,
+      "loss": 3.5374,
+      "step": 3494
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.494545454545455e-06,
+      "loss": 3.6418,
+      "step": 3495
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.493818181818182e-06,
+      "loss": 3.5601,
+      "step": 3496
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.4930909090909094e-06,
+      "loss": 3.473,
+      "step": 3497
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.4923636363636366e-06,
+      "loss": 3.4665,
+      "step": 3498
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.491636363636364e-06,
+      "loss": 3.5665,
+      "step": 3499
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.490909090909091e-06,
+      "loss": 3.6737,
+      "step": 3500
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.4901818181818184e-06,
+      "loss": 3.606,
+      "step": 3501
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.4894545454545456e-06,
+      "loss": 3.8716,
+      "step": 3502
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.488727272727273e-06,
+      "loss": 3.3933,
+      "step": 3503
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.488e-06,
+      "loss": 3.5322,
+      "step": 3504
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.4872727272727274e-06,
+      "loss": 3.8202,
+      "step": 3505
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.4865454545454547e-06,
+      "loss": 3.4104,
+      "step": 3506
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.485818181818182e-06,
+      "loss": 3.416,
+      "step": 3507
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.485090909090909e-06,
+      "loss": 3.4946,
+      "step": 3508
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.484363636363637e-06,
+      "loss": 3.5693,
+      "step": 3509
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.4836363636363637e-06,
+      "loss": 3.4009,
+      "step": 3510
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.482909090909091e-06,
+      "loss": 3.5296,
+      "step": 3511
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.482181818181818e-06,
+      "loss": 3.5778,
+      "step": 3512
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.481454545454546e-06,
+      "loss": 3.6095,
+      "step": 3513
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.480727272727273e-06,
+      "loss": 3.6072,
+      "step": 3514
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 3.7373,
+      "step": 3515
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.4792727272727272e-06,
+      "loss": 3.528,
+      "step": 3516
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.478545454545455e-06,
+      "loss": 3.7475,
+      "step": 3517
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.477818181818182e-06,
+      "loss": 3.4541,
+      "step": 3518
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.4770909090909094e-06,
+      "loss": 3.5158,
+      "step": 3519
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.4763636363636367e-06,
+      "loss": 3.5455,
+      "step": 3520
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.475636363636364e-06,
+      "loss": 3.64,
+      "step": 3521
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.474909090909091e-06,
+      "loss": 3.6964,
+      "step": 3522
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.4741818181818184e-06,
+      "loss": 3.441,
+      "step": 3523
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.4734545454545457e-06,
+      "loss": 3.7242,
+      "step": 3524
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.472727272727273e-06,
+      "loss": 3.4775,
+      "step": 3525
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.4720000000000002e-06,
+      "loss": 3.7025,
+      "step": 3526
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.4712727272727275e-06,
+      "loss": 3.5962,
+      "step": 3527
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.4705454545454547e-06,
+      "loss": 3.6743,
+      "step": 3528
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.469818181818182e-06,
+      "loss": 3.3503,
+      "step": 3529
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.4690909090909092e-06,
+      "loss": 3.5505,
+      "step": 3530
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.4683636363636365e-06,
+      "loss": 3.4482,
+      "step": 3531
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.4676363636363638e-06,
+      "loss": 3.4731,
+      "step": 3532
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.466909090909091e-06,
+      "loss": 3.622,
+      "step": 3533
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.4661818181818183e-06,
+      "loss": 3.6729,
+      "step": 3534
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.4654545454545455e-06,
+      "loss": 3.6054,
+      "step": 3535
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.4647272727272728e-06,
+      "loss": 3.8336,
+      "step": 3536
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.4640000000000005e-06,
+      "loss": 3.4752,
+      "step": 3537
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.4632727272727273e-06,
+      "loss": 3.6568,
+      "step": 3538
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.4625454545454546e-06,
+      "loss": 3.6751,
+      "step": 3539
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.461818181818182e-06,
+      "loss": 3.4633,
+      "step": 3540
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.4610909090909095e-06,
+      "loss": 3.7374,
+      "step": 3541
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.4603636363636367e-06,
+      "loss": 3.6718,
+      "step": 3542
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.4596363636363636e-06,
+      "loss": 3.3836,
+      "step": 3543
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.4589090909090913e-06,
+      "loss": 3.5094,
+      "step": 3544
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.4581818181818185e-06,
+      "loss": 3.4996,
+      "step": 3545
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.4574545454545458e-06,
+      "loss": 3.3876,
+      "step": 3546
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.456727272727273e-06,
+      "loss": 3.5515,
+      "step": 3547
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.4560000000000003e-06,
+      "loss": 3.4716,
+      "step": 3548
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.4552727272727275e-06,
+      "loss": 3.2885,
+      "step": 3549
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.454545454545455e-06,
+      "loss": 3.6281,
+      "step": 3550
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.453818181818182e-06,
+      "loss": 3.4293,
+      "step": 3551
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.4530909090909093e-06,
+      "loss": 3.5695,
+      "step": 3552
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.4523636363636366e-06,
+      "loss": 3.5233,
+      "step": 3553
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.451636363636364e-06,
+      "loss": 3.593,
+      "step": 3554
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.450909090909091e-06,
+      "loss": 3.6641,
+      "step": 3555
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.4501818181818183e-06,
+      "loss": 3.5269,
+      "step": 3556
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.4494545454545456e-06,
+      "loss": 3.386,
+      "step": 3557
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.448727272727273e-06,
+      "loss": 3.817,
+      "step": 3558
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.448e-06,
+      "loss": 3.3379,
+      "step": 3559
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.4472727272727274e-06,
+      "loss": 3.7052,
+      "step": 3560
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.4465454545454546e-06,
+      "loss": 3.8521,
+      "step": 3561
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.445818181818182e-06,
+      "loss": 3.4787,
+      "step": 3562
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.445090909090909e-06,
+      "loss": 3.6801,
+      "step": 3563
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.444363636363637e-06,
+      "loss": 3.6077,
+      "step": 3564
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.443636363636364e-06,
+      "loss": 3.7539,
+      "step": 3565
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.442909090909091e-06,
+      "loss": 3.3503,
+      "step": 3566
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.442181818181818e-06,
+      "loss": 3.6793,
+      "step": 3567
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.441454545454546e-06,
+      "loss": 3.5106,
+      "step": 3568
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.440727272727273e-06,
+      "loss": 3.6088,
+      "step": 3569
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 3.4834,
+      "step": 3570
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.439272727272727e-06,
+      "loss": 3.7011,
+      "step": 3571
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.438545454545455e-06,
+      "loss": 3.6487,
+      "step": 3572
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.437818181818182e-06,
+      "loss": 3.4991,
+      "step": 3573
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.4370909090909094e-06,
+      "loss": 3.7846,
+      "step": 3574
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.4363636363636366e-06,
+      "loss": 3.4475,
+      "step": 3575
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.435636363636364e-06,
+      "loss": 4.0529,
+      "step": 3576
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.434909090909091e-06,
+      "loss": 3.4488,
+      "step": 3577
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.4341818181818184e-06,
+      "loss": 3.6639,
+      "step": 3578
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.4334545454545457e-06,
+      "loss": 3.5269,
+      "step": 3579
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.432727272727273e-06,
+      "loss": 3.4352,
+      "step": 3580
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.432e-06,
+      "loss": 3.6776,
+      "step": 3581
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.4312727272727274e-06,
+      "loss": 3.6584,
+      "step": 3582
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.4305454545454547e-06,
+      "loss": 3.4621,
+      "step": 3583
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.429818181818182e-06,
+      "loss": 3.457,
+      "step": 3584
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.429090909090909e-06,
+      "loss": 3.5343,
+      "step": 3585
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.4283636363636365e-06,
+      "loss": 3.5553,
+      "step": 3586
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.427636363636364e-06,
+      "loss": 3.6002,
+      "step": 3587
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.426909090909091e-06,
+      "loss": 3.5498,
+      "step": 3588
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.4261818181818182e-06,
+      "loss": 3.644,
+      "step": 3589
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.4254545454545455e-06,
+      "loss": 3.3987,
+      "step": 3590
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.424727272727273e-06,
+      "loss": 3.5279,
+      "step": 3591
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.4240000000000004e-06,
+      "loss": 3.6176,
+      "step": 3592
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.4232727272727273e-06,
+      "loss": 3.3215,
+      "step": 3593
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.4225454545454545e-06,
+      "loss": 3.6126,
+      "step": 3594
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.421818181818182e-06,
+      "loss": 3.5597,
+      "step": 3595
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.4210909090909095e-06,
+      "loss": 3.8473,
+      "step": 3596
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.4203636363636367e-06,
+      "loss": 3.382,
+      "step": 3597
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.4196363636363635e-06,
+      "loss": 3.4323,
+      "step": 3598
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.4189090909090912e-06,
+      "loss": 3.3389,
+      "step": 3599
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.4181818181818185e-06,
+      "loss": 3.5779,
+      "step": 3600
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.4174545454545457e-06,
+      "loss": 3.4025,
+      "step": 3601
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.17297297297297298,
+      "eval_loss": 3.525432825088501,
+      "eval_runtime": 143.7462,
+      "eval_samples_per_second": 2.574,
+      "eval_steps_per_second": 0.647,
+      "step": 3601
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.416727272727273e-06,
+      "loss": 3.3477,
+      "step": 3602
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.4160000000000002e-06,
+      "loss": 3.4174,
+      "step": 3603
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.4152727272727275e-06,
+      "loss": 3.398,
+      "step": 3604
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.4145454545454548e-06,
+      "loss": 3.7036,
+      "step": 3605
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.413818181818182e-06,
+      "loss": 3.4664,
+      "step": 3606
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.4130909090909093e-06,
+      "loss": 3.6846,
+      "step": 3607
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.4123636363636365e-06,
+      "loss": 3.6296,
+      "step": 3608
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.4116363636363638e-06,
+      "loss": 3.5091,
+      "step": 3609
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.410909090909091e-06,
+      "loss": 3.4543,
+      "step": 3610
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.4101818181818183e-06,
+      "loss": 3.7015,
+      "step": 3611
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.4094545454545456e-06,
+      "loss": 3.437,
+      "step": 3612
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.408727272727273e-06,
+      "loss": 3.605,
+      "step": 3613
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.408e-06,
+      "loss": 3.5318,
+      "step": 3614
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.4072727272727277e-06,
+      "loss": 3.7078,
+      "step": 3615
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.4065454545454546e-06,
+      "loss": 3.5605,
+      "step": 3616
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.405818181818182e-06,
+      "loss": 3.7233,
+      "step": 3617
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.4050909090909095e-06,
+      "loss": 3.591,
+      "step": 3618
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.4043636363636368e-06,
+      "loss": 3.5863,
+      "step": 3619
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.403636363636364e-06,
+      "loss": 3.4156,
+      "step": 3620
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.402909090909091e-06,
+      "loss": 3.2112,
+      "step": 3621
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.4021818181818185e-06,
+      "loss": 3.4038,
+      "step": 3622
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.401454545454546e-06,
+      "loss": 3.6256,
+      "step": 3623
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.400727272727273e-06,
+      "loss": 3.363,
+      "step": 3624
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.7302,
+      "step": 3625
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.3992727272727276e-06,
+      "loss": 3.3199,
+      "step": 3626
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.398545454545455e-06,
+      "loss": 3.6101,
+      "step": 3627
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.397818181818182e-06,
+      "loss": 3.7507,
+      "step": 3628
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.3970909090909093e-06,
+      "loss": 3.4449,
+      "step": 3629
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.3963636363636366e-06,
+      "loss": 3.7402,
+      "step": 3630
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.395636363636364e-06,
+      "loss": 3.5025,
+      "step": 3631
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.394909090909091e-06,
+      "loss": 3.6047,
+      "step": 3632
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.3941818181818184e-06,
+      "loss": 3.6601,
+      "step": 3633
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.3934545454545456e-06,
+      "loss": 3.6164,
+      "step": 3634
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.392727272727273e-06,
+      "loss": 3.3752,
+      "step": 3635
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.392e-06,
+      "loss": 3.7049,
+      "step": 3636
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.3912727272727274e-06,
+      "loss": 3.6358,
+      "step": 3637
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.3905454545454547e-06,
+      "loss": 3.9546,
+      "step": 3638
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.389818181818182e-06,
+      "loss": 3.4073,
+      "step": 3639
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.389090909090909e-06,
+      "loss": 3.6745,
+      "step": 3640
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.3883636363636364e-06,
+      "loss": 3.5006,
+      "step": 3641
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.387636363636364e-06,
+      "loss": 3.789,
+      "step": 3642
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.386909090909091e-06,
+      "loss": 3.6119,
+      "step": 3643
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.386181818181818e-06,
+      "loss": 3.4353,
+      "step": 3644
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.3854545454545454e-06,
+      "loss": 3.3746,
+      "step": 3645
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.384727272727273e-06,
+      "loss": 3.5512,
+      "step": 3646
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.3840000000000004e-06,
+      "loss": 3.4349,
+      "step": 3647
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.3832727272727272e-06,
+      "loss": 3.5845,
+      "step": 3648
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.3825454545454545e-06,
+      "loss": 3.5994,
+      "step": 3649
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.381818181818182e-06,
+      "loss": 3.4358,
+      "step": 3650
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.3810909090909094e-06,
+      "loss": 3.4449,
+      "step": 3651
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.3803636363636367e-06,
+      "loss": 3.5179,
+      "step": 3652
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.379636363636364e-06,
+      "loss": 3.217,
+      "step": 3653
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.378909090909091e-06,
+      "loss": 3.3622,
+      "step": 3654
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.3781818181818184e-06,
+      "loss": 3.378,
+      "step": 3655
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.3774545454545457e-06,
+      "loss": 3.4159,
+      "step": 3656
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.376727272727273e-06,
+      "loss": 3.6081,
+      "step": 3657
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.376e-06,
+      "loss": 3.4498,
+      "step": 3658
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.3752727272727275e-06,
+      "loss": 3.6704,
+      "step": 3659
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.3745454545454547e-06,
+      "loss": 3.4164,
+      "step": 3660
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.373818181818182e-06,
+      "loss": 3.6679,
+      "step": 3661
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.3730909090909092e-06,
+      "loss": 3.7429,
+      "step": 3662
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.3723636363636365e-06,
+      "loss": 3.4236,
+      "step": 3663
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.3716363636363637e-06,
+      "loss": 3.6616,
+      "step": 3664
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.3709090909090914e-06,
+      "loss": 3.5935,
+      "step": 3665
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.3701818181818183e-06,
+      "loss": 3.6482,
+      "step": 3666
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.3694545454545455e-06,
+      "loss": 3.5314,
+      "step": 3667
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.3687272727272728e-06,
+      "loss": 3.7242,
+      "step": 3668
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.3680000000000005e-06,
+      "loss": 3.373,
+      "step": 3669
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.3672727272727277e-06,
+      "loss": 3.782,
+      "step": 3670
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.3665454545454545e-06,
+      "loss": 3.3761,
+      "step": 3671
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.365818181818182e-06,
+      "loss": 3.2212,
+      "step": 3672
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.3650909090909095e-06,
+      "loss": 3.5118,
+      "step": 3673
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.3643636363636367e-06,
+      "loss": 3.5473,
+      "step": 3674
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.363636363636364e-06,
+      "loss": 3.5094,
+      "step": 3675
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.362909090909091e-06,
+      "loss": 3.4368,
+      "step": 3676
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.3621818181818185e-06,
+      "loss": 3.5221,
+      "step": 3677
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.3614545454545458e-06,
+      "loss": 3.3615,
+      "step": 3678
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.360727272727273e-06,
+      "loss": 3.3598,
+      "step": 3679
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 3.2999,
+      "step": 3680
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.3592727272727275e-06,
+      "loss": 3.45,
+      "step": 3681
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.3585454545454548e-06,
+      "loss": 3.4126,
+      "step": 3682
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.357818181818182e-06,
+      "loss": 3.4601,
+      "step": 3683
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.3570909090909093e-06,
+      "loss": 3.4174,
+      "step": 3684
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.3563636363636366e-06,
+      "loss": 3.5005,
+      "step": 3685
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.355636363636364e-06,
+      "loss": 3.6915,
+      "step": 3686
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.354909090909091e-06,
+      "loss": 3.5962,
+      "step": 3687
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.3541818181818183e-06,
+      "loss": 3.5294,
+      "step": 3688
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.3534545454545456e-06,
+      "loss": 3.5805,
+      "step": 3689
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.352727272727273e-06,
+      "loss": 3.4311,
+      "step": 3690
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.352e-06,
+      "loss": 3.5927,
+      "step": 3691
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.3512727272727278e-06,
+      "loss": 3.2936,
+      "step": 3692
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.3505454545454546e-06,
+      "loss": 3.3017,
+      "step": 3693
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.349818181818182e-06,
+      "loss": 3.3625,
+      "step": 3694
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.349090909090909e-06,
+      "loss": 3.6532,
+      "step": 3695
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.348363636363637e-06,
+      "loss": 3.5263,
+      "step": 3696
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.347636363636364e-06,
+      "loss": 3.5593,
+      "step": 3697
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.346909090909091e-06,
+      "loss": 3.5218,
+      "step": 3698
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.346181818181818e-06,
+      "loss": 3.5983,
+      "step": 3699
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.345454545454546e-06,
+      "loss": 3.5414,
+      "step": 3700
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.344727272727273e-06,
+      "loss": 3.4739,
+      "step": 3701
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.3440000000000003e-06,
+      "loss": 3.5597,
+      "step": 3702
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.343272727272727e-06,
+      "loss": 3.498,
+      "step": 3703
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.342545454545455e-06,
+      "loss": 3.6122,
+      "step": 3704
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.341818181818182e-06,
+      "loss": 3.6025,
+      "step": 3705
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.3410909090909094e-06,
+      "loss": 3.7858,
+      "step": 3706
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.3403636363636366e-06,
+      "loss": 3.6487,
+      "step": 3707
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.339636363636364e-06,
+      "loss": 3.841,
+      "step": 3708
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.338909090909091e-06,
+      "loss": 3.6389,
+      "step": 3709
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.3381818181818184e-06,
+      "loss": 3.569,
+      "step": 3710
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.3374545454545457e-06,
+      "loss": 3.5481,
+      "step": 3711
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.336727272727273e-06,
+      "loss": 3.7009,
+      "step": 3712
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.336e-06,
+      "loss": 3.5214,
+      "step": 3713
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.3352727272727274e-06,
+      "loss": 3.7604,
+      "step": 3714
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.3345454545454547e-06,
+      "loss": 3.2691,
+      "step": 3715
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.333818181818182e-06,
+      "loss": 3.5529,
+      "step": 3716
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.333090909090909e-06,
+      "loss": 3.5559,
+      "step": 3717
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.3323636363636364e-06,
+      "loss": 3.6507,
+      "step": 3718
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.3316363636363637e-06,
+      "loss": 3.6168,
+      "step": 3719
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.3309090909090914e-06,
+      "loss": 3.5637,
+      "step": 3720
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.3301818181818182e-06,
+      "loss": 3.7204,
+      "step": 3721
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.3294545454545455e-06,
+      "loss": 3.389,
+      "step": 3722
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.3287272727272727e-06,
+      "loss": 3.6457,
+      "step": 3723
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.3280000000000004e-06,
+      "loss": 3.4338,
+      "step": 3724
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.3272727272727277e-06,
+      "loss": 3.5129,
+      "step": 3725
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.3265454545454545e-06,
+      "loss": 3.6942,
+      "step": 3726
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.325818181818182e-06,
+      "loss": 3.665,
+      "step": 3727
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.3250909090909094e-06,
+      "loss": 3.723,
+      "step": 3728
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.3243636363636367e-06,
+      "loss": 3.4439,
+      "step": 3729
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.323636363636364e-06,
+      "loss": 3.4795,
+      "step": 3730
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.322909090909091e-06,
+      "loss": 3.5068,
+      "step": 3731
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.3221818181818185e-06,
+      "loss": 3.7707,
+      "step": 3732
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.3214545454545457e-06,
+      "loss": 3.5104,
+      "step": 3733
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.320727272727273e-06,
+      "loss": 3.615,
+      "step": 3734
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 3.4409,
+      "step": 3735
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.3192727272727275e-06,
+      "loss": 3.4918,
+      "step": 3736
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.3185454545454547e-06,
+      "loss": 3.4768,
+      "step": 3737
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.317818181818182e-06,
+      "loss": 3.2869,
+      "step": 3738
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.3170909090909093e-06,
+      "loss": 3.5202,
+      "step": 3739
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.3163636363636365e-06,
+      "loss": 3.4799,
+      "step": 3740
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.3156363636363638e-06,
+      "loss": 3.2619,
+      "step": 3741
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.314909090909091e-06,
+      "loss": 3.4289,
+      "step": 3742
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.3141818181818183e-06,
+      "loss": 3.5057,
+      "step": 3743
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.3134545454545455e-06,
+      "loss": 3.5604,
+      "step": 3744
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.312727272727273e-06,
+      "loss": 3.4291,
+      "step": 3745
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.312e-06,
+      "loss": 3.394,
+      "step": 3746
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.3112727272727277e-06,
+      "loss": 3.8389,
+      "step": 3747
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.3105454545454546e-06,
+      "loss": 3.4336,
+      "step": 3748
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.309818181818182e-06,
+      "loss": 3.5278,
+      "step": 3749
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.309090909090909e-06,
+      "loss": 3.384,
+      "step": 3750
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.3083636363636368e-06,
+      "loss": 3.6742,
+      "step": 3751
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.307636363636364e-06,
+      "loss": 3.5691,
+      "step": 3752
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.306909090909091e-06,
+      "loss": 3.5807,
+      "step": 3753
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.306181818181818e-06,
+      "loss": 3.5796,
+      "step": 3754
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.305454545454546e-06,
+      "loss": 3.4613,
+      "step": 3755
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.304727272727273e-06,
+      "loss": 3.5166,
+      "step": 3756
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.3040000000000003e-06,
+      "loss": 3.6177,
+      "step": 3757
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.303272727272727e-06,
+      "loss": 3.6409,
+      "step": 3758
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.302545454545455e-06,
+      "loss": 3.6238,
+      "step": 3759
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.301818181818182e-06,
+      "loss": 3.5049,
+      "step": 3760
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.3010909090909093e-06,
+      "loss": 3.5825,
+      "step": 3761
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.3003636363636366e-06,
+      "loss": 3.6481,
+      "step": 3762
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.299636363636364e-06,
+      "loss": 3.4489,
+      "step": 3763
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.298909090909091e-06,
+      "loss": 3.4636,
+      "step": 3764
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.2981818181818184e-06,
+      "loss": 3.497,
+      "step": 3765
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.2974545454545456e-06,
+      "loss": 3.5707,
+      "step": 3766
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.296727272727273e-06,
+      "loss": 3.725,
+      "step": 3767
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.296e-06,
+      "loss": 3.5917,
+      "step": 3768
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.2952727272727274e-06,
+      "loss": 3.438,
+      "step": 3769
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.294545454545455e-06,
+      "loss": 3.6161,
+      "step": 3770
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.293818181818182e-06,
+      "loss": 3.6013,
+      "step": 3771
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.293090909090909e-06,
+      "loss": 3.2843,
+      "step": 3772
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.2923636363636364e-06,
+      "loss": 3.7261,
+      "step": 3773
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.291636363636364e-06,
+      "loss": 3.6088,
+      "step": 3774
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.2909090909090913e-06,
+      "loss": 3.6113,
+      "step": 3775
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.290181818181818e-06,
+      "loss": 3.6444,
+      "step": 3776
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.2894545454545454e-06,
+      "loss": 3.4198,
+      "step": 3777
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.288727272727273e-06,
+      "loss": 3.3842,
+      "step": 3778
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.2880000000000004e-06,
+      "loss": 3.4326,
+      "step": 3779
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.2872727272727276e-06,
+      "loss": 3.5141,
+      "step": 3780
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.2865454545454545e-06,
+      "loss": 3.4316,
+      "step": 3781
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.285818181818182e-06,
+      "loss": 3.6167,
+      "step": 3782
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.2850909090909094e-06,
+      "loss": 3.4352,
+      "step": 3783
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.2843636363636367e-06,
+      "loss": 3.5743,
+      "step": 3784
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.283636363636364e-06,
+      "loss": 3.6691,
+      "step": 3785
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.282909090909091e-06,
+      "loss": 3.6247,
+      "step": 3786
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.2821818181818184e-06,
+      "loss": 3.2876,
+      "step": 3787
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.2814545454545457e-06,
+      "loss": 3.338,
+      "step": 3788
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.280727272727273e-06,
+      "loss": 3.742,
+      "step": 3789
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.28e-06,
+      "loss": 3.7688,
+      "step": 3790
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.2792727272727274e-06,
+      "loss": 3.4016,
+      "step": 3791
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.2785454545454547e-06,
+      "loss": 3.495,
+      "step": 3792
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.277818181818182e-06,
+      "loss": 3.4801,
+      "step": 3793
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.2770909090909092e-06,
+      "loss": 3.3322,
+      "step": 3794
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.2763636363636365e-06,
+      "loss": 3.6626,
+      "step": 3795
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.2756363636363637e-06,
+      "loss": 3.4945,
+      "step": 3796
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.2749090909090914e-06,
+      "loss": 3.5415,
+      "step": 3797
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.2741818181818182e-06,
+      "loss": 3.6837,
+      "step": 3798
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.2734545454545455e-06,
+      "loss": 3.5086,
+      "step": 3799
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.2727272727272728e-06,
+      "loss": 3.3399,
+      "step": 3800
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.2720000000000004e-06,
+      "loss": 3.569,
+      "step": 3801
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.2712727272727277e-06,
+      "loss": 3.3834,
+      "step": 3802
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.2705454545454545e-06,
+      "loss": 3.3471,
+      "step": 3803
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.2698181818181818e-06,
+      "loss": 3.6581,
+      "step": 3804
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.2690909090909095e-06,
+      "loss": 3.6361,
+      "step": 3805
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.2683636363636367e-06,
+      "loss": 3.3555,
+      "step": 3806
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.267636363636364e-06,
+      "loss": 3.4082,
+      "step": 3807
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.266909090909091e-06,
+      "loss": 3.7094,
+      "step": 3808
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.2661818181818185e-06,
+      "loss": 3.5591,
+      "step": 3809
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.2654545454545457e-06,
+      "loss": 3.5571,
+      "step": 3810
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.264727272727273e-06,
+      "loss": 3.4779,
+      "step": 3811
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.2640000000000003e-06,
+      "loss": 3.4056,
+      "step": 3812
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.2632727272727275e-06,
+      "loss": 3.7153,
+      "step": 3813
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.2625454545454548e-06,
+      "loss": 3.4424,
+      "step": 3814
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.261818181818182e-06,
+      "loss": 3.5836,
+      "step": 3815
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.2610909090909093e-06,
+      "loss": 3.6321,
+      "step": 3816
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.2603636363636365e-06,
+      "loss": 3.5355,
+      "step": 3817
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.259636363636364e-06,
+      "loss": 3.5902,
+      "step": 3818
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.258909090909091e-06,
+      "loss": 3.4602,
+      "step": 3819
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.2581818181818183e-06,
+      "loss": 3.6363,
+      "step": 3820
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.2574545454545456e-06,
+      "loss": 3.8258,
+      "step": 3821
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.256727272727273e-06,
+      "loss": 3.8474,
+      "step": 3822
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.256e-06,
+      "loss": 3.6202,
+      "step": 3823
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.2552727272727273e-06,
+      "loss": 3.7074,
+      "step": 3824
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.254545454545455e-06,
+      "loss": 3.5342,
+      "step": 3825
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.253818181818182e-06,
+      "loss": 3.6472,
+      "step": 3826
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.253090909090909e-06,
+      "loss": 3.3567,
+      "step": 3827
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.2523636363636364e-06,
+      "loss": 3.7927,
+      "step": 3828
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.251636363636364e-06,
+      "loss": 3.5454,
+      "step": 3829
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.2509090909090913e-06,
+      "loss": 3.6799,
+      "step": 3830
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.250181818181818e-06,
+      "loss": 3.5253,
+      "step": 3831
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.2494545454545454e-06,
+      "loss": 3.4675,
+      "step": 3832
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.248727272727273e-06,
+      "loss": 3.3743,
+      "step": 3833
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.2480000000000003e-06,
+      "loss": 3.4607,
+      "step": 3834
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.2472727272727276e-06,
+      "loss": 3.3449,
+      "step": 3835
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.246545454545455e-06,
+      "loss": 3.5261,
+      "step": 3836
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.245818181818182e-06,
+      "loss": 3.4861,
+      "step": 3837
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.2450909090909094e-06,
+      "loss": 3.6686,
+      "step": 3838
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.2443636363636366e-06,
+      "loss": 3.5675,
+      "step": 3839
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.243636363636364e-06,
+      "loss": 3.2786,
+      "step": 3840
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.242909090909091e-06,
+      "loss": 3.7381,
+      "step": 3841
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.2421818181818184e-06,
+      "loss": 3.1995,
+      "step": 3842
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.2414545454545456e-06,
+      "loss": 3.5874,
+      "step": 3843
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.240727272727273e-06,
+      "loss": 3.5689,
+      "step": 3844
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.24e-06,
+      "loss": 3.3137,
+      "step": 3845
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.2392727272727274e-06,
+      "loss": 3.5597,
+      "step": 3846
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.2385454545454547e-06,
+      "loss": 3.533,
+      "step": 3847
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.237818181818182e-06,
+      "loss": 3.4179,
+      "step": 3848
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.237090909090909e-06,
+      "loss": 3.5644,
+      "step": 3849
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.2363636363636364e-06,
+      "loss": 3.4879,
+      "step": 3850
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.2356363636363637e-06,
+      "loss": 3.7321,
+      "step": 3851
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.2349090909090914e-06,
+      "loss": 3.652,
+      "step": 3852
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.234181818181818e-06,
+      "loss": 3.5711,
+      "step": 3853
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.2334545454545455e-06,
+      "loss": 3.51,
+      "step": 3854
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.2327272727272727e-06,
+      "loss": 3.4268,
+      "step": 3855
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.2320000000000004e-06,
+      "loss": 3.6509,
+      "step": 3856
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.2312727272727277e-06,
+      "loss": 3.4161,
+      "step": 3857
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.2305454545454545e-06,
+      "loss": 3.6052,
+      "step": 3858
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.2298181818181817e-06,
+      "loss": 3.5685,
+      "step": 3859
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.2290909090909094e-06,
+      "loss": 3.4469,
+      "step": 3860
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.2283636363636367e-06,
+      "loss": 3.4751,
+      "step": 3861
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.227636363636364e-06,
+      "loss": 3.6163,
+      "step": 3862
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.2269090909090908e-06,
+      "loss": 3.7556,
+      "step": 3863
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.2261818181818185e-06,
+      "loss": 3.6512,
+      "step": 3864
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.2254545454545457e-06,
+      "loss": 3.5409,
+      "step": 3865
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.224727272727273e-06,
+      "loss": 3.4954,
+      "step": 3866
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.2240000000000002e-06,
+      "loss": 3.4873,
+      "step": 3867
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.2232727272727275e-06,
+      "loss": 3.6087,
+      "step": 3868
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.2225454545454547e-06,
+      "loss": 3.7178,
+      "step": 3869
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.221818181818182e-06,
+      "loss": 3.4974,
+      "step": 3870
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.2210909090909092e-06,
+      "loss": 3.7112,
+      "step": 3871
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.2203636363636365e-06,
+      "loss": 3.4153,
+      "step": 3872
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.2196363636363638e-06,
+      "loss": 3.5948,
+      "step": 3873
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.218909090909091e-06,
+      "loss": 3.6322,
+      "step": 3874
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.2181818181818187e-06,
+      "loss": 3.3861,
+      "step": 3875
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.2174545454545455e-06,
+      "loss": 3.3261,
+      "step": 3876
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.2167272727272728e-06,
+      "loss": 3.5901,
+      "step": 3877
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.216e-06,
+      "loss": 3.5208,
+      "step": 3878
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.16756756756756758,
+      "eval_loss": 3.509823799133301,
+      "eval_runtime": 142.9008,
+      "eval_samples_per_second": 2.589,
+      "eval_steps_per_second": 0.651,
+      "step": 3878
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.2152727272727277e-06,
+      "loss": 3.6482,
+      "step": 3879
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.214545454545455e-06,
+      "loss": 3.4714,
+      "step": 3880
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.213818181818182e-06,
+      "loss": 3.5427,
+      "step": 3881
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.213090909090909e-06,
+      "loss": 3.248,
+      "step": 3882
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.2123636363636368e-06,
+      "loss": 3.4,
+      "step": 3883
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.211636363636364e-06,
+      "loss": 3.4719,
+      "step": 3884
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.2109090909090913e-06,
+      "loss": 3.6758,
+      "step": 3885
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.210181818181818e-06,
+      "loss": 3.3455,
+      "step": 3886
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.2094545454545458e-06,
+      "loss": 3.5456,
+      "step": 3887
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 2.208727272727273e-06,
+      "loss": 3.5532,
+      "step": 3888
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 2.2080000000000003e-06,
+      "loss": 3.3377,
+      "step": 3889
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 2.2072727272727275e-06,
+      "loss": 3.4518,
+      "step": 3890
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 2.206545454545455e-06,
+      "loss": 3.5374,
+      "step": 3891
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 2.205818181818182e-06,
+      "loss": 3.3694,
+      "step": 3892
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 2.2050909090909093e-06,
+      "loss": 3.6207,
+      "step": 3893
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.2043636363636366e-06,
+      "loss": 3.5135,
+      "step": 3894
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.203636363636364e-06,
+      "loss": 3.3489,
+      "step": 3895
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.202909090909091e-06,
+      "loss": 3.4264,
+      "step": 3896
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 2.2021818181818183e-06,
+      "loss": 3.6361,
+      "step": 3897
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 2.2014545454545456e-06,
+      "loss": 3.4959,
+      "step": 3898
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 2.200727272727273e-06,
+      "loss": 3.228,
+      "step": 3899
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 2.2e-06,
+      "loss": 3.5391,
+      "step": 3900
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 2.1992727272727274e-06,
+      "loss": 3.6895,
+      "step": 3901
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 2.1985454545454546e-06,
+      "loss": 3.5811,
+      "step": 3902
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 2.197818181818182e-06,
+      "loss": 3.4719,
+      "step": 3903
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 2.197090909090909e-06,
+      "loss": 3.5023,
+      "step": 3904
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 2.1963636363636364e-06,
+      "loss": 3.4828,
+      "step": 3905
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 2.195636363636364e-06,
+      "loss": 3.6386,
+      "step": 3906
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 2.1949090909090913e-06,
+      "loss": 3.4843,
+      "step": 3907
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 2.194181818181818e-06,
+      "loss": 3.5901,
+      "step": 3908
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 2.1934545454545454e-06,
+      "loss": 3.4771,
+      "step": 3909
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 2.192727272727273e-06,
+      "loss": 3.6714,
+      "step": 3910
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 2.1920000000000004e-06,
+      "loss": 3.4576,
+      "step": 3911
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 2.1912727272727276e-06,
+      "loss": 3.485,
+      "step": 3912
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 2.1905454545454544e-06,
+      "loss": 3.9072,
+      "step": 3913
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 2.189818181818182e-06,
+      "loss": 3.6706,
+      "step": 3914
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 2.1890909090909094e-06,
+      "loss": 3.6995,
+      "step": 3915
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 2.1883636363636366e-06,
+      "loss": 3.3124,
+      "step": 3916
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 2.187636363636364e-06,
+      "loss": 3.6878,
+      "step": 3917
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 2.186909090909091e-06,
+      "loss": 3.5023,
+      "step": 3918
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 2.1861818181818184e-06,
+      "loss": 3.4635,
+      "step": 3919
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 2.1854545454545457e-06,
+      "loss": 3.4287,
+      "step": 3920
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 2.184727272727273e-06,
+      "loss": 3.4349,
+      "step": 3921
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 2.184e-06,
+      "loss": 3.6606,
+      "step": 3922
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 2.1832727272727274e-06,
+      "loss": 3.4362,
+      "step": 3923
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 2.1825454545454547e-06,
+      "loss": 3.5175,
+      "step": 3924
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 2.181818181818182e-06,
+      "loss": 3.4725,
+      "step": 3925
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 2.181090909090909e-06,
+      "loss": 3.4287,
+      "step": 3926
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 2.1803636363636365e-06,
+      "loss": 3.816,
+      "step": 3927
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 2.1796363636363637e-06,
+      "loss": 3.6861,
+      "step": 3928
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 2.178909090909091e-06,
+      "loss": 3.4797,
+      "step": 3929
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 2.1781818181818187e-06,
+      "loss": 3.605,
+      "step": 3930
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 2.1774545454545455e-06,
+      "loss": 3.5497,
+      "step": 3931
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 2.1767272727272727e-06,
+      "loss": 3.3552,
+      "step": 3932
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 2.176e-06,
+      "loss": 3.5411,
+      "step": 3933
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 2.1752727272727277e-06,
+      "loss": 3.6133,
+      "step": 3934
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 2.174545454545455e-06,
+      "loss": 3.8594,
+      "step": 3935
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 2.1738181818181818e-06,
+      "loss": 3.3952,
+      "step": 3936
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 2.173090909090909e-06,
+      "loss": 3.344,
+      "step": 3937
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 2.1723636363636367e-06,
+      "loss": 3.4587,
+      "step": 3938
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 2.171636363636364e-06,
+      "loss": 3.8427,
+      "step": 3939
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 2.1709090909090912e-06,
+      "loss": 3.4193,
+      "step": 3940
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 2.170181818181818e-06,
+      "loss": 3.6359,
+      "step": 3941
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 2.1694545454545457e-06,
+      "loss": 3.2001,
+      "step": 3942
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 2.168727272727273e-06,
+      "loss": 3.1516,
+      "step": 3943
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 2.1680000000000002e-06,
+      "loss": 3.5147,
+      "step": 3944
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 2.1672727272727275e-06,
+      "loss": 3.511,
+      "step": 3945
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 2.1665454545454548e-06,
+      "loss": 3.2995,
+      "step": 3946
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 2.165818181818182e-06,
+      "loss": 3.24,
+      "step": 3947
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 2.1650909090909093e-06,
+      "loss": 3.4686,
+      "step": 3948
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 2.1643636363636365e-06,
+      "loss": 3.4742,
+      "step": 3949
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 2.163636363636364e-06,
+      "loss": 3.4811,
+      "step": 3950
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 2.162909090909091e-06,
+      "loss": 3.7364,
+      "step": 3951
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 2.1621818181818183e-06,
+      "loss": 3.4702,
+      "step": 3952
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 2.1614545454545456e-06,
+      "loss": 3.6719,
+      "step": 3953
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 2.160727272727273e-06,
+      "loss": 3.5639,
+      "step": 3954
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 2.16e-06,
+      "loss": 3.5202,
+      "step": 3955
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 2.1592727272727273e-06,
+      "loss": 3.6903,
+      "step": 3956
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 2.158545454545455e-06,
+      "loss": 3.5263,
+      "step": 3957
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 2.157818181818182e-06,
+      "loss": 3.2945,
+      "step": 3958
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 2.157090909090909e-06,
+      "loss": 3.4774,
+      "step": 3959
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 2.1563636363636364e-06,
+      "loss": 3.37,
+      "step": 3960
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 2.155636363636364e-06,
+      "loss": 3.7535,
+      "step": 3961
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 2.1549090909090913e-06,
+      "loss": 3.6725,
+      "step": 3962
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 2.154181818181818e-06,
+      "loss": 3.2979,
+      "step": 3963
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 2.1534545454545454e-06,
+      "loss": 3.596,
+      "step": 3964
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 2.152727272727273e-06,
+      "loss": 3.8088,
+      "step": 3965
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 2.1520000000000003e-06,
+      "loss": 3.4364,
+      "step": 3966
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 2.1512727272727276e-06,
+      "loss": 3.3862,
+      "step": 3967
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 2.1505454545454544e-06,
+      "loss": 3.3611,
+      "step": 3968
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 2.149818181818182e-06,
+      "loss": 3.7522,
+      "step": 3969
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 2.1490909090909093e-06,
+      "loss": 3.4584,
+      "step": 3970
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 2.1483636363636366e-06,
+      "loss": 3.305,
+      "step": 3971
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 2.147636363636364e-06,
+      "loss": 3.7877,
+      "step": 3972
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 2.146909090909091e-06,
+      "loss": 3.3789,
+      "step": 3973
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.1461818181818184e-06,
+      "loss": 3.6213,
+      "step": 3974
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.1454545454545456e-06,
+      "loss": 3.4393,
+      "step": 3975
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.144727272727273e-06,
+      "loss": 3.5677,
+      "step": 3976
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.144e-06,
+      "loss": 3.7034,
+      "step": 3977
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.1432727272727274e-06,
+      "loss": 3.5295,
+      "step": 3978
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.1425454545454547e-06,
+      "loss": 3.3739,
+      "step": 3979
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.1418181818181823e-06,
+      "loss": 3.5655,
+      "step": 3980
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.141090909090909e-06,
+      "loss": 3.8742,
+      "step": 3981
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.1403636363636364e-06,
+      "loss": 3.4156,
+      "step": 3982
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.1396363636363637e-06,
+      "loss": 3.5785,
+      "step": 3983
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.1389090909090914e-06,
+      "loss": 3.4082,
+      "step": 3984
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.1381818181818186e-06,
+      "loss": 3.4996,
+      "step": 3985
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.1374545454545454e-06,
+      "loss": 3.365,
+      "step": 3986
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.1367272727272727e-06,
+      "loss": 3.4884,
+      "step": 3987
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.1360000000000004e-06,
+      "loss": 3.8333,
+      "step": 3988
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.1352727272727276e-06,
+      "loss": 3.4966,
+      "step": 3989
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.134545454545455e-06,
+      "loss": 3.7075,
+      "step": 3990
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.1338181818181817e-06,
+      "loss": 3.4321,
+      "step": 3991
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.1330909090909094e-06,
+      "loss": 3.5982,
+      "step": 3992
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.1323636363636367e-06,
+      "loss": 3.4399,
+      "step": 3993
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.131636363636364e-06,
+      "loss": 3.5544,
+      "step": 3994
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.130909090909091e-06,
+      "loss": 3.7134,
+      "step": 3995
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.1301818181818184e-06,
+      "loss": 3.3592,
+      "step": 3996
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.1294545454545457e-06,
+      "loss": 3.4734,
+      "step": 3997
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.128727272727273e-06,
+      "loss": 3.7196,
+      "step": 3998
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.128e-06,
+      "loss": 3.8226,
+      "step": 3999
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.1272727272727275e-06,
+      "loss": 3.6262,
+      "step": 4000
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.1265454545454547e-06,
+      "loss": 3.4462,
+      "step": 4001
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.125818181818182e-06,
+      "loss": 3.2793,
+      "step": 4002
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.1250909090909092e-06,
+      "loss": 3.435,
+      "step": 4003
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.1243636363636365e-06,
+      "loss": 3.3804,
+      "step": 4004
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.1236363636363637e-06,
+      "loss": 3.6374,
+      "step": 4005
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.122909090909091e-06,
+      "loss": 3.4228,
+      "step": 4006
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.1221818181818183e-06,
+      "loss": 3.3069,
+      "step": 4007
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.1214545454545455e-06,
+      "loss": 3.2657,
+      "step": 4008
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.1207272727272728e-06,
+      "loss": 3.3323,
+      "step": 4009
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.12e-06,
+      "loss": 3.3373,
+      "step": 4010
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.1192727272727273e-06,
+      "loss": 3.483,
+      "step": 4011
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.118545454545455e-06,
+      "loss": 3.4908,
+      "step": 4012
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.117818181818182e-06,
+      "loss": 3.7765,
+      "step": 4013
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.117090909090909e-06,
+      "loss": 3.5758,
+      "step": 4014
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.1163636363636367e-06,
+      "loss": 3.6486,
+      "step": 4015
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.115636363636364e-06,
+      "loss": 3.5581,
+      "step": 4016
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.1149090909090913e-06,
+      "loss": 3.4893,
+      "step": 4017
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.114181818181818e-06,
+      "loss": 3.7982,
+      "step": 4018
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.1134545454545458e-06,
+      "loss": 3.4915,
+      "step": 4019
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.112727272727273e-06,
+      "loss": 3.7879,
+      "step": 4020
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.1120000000000003e-06,
+      "loss": 3.3896,
+      "step": 4021
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.1112727272727275e-06,
+      "loss": 3.6963,
+      "step": 4022
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.110545454545455e-06,
+      "loss": 3.4972,
+      "step": 4023
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.109818181818182e-06,
+      "loss": 3.5618,
+      "step": 4024
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.1090909090909093e-06,
+      "loss": 3.3876,
+      "step": 4025
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.1083636363636366e-06,
+      "loss": 3.4952,
+      "step": 4026
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.107636363636364e-06,
+      "loss": 3.4144,
+      "step": 4027
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.106909090909091e-06,
+      "loss": 3.3634,
+      "step": 4028
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.1061818181818183e-06,
+      "loss": 3.4178,
+      "step": 4029
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.1054545454545456e-06,
+      "loss": 3.7256,
+      "step": 4030
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.104727272727273e-06,
+      "loss": 3.2803,
+      "step": 4031
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.104e-06,
+      "loss": 3.4344,
+      "step": 4032
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.1032727272727274e-06,
+      "loss": 3.4411,
+      "step": 4033
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.1025454545454546e-06,
+      "loss": 3.4573,
+      "step": 4034
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.1018181818181823e-06,
+      "loss": 3.5336,
+      "step": 4035
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.101090909090909e-06,
+      "loss": 3.6785,
+      "step": 4036
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.1003636363636364e-06,
+      "loss": 3.3908,
+      "step": 4037
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.0996363636363636e-06,
+      "loss": 3.4047,
+      "step": 4038
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.0989090909090913e-06,
+      "loss": 3.1786,
+      "step": 4039
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.0981818181818186e-06,
+      "loss": 3.7858,
+      "step": 4040
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.0974545454545454e-06,
+      "loss": 3.6932,
+      "step": 4041
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.0967272727272727e-06,
+      "loss": 3.6314,
+      "step": 4042
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.0960000000000003e-06,
+      "loss": 3.3444,
+      "step": 4043
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.0952727272727276e-06,
+      "loss": 3.3839,
+      "step": 4044
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.094545454545455e-06,
+      "loss": 3.3094,
+      "step": 4045
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.0938181818181817e-06,
+      "loss": 3.4565,
+      "step": 4046
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.0930909090909094e-06,
+      "loss": 3.6349,
+      "step": 4047
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.0923636363636366e-06,
+      "loss": 3.3917,
+      "step": 4048
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.091636363636364e-06,
+      "loss": 3.4002,
+      "step": 4049
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.090909090909091e-06,
+      "loss": 3.5775,
+      "step": 4050
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.0901818181818184e-06,
+      "loss": 3.3774,
+      "step": 4051
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.0894545454545457e-06,
+      "loss": 3.3741,
+      "step": 4052
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.088727272727273e-06,
+      "loss": 3.5058,
+      "step": 4053
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.088e-06,
+      "loss": 3.5291,
+      "step": 4054
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.0872727272727274e-06,
+      "loss": 3.4527,
+      "step": 4055
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.0865454545454547e-06,
+      "loss": 3.408,
+      "step": 4056
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.085818181818182e-06,
+      "loss": 3.3972,
+      "step": 4057
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.085090909090909e-06,
+      "loss": 3.5682,
+      "step": 4058
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.0843636363636365e-06,
+      "loss": 3.4991,
+      "step": 4059
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.0836363636363637e-06,
+      "loss": 3.5398,
+      "step": 4060
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.082909090909091e-06,
+      "loss": 3.6531,
+      "step": 4061
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.0821818181818186e-06,
+      "loss": 3.5465,
+      "step": 4062
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.0814545454545455e-06,
+      "loss": 3.4608,
+      "step": 4063
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.0807272727272727e-06,
+      "loss": 3.5692,
+      "step": 4064
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.08e-06,
+      "loss": 3.6722,
+      "step": 4065
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.0792727272727277e-06,
+      "loss": 3.3236,
+      "step": 4066
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.078545454545455e-06,
+      "loss": 3.4187,
+      "step": 4067
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.0778181818181818e-06,
+      "loss": 3.3652,
+      "step": 4068
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.077090909090909e-06,
+      "loss": 3.7526,
+      "step": 4069
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.0763636363636367e-06,
+      "loss": 3.6086,
+      "step": 4070
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.075636363636364e-06,
+      "loss": 3.4094,
+      "step": 4071
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.074909090909091e-06,
+      "loss": 3.7193,
+      "step": 4072
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.074181818181818e-06,
+      "loss": 3.5294,
+      "step": 4073
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.0734545454545457e-06,
+      "loss": 3.4183,
+      "step": 4074
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.072727272727273e-06,
+      "loss": 3.5603,
+      "step": 4075
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.0720000000000002e-06,
+      "loss": 3.4572,
+      "step": 4076
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.0712727272727275e-06,
+      "loss": 3.4608,
+      "step": 4077
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.0705454545454547e-06,
+      "loss": 3.5272,
+      "step": 4078
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.069818181818182e-06,
+      "loss": 3.1624,
+      "step": 4079
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.0690909090909093e-06,
+      "loss": 3.3043,
+      "step": 4080
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.0683636363636365e-06,
+      "loss": 3.5885,
+      "step": 4081
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.0676363636363638e-06,
+      "loss": 3.5967,
+      "step": 4082
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.066909090909091e-06,
+      "loss": 3.4542,
+      "step": 4083
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.0661818181818183e-06,
+      "loss": 3.6915,
+      "step": 4084
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.0654545454545455e-06,
+      "loss": 3.5285,
+      "step": 4085
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.064727272727273e-06,
+      "loss": 3.5087,
+      "step": 4086
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.064e-06,
+      "loss": 3.763,
+      "step": 4087
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.0632727272727273e-06,
+      "loss": 3.5179,
+      "step": 4088
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.062545454545455e-06,
+      "loss": 3.2663,
+      "step": 4089
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.0618181818181823e-06,
+      "loss": 3.8395,
+      "step": 4090
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.061090909090909e-06,
+      "loss": 3.5498,
+      "step": 4091
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.0603636363636363e-06,
+      "loss": 3.6,
+      "step": 4092
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.059636363636364e-06,
+      "loss": 3.4345,
+      "step": 4093
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.0589090909090913e-06,
+      "loss": 3.8123,
+      "step": 4094
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.0581818181818185e-06,
+      "loss": 3.4588,
+      "step": 4095
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.0574545454545454e-06,
+      "loss": 3.6726,
+      "step": 4096
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.056727272727273e-06,
+      "loss": 3.7945,
+      "step": 4097
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.0560000000000003e-06,
+      "loss": 3.2811,
+      "step": 4098
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.0552727272727276e-06,
+      "loss": 3.6891,
+      "step": 4099
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.054545454545455e-06,
+      "loss": 3.6435,
+      "step": 4100
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.053818181818182e-06,
+      "loss": 3.5298,
+      "step": 4101
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.0530909090909093e-06,
+      "loss": 3.5228,
+      "step": 4102
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.0523636363636366e-06,
+      "loss": 3.3794,
+      "step": 4103
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.051636363636364e-06,
+      "loss": 3.4463,
+      "step": 4104
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.050909090909091e-06,
+      "loss": 3.3875,
+      "step": 4105
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.0501818181818184e-06,
+      "loss": 3.8656,
+      "step": 4106
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.0494545454545456e-06,
+      "loss": 3.7298,
+      "step": 4107
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.048727272727273e-06,
+      "loss": 3.6582,
+      "step": 4108
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.048e-06,
+      "loss": 3.6378,
+      "step": 4109
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.0472727272727274e-06,
+      "loss": 3.4094,
+      "step": 4110
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.0465454545454546e-06,
+      "loss": 3.4361,
+      "step": 4111
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.045818181818182e-06,
+      "loss": 3.8303,
+      "step": 4112
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.045090909090909e-06,
+      "loss": 3.6222,
+      "step": 4113
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.0443636363636364e-06,
+      "loss": 3.55,
+      "step": 4114
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.0436363636363637e-06,
+      "loss": 3.4918,
+      "step": 4115
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.042909090909091e-06,
+      "loss": 3.6374,
+      "step": 4116
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.0421818181818186e-06,
+      "loss": 3.3872,
+      "step": 4117
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.0414545454545454e-06,
+      "loss": 3.5328,
+      "step": 4118
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.0407272727272727e-06,
+      "loss": 3.493,
+      "step": 4119
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.04e-06,
+      "loss": 3.5566,
+      "step": 4120
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.0392727272727276e-06,
+      "loss": 3.4096,
+      "step": 4121
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.038545454545455e-06,
+      "loss": 3.3845,
+      "step": 4122
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.0378181818181817e-06,
+      "loss": 3.3429,
+      "step": 4123
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.037090909090909e-06,
+      "loss": 3.3983,
+      "step": 4124
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.0363636363636367e-06,
+      "loss": 3.5926,
+      "step": 4125
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.035636363636364e-06,
+      "loss": 3.341,
+      "step": 4126
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.034909090909091e-06,
+      "loss": 3.5056,
+      "step": 4127
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.0341818181818184e-06,
+      "loss": 3.7695,
+      "step": 4128
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.0334545454545457e-06,
+      "loss": 3.3257,
+      "step": 4129
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.032727272727273e-06,
+      "loss": 3.4237,
+      "step": 4130
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.032e-06,
+      "loss": 3.7115,
+      "step": 4131
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.0312727272727275e-06,
+      "loss": 3.7052,
+      "step": 4132
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.0305454545454547e-06,
+      "loss": 3.5929,
+      "step": 4133
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.029818181818182e-06,
+      "loss": 3.7014,
+      "step": 4134
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.0290909090909092e-06,
+      "loss": 3.6071,
+      "step": 4135
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.0283636363636365e-06,
+      "loss": 3.4245,
+      "step": 4136
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.0276363636363637e-06,
+      "loss": 3.3678,
+      "step": 4137
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.026909090909091e-06,
+      "loss": 3.4006,
+      "step": 4138
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.0261818181818182e-06,
+      "loss": 3.5886,
+      "step": 4139
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.025454545454546e-06,
+      "loss": 3.3686,
+      "step": 4140
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.0247272727272728e-06,
+      "loss": 3.6673,
+      "step": 4141
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.024e-06,
+      "loss": 3.2532,
+      "step": 4142
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.0232727272727273e-06,
+      "loss": 3.2796,
+      "step": 4143
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.022545454545455e-06,
+      "loss": 3.676,
+      "step": 4144
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.0218181818181822e-06,
+      "loss": 3.4412,
+      "step": 4145
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.021090909090909e-06,
+      "loss": 3.3826,
+      "step": 4146
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.0203636363636363e-06,
+      "loss": 3.3216,
+      "step": 4147
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.019636363636364e-06,
+      "loss": 3.6824,
+      "step": 4148
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.0189090909090912e-06,
+      "loss": 3.524,
+      "step": 4149
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.0181818181818185e-06,
+      "loss": 3.6506,
+      "step": 4150
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.0174545454545453e-06,
+      "loss": 3.2661,
+      "step": 4151
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.016727272727273e-06,
+      "loss": 3.5833,
+      "step": 4152
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.0160000000000003e-06,
+      "loss": 3.5635,
+      "step": 4153
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.0152727272727275e-06,
+      "loss": 3.493,
+      "step": 4154
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.0145454545454548e-06,
+      "loss": 3.7241,
+      "step": 4155
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.1864864864864865,
+      "eval_loss": 3.4850962162017822,
+      "eval_runtime": 144.2188,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 0.645,
+      "step": 4155
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.013818181818182e-06,
+      "loss": 3.615,
+      "step": 4156
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.0130909090909093e-06,
+      "loss": 3.4291,
+      "step": 4157
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.0123636363636365e-06,
+      "loss": 3.8167,
+      "step": 4158
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.011636363636364e-06,
+      "loss": 3.3099,
+      "step": 4159
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.010909090909091e-06,
+      "loss": 3.5677,
+      "step": 4160
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.0101818181818183e-06,
+      "loss": 3.5543,
+      "step": 4161
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.0094545454545456e-06,
+      "loss": 3.5069,
+      "step": 4162
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.008727272727273e-06,
+      "loss": 3.3409,
+      "step": 4163
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.008e-06,
+      "loss": 3.2968,
+      "step": 4164
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.0072727272727273e-06,
+      "loss": 3.6536,
+      "step": 4165
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.0065454545454546e-06,
+      "loss": 3.5459,
+      "step": 4166
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.0058181818181823e-06,
+      "loss": 3.6289,
+      "step": 4167
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.005090909090909e-06,
+      "loss": 3.3879,
+      "step": 4168
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.0043636363636364e-06,
+      "loss": 3.3274,
+      "step": 4169
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.0036363636363636e-06,
+      "loss": 3.7103,
+      "step": 4170
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.0029090909090913e-06,
+      "loss": 3.5339,
+      "step": 4171
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.0021818181818186e-06,
+      "loss": 3.6833,
+      "step": 4172
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.0014545454545454e-06,
+      "loss": 3.2008,
+      "step": 4173
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.0007272727272727e-06,
+      "loss": 3.4301,
+      "step": 4174
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.3973,
+      "step": 4175
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 1.9992727272727276e-06,
+      "loss": 3.5628,
+      "step": 4176
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 1.998545454545455e-06,
+      "loss": 3.4043,
+      "step": 4177
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 1.9978181818181817e-06,
+      "loss": 3.4686,
+      "step": 4178
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.9970909090909094e-06,
+      "loss": 3.6334,
+      "step": 4179
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.9963636363636366e-06,
+      "loss": 3.8241,
+      "step": 4180
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.995636363636364e-06,
+      "loss": 3.6696,
+      "step": 4181
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.994909090909091e-06,
+      "loss": 3.3008,
+      "step": 4182
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.9941818181818184e-06,
+      "loss": 3.1495,
+      "step": 4183
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.9934545454545456e-06,
+      "loss": 3.6179,
+      "step": 4184
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.992727272727273e-06,
+      "loss": 3.6927,
+      "step": 4185
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.992e-06,
+      "loss": 3.4638,
+      "step": 4186
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.9912727272727274e-06,
+      "loss": 3.3144,
+      "step": 4187
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.9905454545454547e-06,
+      "loss": 3.5682,
+      "step": 4188
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.989818181818182e-06,
+      "loss": 3.4964,
+      "step": 4189
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.989090909090909e-06,
+      "loss": 3.3374,
+      "step": 4190
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.9883636363636364e-06,
+      "loss": 3.4183,
+      "step": 4191
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.9876363636363637e-06,
+      "loss": 3.6195,
+      "step": 4192
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.986909090909091e-06,
+      "loss": 3.4247,
+      "step": 4193
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.986181818181818e-06,
+      "loss": 3.5458,
+      "step": 4194
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.985454545454546e-06,
+      "loss": 3.3845,
+      "step": 4195
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.9847272727272727e-06,
+      "loss": 3.6161,
+      "step": 4196
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.984e-06,
+      "loss": 3.7937,
+      "step": 4197
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.9832727272727277e-06,
+      "loss": 3.4572,
+      "step": 4198
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.982545454545455e-06,
+      "loss": 3.4755,
+      "step": 4199
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.981818181818182e-06,
+      "loss": 3.39,
+      "step": 4200
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.981090909090909e-06,
+      "loss": 3.6209,
+      "step": 4201
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.9803636363636367e-06,
+      "loss": 3.4429,
+      "step": 4202
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.979636363636364e-06,
+      "loss": 3.9566,
+      "step": 4203
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.978909090909091e-06,
+      "loss": 3.5628,
+      "step": 4204
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.9781818181818185e-06,
+      "loss": 3.4099,
+      "step": 4205
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.9774545454545457e-06,
+      "loss": 3.2079,
+      "step": 4206
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.976727272727273e-06,
+      "loss": 3.2569,
+      "step": 4207
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.9760000000000002e-06,
+      "loss": 3.228,
+      "step": 4208
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.9752727272727275e-06,
+      "loss": 3.4035,
+      "step": 4209
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.9745454545454547e-06,
+      "loss": 3.39,
+      "step": 4210
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.973818181818182e-06,
+      "loss": 3.5811,
+      "step": 4211
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.9730909090909093e-06,
+      "loss": 3.6909,
+      "step": 4212
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.9723636363636365e-06,
+      "loss": 3.7249,
+      "step": 4213
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.9716363636363638e-06,
+      "loss": 3.6491,
+      "step": 4214
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.970909090909091e-06,
+      "loss": 3.396,
+      "step": 4215
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.9701818181818183e-06,
+      "loss": 3.4065,
+      "step": 4216
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.9694545454545455e-06,
+      "loss": 3.5167,
+      "step": 4217
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.968727272727273e-06,
+      "loss": 3.7413,
+      "step": 4218
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.968e-06,
+      "loss": 3.8614,
+      "step": 4219
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.9672727272727273e-06,
+      "loss": 3.5068,
+      "step": 4220
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.9665454545454546e-06,
+      "loss": 3.6302,
+      "step": 4221
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.9658181818181822e-06,
+      "loss": 3.3927,
+      "step": 4222
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 1.965090909090909e-06,
+      "loss": 3.5056,
+      "step": 4223
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 1.9643636363636363e-06,
+      "loss": 3.3994,
+      "step": 4224
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 1.9636363636363636e-06,
+      "loss": 3.7143,
+      "step": 4225
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.9629090909090913e-06,
+      "loss": 3.4428,
+      "step": 4226
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.9621818181818185e-06,
+      "loss": 3.611,
+      "step": 4227
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.9614545454545454e-06,
+      "loss": 3.5656,
+      "step": 4228
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.9607272727272726e-06,
+      "loss": 3.2432,
+      "step": 4229
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 3.8496,
+      "step": 4230
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.9592727272727275e-06,
+      "loss": 3.2534,
+      "step": 4231
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 1.958545454545455e-06,
+      "loss": 3.5067,
+      "step": 4232
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 1.9578181818181816e-06,
+      "loss": 3.4539,
+      "step": 4233
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.9570909090909093e-06,
+      "loss": 3.6152,
+      "step": 4234
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.9563636363636366e-06,
+      "loss": 3.5002,
+      "step": 4235
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.955636363636364e-06,
+      "loss": 3.4968,
+      "step": 4236
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.954909090909091e-06,
+      "loss": 3.4132,
+      "step": 4237
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.9541818181818183e-06,
+      "loss": 3.44,
+      "step": 4238
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.9534545454545456e-06,
+      "loss": 3.3916,
+      "step": 4239
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.952727272727273e-06,
+      "loss": 3.5818,
+      "step": 4240
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.952e-06,
+      "loss": 3.3965,
+      "step": 4241
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.9512727272727274e-06,
+      "loss": 3.3302,
+      "step": 4242
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.9505454545454546e-06,
+      "loss": 3.6505,
+      "step": 4243
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.949818181818182e-06,
+      "loss": 3.429,
+      "step": 4244
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.9490909090909096e-06,
+      "loss": 3.4156,
+      "step": 4245
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 1.9483636363636364e-06,
+      "loss": 3.5919,
+      "step": 4246
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 1.9476363636363637e-06,
+      "loss": 3.678,
+      "step": 4247
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.946909090909091e-06,
+      "loss": 3.6524,
+      "step": 4248
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.9461818181818186e-06,
+      "loss": 3.4076,
+      "step": 4249
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.945454545454546e-06,
+      "loss": 3.3107,
+      "step": 4250
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 1.9447272727272727e-06,
+      "loss": 3.7611,
+      "step": 4251
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 1.944e-06,
+      "loss": 3.359,
+      "step": 4252
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 1.9432727272727276e-06,
+      "loss": 3.443,
+      "step": 4253
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.942545454545455e-06,
+      "loss": 3.6964,
+      "step": 4254
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.941818181818182e-06,
+      "loss": 3.3199,
+      "step": 4255
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.941090909090909e-06,
+      "loss": 3.3628,
+      "step": 4256
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.9403636363636366e-06,
+      "loss": 3.8267,
+      "step": 4257
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.939636363636364e-06,
+      "loss": 3.7445,
+      "step": 4258
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 1.938909090909091e-06,
+      "loss": 3.3603,
+      "step": 4259
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 1.9381818181818184e-06,
+      "loss": 3.7059,
+      "step": 4260
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 1.9374545454545457e-06,
+      "loss": 3.2907,
+      "step": 4261
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.936727272727273e-06,
+      "loss": 3.4562,
+      "step": 4262
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.936e-06,
+      "loss": 3.4859,
+      "step": 4263
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.9352727272727274e-06,
+      "loss": 3.3533,
+      "step": 4264
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.9345454545454547e-06,
+      "loss": 3.3934,
+      "step": 4265
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.933818181818182e-06,
+      "loss": 3.4339,
+      "step": 4266
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.933090909090909e-06,
+      "loss": 3.501,
+      "step": 4267
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.9323636363636365e-06,
+      "loss": 3.2587,
+      "step": 4268
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.9316363636363637e-06,
+      "loss": 3.4721,
+      "step": 4269
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 1.930909090909091e-06,
+      "loss": 3.5027,
+      "step": 4270
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 1.9301818181818182e-06,
+      "loss": 3.6338,
+      "step": 4271
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 1.929454545454546e-06,
+      "loss": 3.6104,
+      "step": 4272
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.9287272727272727e-06,
+      "loss": 3.5073,
+      "step": 4273
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.928e-06,
+      "loss": 3.4887,
+      "step": 4274
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.9272727272727273e-06,
+      "loss": 3.7014,
+      "step": 4275
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.926545454545455e-06,
+      "loss": 3.6904,
+      "step": 4276
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.925818181818182e-06,
+      "loss": 3.6576,
+      "step": 4277
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.925090909090909e-06,
+      "loss": 3.5323,
+      "step": 4278
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.9243636363636363e-06,
+      "loss": 3.357,
+      "step": 4279
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.923636363636364e-06,
+      "loss": 3.6488,
+      "step": 4280
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.9229090909090912e-06,
+      "loss": 3.497,
+      "step": 4281
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 1.9221818181818185e-06,
+      "loss": 3.5827,
+      "step": 4282
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 1.9214545454545453e-06,
+      "loss": 3.5437,
+      "step": 4283
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.920727272727273e-06,
+      "loss": 3.4011,
+      "step": 4284
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 3.4724,
+      "step": 4285
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.9192727272727275e-06,
+      "loss": 3.3091,
+      "step": 4286
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 1.9185454545454548e-06,
+      "loss": 3.7776,
+      "step": 4287
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 1.917818181818182e-06,
+      "loss": 3.6332,
+      "step": 4288
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 1.9170909090909093e-06,
+      "loss": 3.3259,
+      "step": 4289
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.9163636363636365e-06,
+      "loss": 3.3762,
+      "step": 4290
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.915636363636364e-06,
+      "loss": 3.2291,
+      "step": 4291
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.914909090909091e-06,
+      "loss": 3.2883,
+      "step": 4292
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 1.9141818181818183e-06,
+      "loss": 3.4176,
+      "step": 4293
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 1.9134545454545456e-06,
+      "loss": 3.4187,
+      "step": 4294
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 1.912727272727273e-06,
+      "loss": 3.4509,
+      "step": 4295
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 1.912e-06,
+      "loss": 3.3261,
+      "step": 4296
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 1.9112727272727273e-06,
+      "loss": 3.3584,
+      "step": 4297
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.9105454545454546e-06,
+      "loss": 3.6613,
+      "step": 4298
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.909818181818182e-06,
+      "loss": 3.5176,
+      "step": 4299
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.9090909090909095e-06,
+      "loss": 3.9326,
+      "step": 4300
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.9083636363636364e-06,
+      "loss": 3.8379,
+      "step": 4301
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.9076363636363636e-06,
+      "loss": 3.6642,
+      "step": 4302
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.9069090909090909e-06,
+      "loss": 3.7356,
+      "step": 4303
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 1.9061818181818183e-06,
+      "loss": 3.4641,
+      "step": 4304
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 1.9054545454545456e-06,
+      "loss": 3.4117,
+      "step": 4305
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 1.9047272727272729e-06,
+      "loss": 3.4371,
+      "step": 4306
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 1.9040000000000003e-06,
+      "loss": 3.7169,
+      "step": 4307
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 1.9032727272727276e-06,
+      "loss": 3.2566,
+      "step": 4308
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.9025454545454546e-06,
+      "loss": 3.382,
+      "step": 4309
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.9018181818181819e-06,
+      "loss": 3.5111,
+      "step": 4310
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.9010909090909093e-06,
+      "loss": 3.5773,
+      "step": 4311
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 1.9003636363636366e-06,
+      "loss": 3.3742,
+      "step": 4312
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 1.8996363636363639e-06,
+      "loss": 3.5802,
+      "step": 4313
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 1.898909090909091e-06,
+      "loss": 3.5385,
+      "step": 4314
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 1.8981818181818184e-06,
+      "loss": 3.5274,
+      "step": 4315
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 1.8974545454545456e-06,
+      "loss": 3.4674,
+      "step": 4316
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 1.8967272727272729e-06,
+      "loss": 3.4344,
+      "step": 4317
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 1.8960000000000001e-06,
+      "loss": 3.5403,
+      "step": 4318
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 1.8952727272727276e-06,
+      "loss": 3.3448,
+      "step": 4319
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.8945454545454547e-06,
+      "loss": 3.4101,
+      "step": 4320
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.893818181818182e-06,
+      "loss": 3.6248,
+      "step": 4321
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.8930909090909092e-06,
+      "loss": 3.8259,
+      "step": 4322
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.8923636363636366e-06,
+      "loss": 3.5598,
+      "step": 4323
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.891636363636364e-06,
+      "loss": 3.7484,
+      "step": 4324
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.890909090909091e-06,
+      "loss": 3.2805,
+      "step": 4325
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 1.8901818181818182e-06,
+      "loss": 3.3599,
+      "step": 4326
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 1.8894545454545457e-06,
+      "loss": 3.5733,
+      "step": 4327
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 1.888727272727273e-06,
+      "loss": 3.2955,
+      "step": 4328
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.8880000000000002e-06,
+      "loss": 3.3517,
+      "step": 4329
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.8872727272727272e-06,
+      "loss": 3.4771,
+      "step": 4330
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.8865454545454547e-06,
+      "loss": 3.7743,
+      "step": 4331
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.885818181818182e-06,
+      "loss": 3.4657,
+      "step": 4332
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.8850909090909092e-06,
+      "loss": 3.3785,
+      "step": 4333
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 1.8843636363636365e-06,
+      "loss": 3.6858,
+      "step": 4334
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 1.883636363636364e-06,
+      "loss": 3.2326,
+      "step": 4335
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 1.882909090909091e-06,
+      "loss": 3.4468,
+      "step": 4336
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.8821818181818182e-06,
+      "loss": 3.5897,
+      "step": 4337
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.8814545454545455e-06,
+      "loss": 3.3807,
+      "step": 4338
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.880727272727273e-06,
+      "loss": 3.454,
+      "step": 4339
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 3.4991,
+      "step": 4340
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 1.8792727272727273e-06,
+      "loss": 3.4034,
+      "step": 4341
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 1.8785454545454545e-06,
+      "loss": 3.452,
+      "step": 4342
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 1.877818181818182e-06,
+      "loss": 3.504,
+      "step": 4343
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 1.8770909090909092e-06,
+      "loss": 3.5154,
+      "step": 4344
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 1.8763636363636365e-06,
+      "loss": 3.6259,
+      "step": 4345
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 1.875636363636364e-06,
+      "loss": 3.3493,
+      "step": 4346
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 1.8749090909090912e-06,
+      "loss": 3.7485,
+      "step": 4347
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.8741818181818183e-06,
+      "loss": 3.3213,
+      "step": 4348
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.8734545454545455e-06,
+      "loss": 3.3062,
+      "step": 4349
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.872727272727273e-06,
+      "loss": 3.6295,
+      "step": 4350
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.8720000000000002e-06,
+      "loss": 3.3229,
+      "step": 4351
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.8712727272727275e-06,
+      "loss": 3.2855,
+      "step": 4352
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.8705454545454545e-06,
+      "loss": 3.5802,
+      "step": 4353
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 1.869818181818182e-06,
+      "loss": 3.3557,
+      "step": 4354
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 1.8690909090909093e-06,
+      "loss": 3.5267,
+      "step": 4355
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.8683636363636365e-06,
+      "loss": 3.8162,
+      "step": 4356
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.8676363636363638e-06,
+      "loss": 3.6356,
+      "step": 4357
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.8669090909090913e-06,
+      "loss": 3.5077,
+      "step": 4358
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.8661818181818183e-06,
+      "loss": 3.5932,
+      "step": 4359
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.8654545454545456e-06,
+      "loss": 3.3308,
+      "step": 4360
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.8647272727272728e-06,
+      "loss": 3.4394,
+      "step": 4361
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.8640000000000003e-06,
+      "loss": 3.5455,
+      "step": 4362
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.8632727272727275e-06,
+      "loss": 3.4246,
+      "step": 4363
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.8625454545454546e-06,
+      "loss": 3.5065,
+      "step": 4364
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 1.8618181818181818e-06,
+      "loss": 3.4363,
+      "step": 4365
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 1.8610909090909093e-06,
+      "loss": 3.7152,
+      "step": 4366
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.8603636363636366e-06,
+      "loss": 3.4873,
+      "step": 4367
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.8596363636363638e-06,
+      "loss": 3.2828,
+      "step": 4368
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.8589090909090909e-06,
+      "loss": 3.2664,
+      "step": 4369
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 1.8581818181818183e-06,
+      "loss": 3.6688,
+      "step": 4370
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 1.8574545454545456e-06,
+      "loss": 3.5023,
+      "step": 4371
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 1.8567272727272728e-06,
+      "loss": 3.3311,
+      "step": 4372
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.856e-06,
+      "loss": 3.5271,
+      "step": 4373
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.8552727272727276e-06,
+      "loss": 3.6172,
+      "step": 4374
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.8545454545454546e-06,
+      "loss": 3.4634,
+      "step": 4375
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 1.8538181818181819e-06,
+      "loss": 3.4517,
+      "step": 4376
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 1.8530909090909091e-06,
+      "loss": 3.406,
+      "step": 4377
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.8523636363636366e-06,
+      "loss": 3.3999,
+      "step": 4378
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.8516363636363639e-06,
+      "loss": 3.7169,
+      "step": 4379
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.850909090909091e-06,
+      "loss": 3.514,
+      "step": 4380
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 1.8501818181818184e-06,
+      "loss": 3.3759,
+      "step": 4381
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 1.8494545454545456e-06,
+      "loss": 3.4405,
+      "step": 4382
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 1.8487272727272729e-06,
+      "loss": 3.5948,
+      "step": 4383
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 1.8480000000000001e-06,
+      "loss": 3.3793,
+      "step": 4384
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 1.8472727272727276e-06,
+      "loss": 3.1982,
+      "step": 4385
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 1.8465454545454546e-06,
+      "loss": 3.3541,
+      "step": 4386
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.845818181818182e-06,
+      "loss": 3.524,
+      "step": 4387
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.8450909090909092e-06,
+      "loss": 3.3377,
+      "step": 4388
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.8443636363636366e-06,
+      "loss": 3.4149,
+      "step": 4389
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 1.8436363636363639e-06,
+      "loss": 3.6434,
+      "step": 4390
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 1.842909090909091e-06,
+      "loss": 3.4075,
+      "step": 4391
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.8421818181818182e-06,
+      "loss": 3.3771,
+      "step": 4392
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.8414545454545457e-06,
+      "loss": 3.3329,
+      "step": 4393
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.840727272727273e-06,
+      "loss": 3.5634,
+      "step": 4394
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 3.3555,
+      "step": 4395
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 1.8392727272727272e-06,
+      "loss": 3.5402,
+      "step": 4396
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 1.838545454545455e-06,
+      "loss": 3.5108,
+      "step": 4397
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.837818181818182e-06,
+      "loss": 3.5108,
+      "step": 4398
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.8370909090909092e-06,
+      "loss": 3.3913,
+      "step": 4399
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.8363636363636365e-06,
+      "loss": 3.4063,
+      "step": 4400
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 1.835636363636364e-06,
+      "loss": 3.3633,
+      "step": 4401
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 1.8349090909090912e-06,
+      "loss": 3.3998,
+      "step": 4402
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.8341818181818182e-06,
+      "loss": 3.7642,
+      "step": 4403
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.8334545454545455e-06,
+      "loss": 3.1674,
+      "step": 4404
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.832727272727273e-06,
+      "loss": 3.656,
+      "step": 4405
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 1.8320000000000002e-06,
+      "loss": 3.6082,
+      "step": 4406
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 1.8312727272727275e-06,
+      "loss": 3.6796,
+      "step": 4407
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 1.8305454545454545e-06,
+      "loss": 3.5172,
+      "step": 4408
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.829818181818182e-06,
+      "loss": 3.3592,
+      "step": 4409
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.8290909090909092e-06,
+      "loss": 3.3011,
+      "step": 4410
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.8283636363636365e-06,
+      "loss": 3.3566,
+      "step": 4411
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.8276363636363637e-06,
+      "loss": 3.4373,
+      "step": 4412
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.8269090909090912e-06,
+      "loss": 3.5361,
+      "step": 4413
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.8261818181818183e-06,
+      "loss": 3.6232,
+      "step": 4414
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.8254545454545455e-06,
+      "loss": 3.3926,
+      "step": 4415
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.824727272727273e-06,
+      "loss": 3.4591,
+      "step": 4416
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 1.8240000000000002e-06,
+      "loss": 3.335,
+      "step": 4417
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 1.8232727272727275e-06,
+      "loss": 3.3232,
+      "step": 4418
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 1.8225454545454545e-06,
+      "loss": 3.3501,
+      "step": 4419
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.821818181818182e-06,
+      "loss": 3.4656,
+      "step": 4420
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.8210909090909093e-06,
+      "loss": 3.4599,
+      "step": 4421
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.8203636363636365e-06,
+      "loss": 3.693,
+      "step": 4422
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.8196363636363638e-06,
+      "loss": 3.323,
+      "step": 4423
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.8189090909090912e-06,
+      "loss": 3.7289,
+      "step": 4424
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 3.5074,
+      "step": 4425
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 1.8174545454545455e-06,
+      "loss": 3.3413,
+      "step": 4426
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 1.8167272727272728e-06,
+      "loss": 3.2168,
+      "step": 4427
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.8160000000000003e-06,
+      "loss": 3.3057,
+      "step": 4428
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.8152727272727275e-06,
+      "loss": 3.2964,
+      "step": 4429
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.8145454545454546e-06,
+      "loss": 3.4701,
+      "step": 4430
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 1.8138181818181818e-06,
+      "loss": 3.5282,
+      "step": 4431
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 1.8130909090909093e-06,
+      "loss": 3.2338,
+      "step": 4432
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.17027027027027028,
+      "eval_loss": 3.458405017852783,
+      "eval_runtime": 143.138,
+      "eval_samples_per_second": 2.585,
+      "eval_steps_per_second": 0.65,
+      "step": 4432
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 1.8123636363636366e-06,
+      "loss": 3.3074,
+      "step": 4433
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 1.8116363636363638e-06,
+      "loss": 3.347,
+      "step": 4434
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 1.8109090909090909e-06,
+      "loss": 3.3817,
+      "step": 4435
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 1.8101818181818183e-06,
+      "loss": 3.4262,
+      "step": 4436
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 1.8094545454545456e-06,
+      "loss": 3.4069,
+      "step": 4437
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 1.8087272727272728e-06,
+      "loss": 3.6127,
+      "step": 4438
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 1.808e-06,
+      "loss": 3.4932,
+      "step": 4439
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 1.8072727272727276e-06,
+      "loss": 3.6224,
+      "step": 4440
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 1.8065454545454546e-06,
+      "loss": 3.4797,
+      "step": 4441
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 1.8058181818181819e-06,
+      "loss": 3.578,
+      "step": 4442
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 1.8050909090909091e-06,
+      "loss": 3.7681,
+      "step": 4443
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 1.8043636363636366e-06,
+      "loss": 3.3901,
+      "step": 4444
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 1.8036363636363638e-06,
+      "loss": 3.4332,
+      "step": 4445
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 1.8029090909090909e-06,
+      "loss": 3.7841,
+      "step": 4446
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 1.8021818181818181e-06,
+      "loss": 3.228,
+      "step": 4447
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 1.8014545454545456e-06,
+      "loss": 3.3737,
+      "step": 4448
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 1.8007272727272729e-06,
+      "loss": 3.3399,
+      "step": 4449
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 3.3253,
+      "step": 4450
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 1.7992727272727274e-06,
+      "loss": 3.3294,
+      "step": 4451
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 1.7985454545454549e-06,
+      "loss": 3.8163,
+      "step": 4452
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 1.797818181818182e-06,
+      "loss": 3.5138,
+      "step": 4453
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 1.7970909090909092e-06,
+      "loss": 3.2376,
+      "step": 4454
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 1.7963636363636366e-06,
+      "loss": 3.354,
+      "step": 4455
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 1.7956363636363639e-06,
+      "loss": 3.6623,
+      "step": 4456
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 1.7949090909090911e-06,
+      "loss": 3.3044,
+      "step": 4457
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 1.7941818181818182e-06,
+      "loss": 3.4408,
+      "step": 4458
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 1.7934545454545457e-06,
+      "loss": 3.4523,
+      "step": 4459
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 1.792727272727273e-06,
+      "loss": 3.3823,
+      "step": 4460
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 1.7920000000000002e-06,
+      "loss": 3.686,
+      "step": 4461
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 1.7912727272727274e-06,
+      "loss": 3.5262,
+      "step": 4462
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 1.7905454545454549e-06,
+      "loss": 3.3939,
+      "step": 4463
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 1.789818181818182e-06,
+      "loss": 3.7695,
+      "step": 4464
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 1.7890909090909092e-06,
+      "loss": 3.4047,
+      "step": 4465
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 1.7883636363636364e-06,
+      "loss": 3.363,
+      "step": 4466
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 1.787636363636364e-06,
+      "loss": 3.4663,
+      "step": 4467
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 1.7869090909090912e-06,
+      "loss": 3.7137,
+      "step": 4468
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 1.7861818181818182e-06,
+      "loss": 3.4127,
+      "step": 4469
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 1.7854545454545455e-06,
+      "loss": 3.6678,
+      "step": 4470
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 1.784727272727273e-06,
+      "loss": 3.6311,
+      "step": 4471
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 1.7840000000000002e-06,
+      "loss": 3.6997,
+      "step": 4472
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 1.7832727272727275e-06,
+      "loss": 3.6476,
+      "step": 4473
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 1.7825454545454545e-06,
+      "loss": 3.3088,
+      "step": 4474
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 1.781818181818182e-06,
+      "loss": 3.431,
+      "step": 4475
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 1.7810909090909092e-06,
+      "loss": 3.3803,
+      "step": 4476
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 1.7803636363636365e-06,
+      "loss": 3.6748,
+      "step": 4477
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 1.7796363636363637e-06,
+      "loss": 3.5638,
+      "step": 4478
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 1.7789090909090912e-06,
+      "loss": 3.5812,
+      "step": 4479
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 1.7781818181818183e-06,
+      "loss": 3.5195,
+      "step": 4480
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 1.7774545454545455e-06,
+      "loss": 3.2891,
+      "step": 4481
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 1.7767272727272728e-06,
+      "loss": 3.533,
+      "step": 4482
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 1.7760000000000002e-06,
+      "loss": 3.5147,
+      "step": 4483
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 1.7752727272727275e-06,
+      "loss": 3.6295,
+      "step": 4484
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 1.7745454545454545e-06,
+      "loss": 3.4756,
+      "step": 4485
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 1.7738181818181818e-06,
+      "loss": 3.525,
+      "step": 4486
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 1.7730909090909093e-06,
+      "loss": 3.5307,
+      "step": 4487
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 1.7723636363636365e-06,
+      "loss": 3.2665,
+      "step": 4488
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 1.7716363636363638e-06,
+      "loss": 3.4083,
+      "step": 4489
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 1.7709090909090912e-06,
+      "loss": 3.3607,
+      "step": 4490
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 1.7701818181818183e-06,
+      "loss": 3.2783,
+      "step": 4491
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 1.7694545454545455e-06,
+      "loss": 3.534,
+      "step": 4492
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 1.7687272727272728e-06,
+      "loss": 3.2155,
+      "step": 4493
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 1.7680000000000003e-06,
+      "loss": 3.7327,
+      "step": 4494
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 1.7672727272727275e-06,
+      "loss": 3.2159,
+      "step": 4495
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 1.7665454545454546e-06,
+      "loss": 3.2434,
+      "step": 4496
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 1.7658181818181818e-06,
+      "loss": 3.6957,
+      "step": 4497
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.7650909090909093e-06,
+      "loss": 3.4629,
+      "step": 4498
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.7643636363636366e-06,
+      "loss": 3.3739,
+      "step": 4499
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.7636363636363638e-06,
+      "loss": 3.3183,
+      "step": 4500
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.762909090909091e-06,
+      "loss": 3.4649,
+      "step": 4501
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.7621818181818185e-06,
+      "loss": 3.3215,
+      "step": 4502
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.7614545454545456e-06,
+      "loss": 3.1816,
+      "step": 4503
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.7607272727272728e-06,
+      "loss": 3.616,
+      "step": 4504
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.76e-06,
+      "loss": 3.384,
+      "step": 4505
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.7592727272727276e-06,
+      "loss": 3.4068,
+      "step": 4506
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.7585454545454548e-06,
+      "loss": 3.3264,
+      "step": 4507
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.7578181818181819e-06,
+      "loss": 3.4851,
+      "step": 4508
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.7570909090909091e-06,
+      "loss": 3.5934,
+      "step": 4509
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.7563636363636366e-06,
+      "loss": 3.2928,
+      "step": 4510
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.7556363636363638e-06,
+      "loss": 3.4137,
+      "step": 4511
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.754909090909091e-06,
+      "loss": 3.5836,
+      "step": 4512
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 1.7541818181818181e-06,
+      "loss": 3.6658,
+      "step": 4513
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.7534545454545456e-06,
+      "loss": 3.5809,
+      "step": 4514
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.7527272727272729e-06,
+      "loss": 3.7608,
+      "step": 4515
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.7520000000000001e-06,
+      "loss": 3.5973,
+      "step": 4516
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.7512727272727274e-06,
+      "loss": 3.4076,
+      "step": 4517
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.7505454545454548e-06,
+      "loss": 3.6038,
+      "step": 4518
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.7498181818181819e-06,
+      "loss": 3.5941,
+      "step": 4519
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.7490909090909092e-06,
+      "loss": 3.2774,
+      "step": 4520
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.7483636363636364e-06,
+      "loss": 3.5812,
+      "step": 4521
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.7476363636363639e-06,
+      "loss": 3.4412,
+      "step": 4522
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.7469090909090911e-06,
+      "loss": 3.6463,
+      "step": 4523
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.7461818181818182e-06,
+      "loss": 3.5273,
+      "step": 4524
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.7454545454545456e-06,
+      "loss": 3.5971,
+      "step": 4525
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.744727272727273e-06,
+      "loss": 3.408,
+      "step": 4526
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.7440000000000002e-06,
+      "loss": 3.4221,
+      "step": 4527
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.7432727272727274e-06,
+      "loss": 3.5385,
+      "step": 4528
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.7425454545454549e-06,
+      "loss": 3.4544,
+      "step": 4529
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.741818181818182e-06,
+      "loss": 3.2895,
+      "step": 4530
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.7410909090909092e-06,
+      "loss": 3.2361,
+      "step": 4531
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.7403636363636364e-06,
+      "loss": 3.4719,
+      "step": 4532
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.739636363636364e-06,
+      "loss": 3.6658,
+      "step": 4533
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.7389090909090912e-06,
+      "loss": 3.5018,
+      "step": 4534
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.7381818181818182e-06,
+      "loss": 3.3359,
+      "step": 4535
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.7374545454545455e-06,
+      "loss": 3.3784,
+      "step": 4536
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.736727272727273e-06,
+      "loss": 3.4501,
+      "step": 4537
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.7360000000000002e-06,
+      "loss": 3.4771,
+      "step": 4538
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.7352727272727274e-06,
+      "loss": 3.4971,
+      "step": 4539
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.7345454545454545e-06,
+      "loss": 3.7744,
+      "step": 4540
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.733818181818182e-06,
+      "loss": 3.6219,
+      "step": 4541
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.7330909090909092e-06,
+      "loss": 3.5175,
+      "step": 4542
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.7323636363636365e-06,
+      "loss": 3.582,
+      "step": 4543
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 1.7316363636363637e-06,
+      "loss": 3.2335,
+      "step": 4544
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.7309090909090912e-06,
+      "loss": 3.4033,
+      "step": 4545
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.7301818181818185e-06,
+      "loss": 3.8515,
+      "step": 4546
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.7294545454545455e-06,
+      "loss": 3.5497,
+      "step": 4547
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.7287272727272728e-06,
+      "loss": 3.5395,
+      "step": 4548
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.7280000000000002e-06,
+      "loss": 3.1683,
+      "step": 4549
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.7272727272727275e-06,
+      "loss": 3.3743,
+      "step": 4550
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.7265454545454547e-06,
+      "loss": 3.2923,
+      "step": 4551
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.7258181818181818e-06,
+      "loss": 3.5061,
+      "step": 4552
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.7250909090909093e-06,
+      "loss": 3.5285,
+      "step": 4553
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.7243636363636365e-06,
+      "loss": 3.4794,
+      "step": 4554
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.7236363636363638e-06,
+      "loss": 3.4992,
+      "step": 4555
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.722909090909091e-06,
+      "loss": 3.3889,
+      "step": 4556
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.7221818181818185e-06,
+      "loss": 3.1863,
+      "step": 4557
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.7214545454545455e-06,
+      "loss": 3.682,
+      "step": 4558
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.7207272727272728e-06,
+      "loss": 3.2787,
+      "step": 4559
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.72e-06,
+      "loss": 3.5739,
+      "step": 4560
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.7192727272727275e-06,
+      "loss": 3.4177,
+      "step": 4561
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.7185454545454548e-06,
+      "loss": 3.787,
+      "step": 4562
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.7178181818181818e-06,
+      "loss": 3.3499,
+      "step": 4563
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.7170909090909093e-06,
+      "loss": 3.4488,
+      "step": 4564
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.7163636363636365e-06,
+      "loss": 3.3797,
+      "step": 4565
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.7156363636363638e-06,
+      "loss": 3.3849,
+      "step": 4566
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.714909090909091e-06,
+      "loss": 3.4321,
+      "step": 4567
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.7141818181818185e-06,
+      "loss": 3.4573,
+      "step": 4568
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.7134545454545456e-06,
+      "loss": 3.3712,
+      "step": 4569
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.7127272727272728e-06,
+      "loss": 3.3501,
+      "step": 4570
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.712e-06,
+      "loss": 3.4197,
+      "step": 4571
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.7112727272727276e-06,
+      "loss": 3.455,
+      "step": 4572
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.7105454545454548e-06,
+      "loss": 3.5438,
+      "step": 4573
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.7098181818181819e-06,
+      "loss": 3.7015,
+      "step": 4574
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.7090909090909091e-06,
+      "loss": 3.6374,
+      "step": 4575
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.7083636363636366e-06,
+      "loss": 3.263,
+      "step": 4576
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.7076363636363638e-06,
+      "loss": 3.4289,
+      "step": 4577
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.706909090909091e-06,
+      "loss": 3.5838,
+      "step": 4578
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.7061818181818181e-06,
+      "loss": 3.2975,
+      "step": 4579
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.7054545454545456e-06,
+      "loss": 3.3077,
+      "step": 4580
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.7047272727272729e-06,
+      "loss": 3.3703,
+      "step": 4581
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.7040000000000001e-06,
+      "loss": 3.4918,
+      "step": 4582
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.7032727272727274e-06,
+      "loss": 3.1396,
+      "step": 4583
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.7025454545454548e-06,
+      "loss": 3.6524,
+      "step": 4584
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.7018181818181819e-06,
+      "loss": 3.3796,
+      "step": 4585
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.7010909090909091e-06,
+      "loss": 3.9603,
+      "step": 4586
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.7003636363636364e-06,
+      "loss": 3.2634,
+      "step": 4587
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.6996363636363639e-06,
+      "loss": 3.6195,
+      "step": 4588
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.6989090909090911e-06,
+      "loss": 3.5322,
+      "step": 4589
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.6981818181818182e-06,
+      "loss": 3.5875,
+      "step": 4590
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 1.6974545454545454e-06,
+      "loss": 3.3127,
+      "step": 4591
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.696727272727273e-06,
+      "loss": 3.377,
+      "step": 4592
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.6960000000000002e-06,
+      "loss": 3.3517,
+      "step": 4593
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.6952727272727274e-06,
+      "loss": 3.5214,
+      "step": 4594
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.6945454545454545e-06,
+      "loss": 3.2942,
+      "step": 4595
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.6938181818181821e-06,
+      "loss": 3.574,
+      "step": 4596
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.6930909090909092e-06,
+      "loss": 3.4561,
+      "step": 4597
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.6923636363636364e-06,
+      "loss": 3.382,
+      "step": 4598
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.691636363636364e-06,
+      "loss": 3.4818,
+      "step": 4599
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.6909090909090912e-06,
+      "loss": 3.4061,
+      "step": 4600
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.6901818181818184e-06,
+      "loss": 3.6684,
+      "step": 4601
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.6894545454545455e-06,
+      "loss": 3.7164,
+      "step": 4602
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.688727272727273e-06,
+      "loss": 3.2625,
+      "step": 4603
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.6880000000000002e-06,
+      "loss": 3.3782,
+      "step": 4604
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.6872727272727274e-06,
+      "loss": 3.2965,
+      "step": 4605
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 1.6865454545454547e-06,
+      "loss": 3.5019,
+      "step": 4606
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 1.6858181818181822e-06,
+      "loss": 3.3663,
+      "step": 4607
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.6850909090909092e-06,
+      "loss": 3.4547,
+      "step": 4608
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.6843636363636365e-06,
+      "loss": 3.4722,
+      "step": 4609
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.6836363636363637e-06,
+      "loss": 3.2204,
+      "step": 4610
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.6829090909090912e-06,
+      "loss": 3.4656,
+      "step": 4611
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.6821818181818185e-06,
+      "loss": 3.8617,
+      "step": 4612
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.6814545454545455e-06,
+      "loss": 3.4415,
+      "step": 4613
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.6807272727272728e-06,
+      "loss": 3.5875,
+      "step": 4614
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 3.4872,
+      "step": 4615
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.6792727272727275e-06,
+      "loss": 3.7777,
+      "step": 4616
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.6785454545454547e-06,
+      "loss": 3.2328,
+      "step": 4617
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.6778181818181818e-06,
+      "loss": 3.6562,
+      "step": 4618
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.6770909090909092e-06,
+      "loss": 3.2018,
+      "step": 4619
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.6763636363636365e-06,
+      "loss": 3.4411,
+      "step": 4620
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 1.6756363636363638e-06,
+      "loss": 3.3805,
+      "step": 4621
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.674909090909091e-06,
+      "loss": 3.3958,
+      "step": 4622
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.6741818181818185e-06,
+      "loss": 3.3938,
+      "step": 4623
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.6734545454545455e-06,
+      "loss": 3.4282,
+      "step": 4624
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.6727272727272728e-06,
+      "loss": 3.364,
+      "step": 4625
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.672e-06,
+      "loss": 3.7387,
+      "step": 4626
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.6712727272727275e-06,
+      "loss": 3.363,
+      "step": 4627
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.6705454545454548e-06,
+      "loss": 3.4126,
+      "step": 4628
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.6698181818181818e-06,
+      "loss": 3.4557,
+      "step": 4629
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.669090909090909e-06,
+      "loss": 3.4804,
+      "step": 4630
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.6683636363636365e-06,
+      "loss": 3.4301,
+      "step": 4631
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.6676363636363638e-06,
+      "loss": 3.5513,
+      "step": 4632
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.666909090909091e-06,
+      "loss": 3.1861,
+      "step": 4633
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.6661818181818185e-06,
+      "loss": 3.454,
+      "step": 4634
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.6654545454545456e-06,
+      "loss": 3.5082,
+      "step": 4635
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.6647272727272728e-06,
+      "loss": 3.6031,
+      "step": 4636
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.664e-06,
+      "loss": 3.5359,
+      "step": 4637
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 1.6632727272727275e-06,
+      "loss": 3.5417,
+      "step": 4638
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.6625454545454548e-06,
+      "loss": 3.5523,
+      "step": 4639
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.6618181818181818e-06,
+      "loss": 3.634,
+      "step": 4640
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.661090909090909e-06,
+      "loss": 3.7337,
+      "step": 4641
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.6603636363636366e-06,
+      "loss": 3.3968,
+      "step": 4642
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.6596363636363638e-06,
+      "loss": 3.4406,
+      "step": 4643
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.658909090909091e-06,
+      "loss": 3.3712,
+      "step": 4644
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.6581818181818181e-06,
+      "loss": 3.4515,
+      "step": 4645
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.6574545454545458e-06,
+      "loss": 3.3041,
+      "step": 4646
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.6567272727272729e-06,
+      "loss": 3.5987,
+      "step": 4647
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.6560000000000001e-06,
+      "loss": 3.3485,
+      "step": 4648
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.6552727272727274e-06,
+      "loss": 3.778,
+      "step": 4649
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.6545454545454548e-06,
+      "loss": 3.5842,
+      "step": 4650
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.653818181818182e-06,
+      "loss": 3.6504,
+      "step": 4651
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 1.6530909090909091e-06,
+      "loss": 3.3677,
+      "step": 4652
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.6523636363636364e-06,
+      "loss": 3.1612,
+      "step": 4653
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.6516363636363639e-06,
+      "loss": 3.4034,
+      "step": 4654
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.6509090909090911e-06,
+      "loss": 3.2508,
+      "step": 4655
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.6501818181818184e-06,
+      "loss": 3.4457,
+      "step": 4656
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.6494545454545454e-06,
+      "loss": 3.3542,
+      "step": 4657
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.6487272727272729e-06,
+      "loss": 3.2713,
+      "step": 4658
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.6480000000000001e-06,
+      "loss": 3.231,
+      "step": 4659
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.6472727272727274e-06,
+      "loss": 3.3116,
+      "step": 4660
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.6465454545454547e-06,
+      "loss": 3.6108,
+      "step": 4661
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.6458181818181821e-06,
+      "loss": 3.4082,
+      "step": 4662
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.6450909090909092e-06,
+      "loss": 3.3034,
+      "step": 4663
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.6443636363636364e-06,
+      "loss": 3.8163,
+      "step": 4664
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.6436363636363637e-06,
+      "loss": 3.4716,
+      "step": 4665
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.6429090909090912e-06,
+      "loss": 3.2005,
+      "step": 4666
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.6421818181818184e-06,
+      "loss": 3.4712,
+      "step": 4667
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.6414545454545455e-06,
+      "loss": 3.5323,
+      "step": 4668
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.6407272727272727e-06,
+      "loss": 3.5034,
+      "step": 4669
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 3.5803,
+      "step": 4670
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.6392727272727274e-06,
+      "loss": 3.2334,
+      "step": 4671
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.6385454545454547e-06,
+      "loss": 3.5468,
+      "step": 4672
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.6378181818181822e-06,
+      "loss": 3.7412,
+      "step": 4673
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.6370909090909092e-06,
+      "loss": 3.4538,
+      "step": 4674
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.6363636363636365e-06,
+      "loss": 3.2666,
+      "step": 4675
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.6356363636363637e-06,
+      "loss": 3.5332,
+      "step": 4676
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.6349090909090912e-06,
+      "loss": 3.1975,
+      "step": 4677
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.6341818181818184e-06,
+      "loss": 3.5988,
+      "step": 4678
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.6334545454545455e-06,
+      "loss": 3.9488,
+      "step": 4679
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.6327272727272727e-06,
+      "loss": 3.2659,
+      "step": 4680
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.6320000000000002e-06,
+      "loss": 3.6132,
+      "step": 4681
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.6312727272727275e-06,
+      "loss": 3.5099,
+      "step": 4682
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.6305454545454547e-06,
+      "loss": 3.3583,
+      "step": 4683
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.6298181818181818e-06,
+      "loss": 3.612,
+      "step": 4684
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.6290909090909092e-06,
+      "loss": 3.5139,
+      "step": 4685
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.6283636363636365e-06,
+      "loss": 3.4672,
+      "step": 4686
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.6276363636363638e-06,
+      "loss": 3.5769,
+      "step": 4687
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.626909090909091e-06,
+      "loss": 3.4491,
+      "step": 4688
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 1.6261818181818185e-06,
+      "loss": 3.2723,
+      "step": 4689
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 1.6254545454545455e-06,
+      "loss": 3.2686,
+      "step": 4690
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.6247272727272728e-06,
+      "loss": 3.4557,
+      "step": 4691
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.624e-06,
+      "loss": 3.5423,
+      "step": 4692
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.6232727272727275e-06,
+      "loss": 3.6074,
+      "step": 4693
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.6225454545454548e-06,
+      "loss": 3.4536,
+      "step": 4694
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.6218181818181818e-06,
+      "loss": 3.2591,
+      "step": 4695
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.621090909090909e-06,
+      "loss": 3.4317,
+      "step": 4696
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.6203636363636365e-06,
+      "loss": 3.3564,
+      "step": 4697
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.6196363636363638e-06,
+      "loss": 3.3466,
+      "step": 4698
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.618909090909091e-06,
+      "loss": 3.3218,
+      "step": 4699
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.618181818181818e-06,
+      "loss": 3.3264,
+      "step": 4700
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.6174545454545458e-06,
+      "loss": 3.5637,
+      "step": 4701
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.6167272727272728e-06,
+      "loss": 3.4715,
+      "step": 4702
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.616e-06,
+      "loss": 3.7355,
+      "step": 4703
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.6152727272727273e-06,
+      "loss": 3.381,
+      "step": 4704
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.6145454545454548e-06,
+      "loss": 3.2971,
+      "step": 4705
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.613818181818182e-06,
+      "loss": 3.2479,
+      "step": 4706
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.613090909090909e-06,
+      "loss": 3.6456,
+      "step": 4707
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.6123636363636366e-06,
+      "loss": 3.3779,
+      "step": 4708
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.6116363636363638e-06,
+      "loss": 3.2244,
+      "step": 4709
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.15945945945945947,
+      "eval_loss": 3.44972825050354,
+      "eval_runtime": 143.6192,
+      "eval_samples_per_second": 2.576,
+      "eval_steps_per_second": 0.648,
+      "step": 4709
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.610909090909091e-06,
+      "loss": 3.6079,
+      "step": 4710
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.6101818181818183e-06,
+      "loss": 3.2757,
+      "step": 4711
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.6094545454545458e-06,
+      "loss": 3.5687,
+      "step": 4712
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.6087272727272728e-06,
+      "loss": 3.5477,
+      "step": 4713
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.608e-06,
+      "loss": 3.3837,
+      "step": 4714
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.6072727272727274e-06,
+      "loss": 3.3891,
+      "step": 4715
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.6065454545454548e-06,
+      "loss": 3.4035,
+      "step": 4716
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.605818181818182e-06,
+      "loss": 3.6682,
+      "step": 4717
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.6050909090909091e-06,
+      "loss": 3.388,
+      "step": 4718
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.6043636363636364e-06,
+      "loss": 3.14,
+      "step": 4719
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.6036363636363639e-06,
+      "loss": 3.2924,
+      "step": 4720
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 1.6029090909090911e-06,
+      "loss": 3.7094,
+      "step": 4721
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.6021818181818184e-06,
+      "loss": 3.5189,
+      "step": 4722
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.6014545454545454e-06,
+      "loss": 3.4716,
+      "step": 4723
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.6007272727272729e-06,
+      "loss": 3.7913,
+      "step": 4724
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 3.2232,
+      "step": 4725
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.5992727272727274e-06,
+      "loss": 3.4165,
+      "step": 4726
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.5985454545454547e-06,
+      "loss": 3.8418,
+      "step": 4727
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.5978181818181821e-06,
+      "loss": 3.4333,
+      "step": 4728
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.5970909090909092e-06,
+      "loss": 3.3427,
+      "step": 4729
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.5963636363636364e-06,
+      "loss": 3.5226,
+      "step": 4730
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.5956363636363637e-06,
+      "loss": 3.441,
+      "step": 4731
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.5949090909090911e-06,
+      "loss": 3.2405,
+      "step": 4732
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.5941818181818184e-06,
+      "loss": 3.6867,
+      "step": 4733
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.5934545454545454e-06,
+      "loss": 3.2055,
+      "step": 4734
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.5927272727272727e-06,
+      "loss": 3.4355,
+      "step": 4735
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.5920000000000002e-06,
+      "loss": 3.3342,
+      "step": 4736
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.5912727272727274e-06,
+      "loss": 3.4231,
+      "step": 4737
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 1.5905454545454547e-06,
+      "loss": 3.4316,
+      "step": 4738
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.5898181818181817e-06,
+      "loss": 3.5669,
+      "step": 4739
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.5890909090909092e-06,
+      "loss": 3.2137,
+      "step": 4740
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.5883636363636365e-06,
+      "loss": 3.7207,
+      "step": 4741
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.5876363636363637e-06,
+      "loss": 3.2353,
+      "step": 4742
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.5869090909090912e-06,
+      "loss": 3.854,
+      "step": 4743
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.5861818181818184e-06,
+      "loss": 3.2971,
+      "step": 4744
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.5854545454545455e-06,
+      "loss": 3.3812,
+      "step": 4745
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 1.5847272727272727e-06,
+      "loss": 3.6155,
+      "step": 4746
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.5840000000000002e-06,
+      "loss": 3.4752,
+      "step": 4747
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.5832727272727275e-06,
+      "loss": 3.432,
+      "step": 4748
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.5825454545454547e-06,
+      "loss": 3.3311,
+      "step": 4749
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 1.5818181818181818e-06,
+      "loss": 3.6685,
+      "step": 4750
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 1.5810909090909094e-06,
+      "loss": 3.2689,
+      "step": 4751
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.5803636363636365e-06,
+      "loss": 3.5768,
+      "step": 4752
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.5796363636363637e-06,
+      "loss": 3.5581,
+      "step": 4753
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.578909090909091e-06,
+      "loss": 3.3583,
+      "step": 4754
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.5781818181818185e-06,
+      "loss": 3.3616,
+      "step": 4755
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.5774545454545457e-06,
+      "loss": 3.594,
+      "step": 4756
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.5767272727272728e-06,
+      "loss": 3.6285,
+      "step": 4757
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.576e-06,
+      "loss": 3.2533,
+      "step": 4758
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.5752727272727275e-06,
+      "loss": 3.335,
+      "step": 4759
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.5745454545454548e-06,
+      "loss": 3.556,
+      "step": 4760
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.573818181818182e-06,
+      "loss": 3.358,
+      "step": 4761
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.573090909090909e-06,
+      "loss": 3.5439,
+      "step": 4762
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.5723636363636365e-06,
+      "loss": 3.3709,
+      "step": 4763
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.5716363636363638e-06,
+      "loss": 3.1965,
+      "step": 4764
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.570909090909091e-06,
+      "loss": 3.5594,
+      "step": 4765
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.5701818181818183e-06,
+      "loss": 3.641,
+      "step": 4766
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.5694545454545458e-06,
+      "loss": 3.3099,
+      "step": 4767
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.5687272727272728e-06,
+      "loss": 3.4906,
+      "step": 4768
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.568e-06,
+      "loss": 3.4224,
+      "step": 4769
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.5672727272727273e-06,
+      "loss": 3.3392,
+      "step": 4770
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.5665454545454548e-06,
+      "loss": 3.5752,
+      "step": 4771
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.565818181818182e-06,
+      "loss": 3.427,
+      "step": 4772
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.565090909090909e-06,
+      "loss": 3.124,
+      "step": 4773
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.5643636363636363e-06,
+      "loss": 3.4719,
+      "step": 4774
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.5636363636363638e-06,
+      "loss": 3.356,
+      "step": 4775
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.562909090909091e-06,
+      "loss": 3.4003,
+      "step": 4776
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.5621818181818183e-06,
+      "loss": 3.3209,
+      "step": 4777
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.5614545454545454e-06,
+      "loss": 3.2112,
+      "step": 4778
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.5607272727272728e-06,
+      "loss": 3.2878,
+      "step": 4779
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.56e-06,
+      "loss": 3.5847,
+      "step": 4780
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.5592727272727274e-06,
+      "loss": 3.729,
+      "step": 4781
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.5585454545454548e-06,
+      "loss": 3.3995,
+      "step": 4782
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.557818181818182e-06,
+      "loss": 3.5546,
+      "step": 4783
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.5570909090909091e-06,
+      "loss": 3.326,
+      "step": 4784
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.5563636363636364e-06,
+      "loss": 3.7201,
+      "step": 4785
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.5556363636363639e-06,
+      "loss": 3.3938,
+      "step": 4786
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.5549090909090911e-06,
+      "loss": 3.5788,
+      "step": 4787
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.5541818181818184e-06,
+      "loss": 3.3373,
+      "step": 4788
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.5534545454545454e-06,
+      "loss": 3.4106,
+      "step": 4789
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.5527272727272729e-06,
+      "loss": 3.6174,
+      "step": 4790
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.5520000000000001e-06,
+      "loss": 3.4307,
+      "step": 4791
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.5512727272727274e-06,
+      "loss": 3.4366,
+      "step": 4792
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.5505454545454546e-06,
+      "loss": 3.3086,
+      "step": 4793
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.5498181818181821e-06,
+      "loss": 3.8186,
+      "step": 4794
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.5490909090909092e-06,
+      "loss": 3.4859,
+      "step": 4795
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.5483636363636364e-06,
+      "loss": 3.5214,
+      "step": 4796
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.5476363636363637e-06,
+      "loss": 3.3899,
+      "step": 4797
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.5469090909090911e-06,
+      "loss": 3.9106,
+      "step": 4798
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.5461818181818184e-06,
+      "loss": 3.4236,
+      "step": 4799
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.5454545454545454e-06,
+      "loss": 3.3913,
+      "step": 4800
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.5447272727272727e-06,
+      "loss": 3.5841,
+      "step": 4801
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.5440000000000002e-06,
+      "loss": 3.4559,
+      "step": 4802
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.5432727272727274e-06,
+      "loss": 3.2077,
+      "step": 4803
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.5425454545454547e-06,
+      "loss": 3.1751,
+      "step": 4804
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.5418181818181817e-06,
+      "loss": 3.4373,
+      "step": 4805
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.5410909090909094e-06,
+      "loss": 3.8029,
+      "step": 4806
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.5403636363636365e-06,
+      "loss": 3.4498,
+      "step": 4807
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.5396363636363637e-06,
+      "loss": 3.4136,
+      "step": 4808
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.538909090909091e-06,
+      "loss": 3.2118,
+      "step": 4809
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.5381818181818184e-06,
+      "loss": 3.5335,
+      "step": 4810
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.5374545454545457e-06,
+      "loss": 3.4339,
+      "step": 4811
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.5367272727272727e-06,
+      "loss": 3.2865,
+      "step": 4812
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.536e-06,
+      "loss": 3.5472,
+      "step": 4813
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.5352727272727275e-06,
+      "loss": 3.2965,
+      "step": 4814
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.5345454545454547e-06,
+      "loss": 3.5496,
+      "step": 4815
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.533818181818182e-06,
+      "loss": 3.6936,
+      "step": 4816
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.5330909090909094e-06,
+      "loss": 3.503,
+      "step": 4817
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.5323636363636365e-06,
+      "loss": 3.3279,
+      "step": 4818
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.5316363636363637e-06,
+      "loss": 3.3831,
+      "step": 4819
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.530909090909091e-06,
+      "loss": 3.3939,
+      "step": 4820
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 1.5301818181818185e-06,
+      "loss": 3.9612,
+      "step": 4821
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.5294545454545457e-06,
+      "loss": 3.7434,
+      "step": 4822
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.5287272727272728e-06,
+      "loss": 3.8907,
+      "step": 4823
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.528e-06,
+      "loss": 3.6046,
+      "step": 4824
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.5272727272727275e-06,
+      "loss": 3.5326,
+      "step": 4825
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.5265454545454547e-06,
+      "loss": 3.5618,
+      "step": 4826
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.525818181818182e-06,
+      "loss": 3.2908,
+      "step": 4827
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.525090909090909e-06,
+      "loss": 3.348,
+      "step": 4828
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.5243636363636365e-06,
+      "loss": 3.2298,
+      "step": 4829
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.5236363636363638e-06,
+      "loss": 3.4695,
+      "step": 4830
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.522909090909091e-06,
+      "loss": 3.5993,
+      "step": 4831
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.5221818181818183e-06,
+      "loss": 3.6589,
+      "step": 4832
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.5214545454545458e-06,
+      "loss": 3.206,
+      "step": 4833
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.5207272727272728e-06,
+      "loss": 3.3393,
+      "step": 4834
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.52e-06,
+      "loss": 3.3122,
+      "step": 4835
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.5192727272727273e-06,
+      "loss": 3.7211,
+      "step": 4836
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.5185454545454548e-06,
+      "loss": 3.3946,
+      "step": 4837
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.517818181818182e-06,
+      "loss": 3.4489,
+      "step": 4838
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.517090909090909e-06,
+      "loss": 3.7042,
+      "step": 4839
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.5163636363636363e-06,
+      "loss": 3.4734,
+      "step": 4840
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.5156363636363638e-06,
+      "loss": 3.5086,
+      "step": 4841
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.514909090909091e-06,
+      "loss": 3.444,
+      "step": 4842
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.5141818181818183e-06,
+      "loss": 3.5897,
+      "step": 4843
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.5134545454545454e-06,
+      "loss": 3.5639,
+      "step": 4844
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.5127272727272728e-06,
+      "loss": 3.4986,
+      "step": 4845
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 1.512e-06,
+      "loss": 3.646,
+      "step": 4846
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.5112727272727273e-06,
+      "loss": 3.6958,
+      "step": 4847
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.5105454545454546e-06,
+      "loss": 3.4286,
+      "step": 4848
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.509818181818182e-06,
+      "loss": 3.3352,
+      "step": 4849
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.5090909090909091e-06,
+      "loss": 3.4789,
+      "step": 4850
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 1.5083636363636364e-06,
+      "loss": 3.4365,
+      "step": 4851
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.5076363636363638e-06,
+      "loss": 3.5209,
+      "step": 4852
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.506909090909091e-06,
+      "loss": 3.3496,
+      "step": 4853
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.5061818181818184e-06,
+      "loss": 3.2735,
+      "step": 4854
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.5054545454545454e-06,
+      "loss": 3.5502,
+      "step": 4855
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.504727272727273e-06,
+      "loss": 3.5185,
+      "step": 4856
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.5040000000000001e-06,
+      "loss": 3.1639,
+      "step": 4857
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.5032727272727274e-06,
+      "loss": 3.2811,
+      "step": 4858
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.5025454545454546e-06,
+      "loss": 3.4157,
+      "step": 4859
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.5018181818181821e-06,
+      "loss": 3.3624,
+      "step": 4860
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.5010909090909094e-06,
+      "loss": 3.6414,
+      "step": 4861
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.5003636363636364e-06,
+      "loss": 3.3737,
+      "step": 4862
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.4996363636363637e-06,
+      "loss": 3.2633,
+      "step": 4863
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.4989090909090911e-06,
+      "loss": 3.3719,
+      "step": 4864
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.4981818181818184e-06,
+      "loss": 3.5367,
+      "step": 4865
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.4974545454545456e-06,
+      "loss": 3.7023,
+      "step": 4866
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.4967272727272727e-06,
+      "loss": 3.9143,
+      "step": 4867
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.4960000000000002e-06,
+      "loss": 3.1246,
+      "step": 4868
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.4952727272727274e-06,
+      "loss": 3.5047,
+      "step": 4869
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.4945454545454547e-06,
+      "loss": 3.4295,
+      "step": 4870
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.493818181818182e-06,
+      "loss": 3.5122,
+      "step": 4871
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.4930909090909094e-06,
+      "loss": 3.3579,
+      "step": 4872
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.4923636363636364e-06,
+      "loss": 3.373,
+      "step": 4873
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.4916363636363637e-06,
+      "loss": 3.2126,
+      "step": 4874
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.490909090909091e-06,
+      "loss": 3.2009,
+      "step": 4875
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.4901818181818184e-06,
+      "loss": 3.4183,
+      "step": 4876
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.4894545454545457e-06,
+      "loss": 3.393,
+      "step": 4877
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.4887272727272727e-06,
+      "loss": 3.5687,
+      "step": 4878
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.488e-06,
+      "loss": 3.3867,
+      "step": 4879
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.4872727272727275e-06,
+      "loss": 3.4442,
+      "step": 4880
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.4865454545454547e-06,
+      "loss": 3.3209,
+      "step": 4881
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.485818181818182e-06,
+      "loss": 3.666,
+      "step": 4882
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.485090909090909e-06,
+      "loss": 3.4965,
+      "step": 4883
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.4843636363636365e-06,
+      "loss": 3.328,
+      "step": 4884
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.4836363636363637e-06,
+      "loss": 3.318,
+      "step": 4885
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.482909090909091e-06,
+      "loss": 3.2386,
+      "step": 4886
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.4821818181818182e-06,
+      "loss": 3.4493,
+      "step": 4887
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.4814545454545457e-06,
+      "loss": 3.6731,
+      "step": 4888
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.4807272727272728e-06,
+      "loss": 3.6449,
+      "step": 4889
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.48e-06,
+      "loss": 3.556,
+      "step": 4890
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.4792727272727275e-06,
+      "loss": 3.4808,
+      "step": 4891
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.4785454545454547e-06,
+      "loss": 3.3702,
+      "step": 4892
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.477818181818182e-06,
+      "loss": 3.3796,
+      "step": 4893
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.477090909090909e-06,
+      "loss": 3.3032,
+      "step": 4894
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.4763636363636365e-06,
+      "loss": 3.3703,
+      "step": 4895
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.4756363636363638e-06,
+      "loss": 3.3208,
+      "step": 4896
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.474909090909091e-06,
+      "loss": 3.3211,
+      "step": 4897
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.4741818181818183e-06,
+      "loss": 3.3915,
+      "step": 4898
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.4734545454545458e-06,
+      "loss": 3.3336,
+      "step": 4899
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.4727272727272728e-06,
+      "loss": 3.3715,
+      "step": 4900
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.472e-06,
+      "loss": 3.3141,
+      "step": 4901
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.4712727272727273e-06,
+      "loss": 3.6933,
+      "step": 4902
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.4705454545454548e-06,
+      "loss": 3.6723,
+      "step": 4903
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.469818181818182e-06,
+      "loss": 3.4435,
+      "step": 4904
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.469090909090909e-06,
+      "loss": 3.3957,
+      "step": 4905
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.4683636363636363e-06,
+      "loss": 3.5439,
+      "step": 4906
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.4676363636363638e-06,
+      "loss": 3.3412,
+      "step": 4907
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.466909090909091e-06,
+      "loss": 3.4045,
+      "step": 4908
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.4661818181818183e-06,
+      "loss": 3.2396,
+      "step": 4909
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.4654545454545454e-06,
+      "loss": 3.4363,
+      "step": 4910
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.464727272727273e-06,
+      "loss": 3.5077,
+      "step": 4911
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.464e-06,
+      "loss": 3.3685,
+      "step": 4912
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.4632727272727273e-06,
+      "loss": 3.3986,
+      "step": 4913
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.4625454545454546e-06,
+      "loss": 3.7888,
+      "step": 4914
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.461818181818182e-06,
+      "loss": 3.425,
+      "step": 4915
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.4610909090909093e-06,
+      "loss": 3.4872,
+      "step": 4916
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.4603636363636364e-06,
+      "loss": 3.3855,
+      "step": 4917
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.4596363636363636e-06,
+      "loss": 3.3378,
+      "step": 4918
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.458909090909091e-06,
+      "loss": 3.3861,
+      "step": 4919
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.4581818181818184e-06,
+      "loss": 3.5173,
+      "step": 4920
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.4574545454545456e-06,
+      "loss": 3.6016,
+      "step": 4921
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.4567272727272727e-06,
+      "loss": 3.4968,
+      "step": 4922
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.4560000000000001e-06,
+      "loss": 3.383,
+      "step": 4923
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.4552727272727274e-06,
+      "loss": 3.3964,
+      "step": 4924
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.4545454545454546e-06,
+      "loss": 3.5542,
+      "step": 4925
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.453818181818182e-06,
+      "loss": 3.458,
+      "step": 4926
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.4530909090909094e-06,
+      "loss": 3.3328,
+      "step": 4927
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.4523636363636364e-06,
+      "loss": 3.4026,
+      "step": 4928
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.4516363636363637e-06,
+      "loss": 3.5164,
+      "step": 4929
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.4509090909090911e-06,
+      "loss": 3.5502,
+      "step": 4930
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.4501818181818184e-06,
+      "loss": 3.5689,
+      "step": 4931
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.4494545454545456e-06,
+      "loss": 3.3678,
+      "step": 4932
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.4487272727272727e-06,
+      "loss": 3.4264,
+      "step": 4933
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.4480000000000002e-06,
+      "loss": 3.2959,
+      "step": 4934
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.4472727272727274e-06,
+      "loss": 3.5106,
+      "step": 4935
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.4465454545454547e-06,
+      "loss": 3.358,
+      "step": 4936
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.445818181818182e-06,
+      "loss": 3.272,
+      "step": 4937
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.4450909090909094e-06,
+      "loss": 3.0996,
+      "step": 4938
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.4443636363636364e-06,
+      "loss": 3.5457,
+      "step": 4939
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.4436363636363637e-06,
+      "loss": 3.5068,
+      "step": 4940
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.442909090909091e-06,
+      "loss": 3.5623,
+      "step": 4941
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.4421818181818184e-06,
+      "loss": 3.437,
+      "step": 4942
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.4414545454545457e-06,
+      "loss": 3.3938,
+      "step": 4943
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.4407272727272727e-06,
+      "loss": 3.6111,
+      "step": 4944
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.44e-06,
+      "loss": 3.5306,
+      "step": 4945
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.4392727272727274e-06,
+      "loss": 3.3167,
+      "step": 4946
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.4385454545454547e-06,
+      "loss": 3.3215,
+      "step": 4947
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.437818181818182e-06,
+      "loss": 3.5474,
+      "step": 4948
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.437090909090909e-06,
+      "loss": 3.4655,
+      "step": 4949
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.4363636363636365e-06,
+      "loss": 3.5726,
+      "step": 4950
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.4356363636363637e-06,
+      "loss": 3.5097,
+      "step": 4951
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.434909090909091e-06,
+      "loss": 3.3129,
+      "step": 4952
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.4341818181818182e-06,
+      "loss": 3.351,
+      "step": 4953
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.4334545454545457e-06,
+      "loss": 3.3467,
+      "step": 4954
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.4327272727272728e-06,
+      "loss": 3.3445,
+      "step": 4955
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.432e-06,
+      "loss": 3.411,
+      "step": 4956
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.4312727272727273e-06,
+      "loss": 3.2561,
+      "step": 4957
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.4305454545454547e-06,
+      "loss": 3.2259,
+      "step": 4958
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.429818181818182e-06,
+      "loss": 3.142,
+      "step": 4959
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.429090909090909e-06,
+      "loss": 3.3798,
+      "step": 4960
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.4283636363636367e-06,
+      "loss": 3.4894,
+      "step": 4961
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.4276363636363638e-06,
+      "loss": 3.2118,
+      "step": 4962
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.426909090909091e-06,
+      "loss": 3.2251,
+      "step": 4963
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.4261818181818183e-06,
+      "loss": 3.4268,
+      "step": 4964
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.4254545454545457e-06,
+      "loss": 3.5663,
+      "step": 4965
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.424727272727273e-06,
+      "loss": 3.5513,
+      "step": 4966
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.424e-06,
+      "loss": 3.2895,
+      "step": 4967
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.4232727272727273e-06,
+      "loss": 3.2976,
+      "step": 4968
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.4225454545454548e-06,
+      "loss": 3.2187,
+      "step": 4969
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.421818181818182e-06,
+      "loss": 3.4647,
+      "step": 4970
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.4210909090909093e-06,
+      "loss": 3.6464,
+      "step": 4971
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.4203636363636363e-06,
+      "loss": 3.6278,
+      "step": 4972
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.4196363636363638e-06,
+      "loss": 3.4772,
+      "step": 4973
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.418909090909091e-06,
+      "loss": 3.3664,
+      "step": 4974
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.4181818181818183e-06,
+      "loss": 3.3906,
+      "step": 4975
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.4174545454545456e-06,
+      "loss": 3.4138,
+      "step": 4976
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.416727272727273e-06,
+      "loss": 3.177,
+      "step": 4977
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.416e-06,
+      "loss": 3.7338,
+      "step": 4978
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.4152727272727273e-06,
+      "loss": 3.3562,
+      "step": 4979
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.4145454545454546e-06,
+      "loss": 3.406,
+      "step": 4980
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.413818181818182e-06,
+      "loss": 3.3188,
+      "step": 4981
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.4130909090909093e-06,
+      "loss": 3.2854,
+      "step": 4982
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.4123636363636364e-06,
+      "loss": 3.2958,
+      "step": 4983
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.4116363636363636e-06,
+      "loss": 3.1311,
+      "step": 4984
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.410909090909091e-06,
+      "loss": 3.6585,
+      "step": 4985
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.4101818181818183e-06,
+      "loss": 3.3458,
+      "step": 4986
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.1972972972972973,
+      "eval_loss": 3.4123165607452393,
+      "eval_runtime": 144.0861,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 0.645,
+      "step": 4986
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.4094545454545456e-06,
+      "loss": 3.412,
+      "step": 4987
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.4087272727272726e-06,
+      "loss": 3.3351,
+      "step": 4988
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.4080000000000001e-06,
+      "loss": 3.4692,
+      "step": 4989
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.4072727272727274e-06,
+      "loss": 3.2107,
+      "step": 4990
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.4065454545454546e-06,
+      "loss": 3.3303,
+      "step": 4991
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.4058181818181819e-06,
+      "loss": 3.2749,
+      "step": 4992
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.4050909090909094e-06,
+      "loss": 3.3228,
+      "step": 4993
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.4043636363636364e-06,
+      "loss": 3.567,
+      "step": 4994
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.4036363636363637e-06,
+      "loss": 3.3768,
+      "step": 4995
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.402909090909091e-06,
+      "loss": 3.2845,
+      "step": 4996
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.4021818181818184e-06,
+      "loss": 3.5246,
+      "step": 4997
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.4014545454545456e-06,
+      "loss": 3.3859,
+      "step": 4998
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.4007272727272727e-06,
+      "loss": 3.3785,
+      "step": 4999
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.3621,
+      "step": 5000
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.3992727272727274e-06,
+      "loss": 3.5723,
+      "step": 5001
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.3985454545454547e-06,
+      "loss": 3.361,
+      "step": 5002
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.397818181818182e-06,
+      "loss": 3.4218,
+      "step": 5003
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.3970909090909094e-06,
+      "loss": 3.3222,
+      "step": 5004
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.3963636363636364e-06,
+      "loss": 3.4606,
+      "step": 5005
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.3956363636363637e-06,
+      "loss": 3.3838,
+      "step": 5006
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.394909090909091e-06,
+      "loss": 3.2459,
+      "step": 5007
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.3941818181818184e-06,
+      "loss": 3.3012,
+      "step": 5008
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.3934545454545457e-06,
+      "loss": 3.5139,
+      "step": 5009
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.3927272727272727e-06,
+      "loss": 3.5242,
+      "step": 5010
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.392e-06,
+      "loss": 3.6343,
+      "step": 5011
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.3912727272727274e-06,
+      "loss": 3.2414,
+      "step": 5012
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.3905454545454547e-06,
+      "loss": 3.2235,
+      "step": 5013
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.389818181818182e-06,
+      "loss": 3.4316,
+      "step": 5014
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.389090909090909e-06,
+      "loss": 3.1832,
+      "step": 5015
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.3883636363636367e-06,
+      "loss": 3.4239,
+      "step": 5016
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.3876363636363637e-06,
+      "loss": 3.3626,
+      "step": 5017
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.386909090909091e-06,
+      "loss": 3.3808,
+      "step": 5018
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.3861818181818182e-06,
+      "loss": 3.357,
+      "step": 5019
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.3854545454545457e-06,
+      "loss": 3.7329,
+      "step": 5020
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 1.384727272727273e-06,
+      "loss": 3.2679,
+      "step": 5021
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 1.384e-06,
+      "loss": 3.2361,
+      "step": 5022
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 1.3832727272727273e-06,
+      "loss": 3.1629,
+      "step": 5023
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 1.3825454545454547e-06,
+      "loss": 3.2796,
+      "step": 5024
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 1.381818181818182e-06,
+      "loss": 3.2482,
+      "step": 5025
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 1.3810909090909092e-06,
+      "loss": 3.3262,
+      "step": 5026
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 1.3803636363636363e-06,
+      "loss": 3.1769,
+      "step": 5027
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 1.3796363636363638e-06,
+      "loss": 3.2983,
+      "step": 5028
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 1.378909090909091e-06,
+      "loss": 3.4303,
+      "step": 5029
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 1.3781818181818183e-06,
+      "loss": 3.7122,
+      "step": 5030
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 1.3774545454545455e-06,
+      "loss": 3.1625,
+      "step": 5031
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 1.376727272727273e-06,
+      "loss": 3.3098,
+      "step": 5032
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 1.376e-06,
+      "loss": 3.7278,
+      "step": 5033
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 1.3752727272727273e-06,
+      "loss": 3.3586,
+      "step": 5034
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 1.3745454545454548e-06,
+      "loss": 3.4971,
+      "step": 5035
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 1.373818181818182e-06,
+      "loss": 3.6393,
+      "step": 5036
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 1.3730909090909093e-06,
+      "loss": 3.171,
+      "step": 5037
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 1.3723636363636363e-06,
+      "loss": 3.8535,
+      "step": 5038
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 1.3716363636363638e-06,
+      "loss": 3.0381,
+      "step": 5039
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 1.370909090909091e-06,
+      "loss": 3.8137,
+      "step": 5040
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 1.3701818181818183e-06,
+      "loss": 3.2765,
+      "step": 5041
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 1.3694545454545456e-06,
+      "loss": 3.5458,
+      "step": 5042
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 1.368727272727273e-06,
+      "loss": 3.6912,
+      "step": 5043
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 1.368e-06,
+      "loss": 3.4521,
+      "step": 5044
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 1.3672727272727273e-06,
+      "loss": 3.6082,
+      "step": 5045
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 1.3665454545454546e-06,
+      "loss": 3.3956,
+      "step": 5046
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 1.365818181818182e-06,
+      "loss": 3.377,
+      "step": 5047
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 1.3650909090909093e-06,
+      "loss": 3.3031,
+      "step": 5048
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 1.3643636363636364e-06,
+      "loss": 3.2183,
+      "step": 5049
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 1.3636363636363636e-06,
+      "loss": 3.4639,
+      "step": 5050
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 1.362909090909091e-06,
+      "loss": 3.5294,
+      "step": 5051
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 1.3621818181818183e-06,
+      "loss": 3.5743,
+      "step": 5052
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 1.3614545454545456e-06,
+      "loss": 3.4588,
+      "step": 5053
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 1.3607272727272726e-06,
+      "loss": 3.259,
+      "step": 5054
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 3.3672,
+      "step": 5055
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 1.3592727272727274e-06,
+      "loss": 3.3324,
+      "step": 5056
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 1.3585454545454546e-06,
+      "loss": 3.3629,
+      "step": 5057
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 1.3578181818181819e-06,
+      "loss": 3.3973,
+      "step": 5058
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 1.3570909090909093e-06,
+      "loss": 3.4072,
+      "step": 5059
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 1.3563636363636364e-06,
+      "loss": 3.6937,
+      "step": 5060
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 1.3556363636363636e-06,
+      "loss": 3.5826,
+      "step": 5061
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 1.354909090909091e-06,
+      "loss": 3.6606,
+      "step": 5062
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 1.3541818181818184e-06,
+      "loss": 3.3594,
+      "step": 5063
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 1.3534545454545456e-06,
+      "loss": 3.3173,
+      "step": 5064
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 1.3527272727272727e-06,
+      "loss": 3.3709,
+      "step": 5065
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 1.352e-06,
+      "loss": 3.602,
+      "step": 5066
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 1.3512727272727274e-06,
+      "loss": 3.4214,
+      "step": 5067
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 1.3505454545454547e-06,
+      "loss": 3.4,
+      "step": 5068
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 1.349818181818182e-06,
+      "loss": 3.6379,
+      "step": 5069
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 1.3490909090909094e-06,
+      "loss": 3.5654,
+      "step": 5070
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 1.3483636363636366e-06,
+      "loss": 3.108,
+      "step": 5071
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 1.3476363636363637e-06,
+      "loss": 3.6044,
+      "step": 5072
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 1.346909090909091e-06,
+      "loss": 3.663,
+      "step": 5073
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 1.3461818181818184e-06,
+      "loss": 3.8188,
+      "step": 5074
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 1.3454545454545457e-06,
+      "loss": 3.2286,
+      "step": 5075
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 1.344727272727273e-06,
+      "loss": 3.1921,
+      "step": 5076
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 1.344e-06,
+      "loss": 3.3251,
+      "step": 5077
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 1.3432727272727274e-06,
+      "loss": 3.1548,
+      "step": 5078
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 1.3425454545454547e-06,
+      "loss": 3.2871,
+      "step": 5079
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 1.341818181818182e-06,
+      "loss": 3.7471,
+      "step": 5080
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 1.3410909090909092e-06,
+      "loss": 3.3666,
+      "step": 5081
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 1.3403636363636367e-06,
+      "loss": 3.3035,
+      "step": 5082
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 1.3396363636363637e-06,
+      "loss": 3.7639,
+      "step": 5083
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 1.338909090909091e-06,
+      "loss": 3.4629,
+      "step": 5084
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 1.3381818181818182e-06,
+      "loss": 3.277,
+      "step": 5085
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 1.3374545454545457e-06,
+      "loss": 3.2222,
+      "step": 5086
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 1.336727272727273e-06,
+      "loss": 3.6794,
+      "step": 5087
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 1.336e-06,
+      "loss": 3.4357,
+      "step": 5088
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 1.3352727272727273e-06,
+      "loss": 3.466,
+      "step": 5089
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 1.3345454545454547e-06,
+      "loss": 3.3323,
+      "step": 5090
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 1.333818181818182e-06,
+      "loss": 3.4495,
+      "step": 5091
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 1.3330909090909092e-06,
+      "loss": 3.272,
+      "step": 5092
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 1.3323636363636363e-06,
+      "loss": 3.6056,
+      "step": 5093
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 1.3316363636363638e-06,
+      "loss": 3.3845,
+      "step": 5094
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 1.330909090909091e-06,
+      "loss": 3.57,
+      "step": 5095
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 1.3301818181818183e-06,
+      "loss": 3.5051,
+      "step": 5096
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 1.3294545454545455e-06,
+      "loss": 3.6612,
+      "step": 5097
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 1.328727272727273e-06,
+      "loss": 3.4955,
+      "step": 5098
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 1.328e-06,
+      "loss": 3.4774,
+      "step": 5099
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 1.3272727272727273e-06,
+      "loss": 3.5427,
+      "step": 5100
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 1.3265454545454545e-06,
+      "loss": 3.3249,
+      "step": 5101
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 1.325818181818182e-06,
+      "loss": 3.3667,
+      "step": 5102
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 1.3250909090909093e-06,
+      "loss": 3.5111,
+      "step": 5103
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 1.3243636363636363e-06,
+      "loss": 3.3421,
+      "step": 5104
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 1.3236363636363636e-06,
+      "loss": 3.0765,
+      "step": 5105
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 1.322909090909091e-06,
+      "loss": 3.915,
+      "step": 5106
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 1.3221818181818183e-06,
+      "loss": 3.4743,
+      "step": 5107
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 1.3214545454545456e-06,
+      "loss": 3.3409,
+      "step": 5108
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 1.320727272727273e-06,
+      "loss": 3.2438,
+      "step": 5109
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 1.32e-06,
+      "loss": 3.442,
+      "step": 5110
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 1.3192727272727273e-06,
+      "loss": 3.3496,
+      "step": 5111
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 1.3185454545454546e-06,
+      "loss": 3.4595,
+      "step": 5112
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 1.317818181818182e-06,
+      "loss": 3.2003,
+      "step": 5113
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 1.3170909090909093e-06,
+      "loss": 3.46,
+      "step": 5114
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 1.3163636363636364e-06,
+      "loss": 3.2892,
+      "step": 5115
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 1.3156363636363636e-06,
+      "loss": 3.5111,
+      "step": 5116
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 1.314909090909091e-06,
+      "loss": 3.3351,
+      "step": 5117
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 1.3141818181818183e-06,
+      "loss": 3.7521,
+      "step": 5118
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 1.3134545454545456e-06,
+      "loss": 3.2978,
+      "step": 5119
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 1.3127272727272726e-06,
+      "loss": 3.3067,
+      "step": 5120
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 1.3120000000000003e-06,
+      "loss": 3.5578,
+      "step": 5121
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 1.3112727272727274e-06,
+      "loss": 3.3174,
+      "step": 5122
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 1.3105454545454546e-06,
+      "loss": 3.3103,
+      "step": 5123
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 1.3098181818181819e-06,
+      "loss": 3.372,
+      "step": 5124
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 1.3090909090909093e-06,
+      "loss": 3.4722,
+      "step": 5125
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 1.3083636363636366e-06,
+      "loss": 3.537,
+      "step": 5126
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 1.3076363636363636e-06,
+      "loss": 3.8648,
+      "step": 5127
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 1.306909090909091e-06,
+      "loss": 3.2838,
+      "step": 5128
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 1.3061818181818184e-06,
+      "loss": 3.3383,
+      "step": 5129
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 1.3054545454545456e-06,
+      "loss": 3.1465,
+      "step": 5130
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 1.3047272727272729e-06,
+      "loss": 3.6639,
+      "step": 5131
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 1.304e-06,
+      "loss": 3.2721,
+      "step": 5132
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 1.3032727272727274e-06,
+      "loss": 3.5854,
+      "step": 5133
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 1.3025454545454546e-06,
+      "loss": 3.1962,
+      "step": 5134
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 1.301818181818182e-06,
+      "loss": 3.4407,
+      "step": 5135
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 1.3010909090909092e-06,
+      "loss": 3.4932,
+      "step": 5136
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 1.3003636363636366e-06,
+      "loss": 3.8253,
+      "step": 5137
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 1.2996363636363637e-06,
+      "loss": 3.5092,
+      "step": 5138
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 1.298909090909091e-06,
+      "loss": 3.5409,
+      "step": 5139
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 1.2981818181818182e-06,
+      "loss": 3.4244,
+      "step": 5140
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 1.2974545454545457e-06,
+      "loss": 3.6679,
+      "step": 5141
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 1.296727272727273e-06,
+      "loss": 3.3326,
+      "step": 5142
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 1.296e-06,
+      "loss": 3.2943,
+      "step": 5143
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 1.2952727272727274e-06,
+      "loss": 3.513,
+      "step": 5144
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 1.2945454545454547e-06,
+      "loss": 3.2214,
+      "step": 5145
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 1.293818181818182e-06,
+      "loss": 3.4171,
+      "step": 5146
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 1.2930909090909092e-06,
+      "loss": 3.731,
+      "step": 5147
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 1.2923636363636367e-06,
+      "loss": 3.2848,
+      "step": 5148
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 1.2916363636363637e-06,
+      "loss": 3.6465,
+      "step": 5149
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 1.290909090909091e-06,
+      "loss": 3.479,
+      "step": 5150
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 1.2901818181818182e-06,
+      "loss": 3.932,
+      "step": 5151
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 1.2894545454545457e-06,
+      "loss": 3.5495,
+      "step": 5152
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 1.288727272727273e-06,
+      "loss": 3.5,
+      "step": 5153
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 1.288e-06,
+      "loss": 3.5637,
+      "step": 5154
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 1.2872727272727272e-06,
+      "loss": 3.3209,
+      "step": 5155
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 1.2865454545454547e-06,
+      "loss": 3.3377,
+      "step": 5156
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 1.285818181818182e-06,
+      "loss": 3.2658,
+      "step": 5157
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 1.2850909090909092e-06,
+      "loss": 3.3277,
+      "step": 5158
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 1.2843636363636363e-06,
+      "loss": 3.6884,
+      "step": 5159
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 1.2836363636363637e-06,
+      "loss": 3.511,
+      "step": 5160
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 1.282909090909091e-06,
+      "loss": 3.2835,
+      "step": 5161
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 1.2821818181818183e-06,
+      "loss": 3.3425,
+      "step": 5162
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 1.2814545454545455e-06,
+      "loss": 3.6717,
+      "step": 5163
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 1.280727272727273e-06,
+      "loss": 3.5643,
+      "step": 5164
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 1.28e-06,
+      "loss": 3.2435,
+      "step": 5165
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 1.2792727272727273e-06,
+      "loss": 3.527,
+      "step": 5166
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 1.2785454545454545e-06,
+      "loss": 3.2643,
+      "step": 5167
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 1.277818181818182e-06,
+      "loss": 3.36,
+      "step": 5168
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 1.2770909090909093e-06,
+      "loss": 3.2132,
+      "step": 5169
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 1.2763636363636365e-06,
+      "loss": 3.4573,
+      "step": 5170
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 1.2756363636363636e-06,
+      "loss": 3.3968,
+      "step": 5171
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 1.274909090909091e-06,
+      "loss": 3.2779,
+      "step": 5172
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 1.2741818181818183e-06,
+      "loss": 3.5194,
+      "step": 5173
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 1.2734545454545455e-06,
+      "loss": 3.3779,
+      "step": 5174
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 1.2727272727272728e-06,
+      "loss": 3.3045,
+      "step": 5175
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 1.2720000000000003e-06,
+      "loss": 3.634,
+      "step": 5176
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 1.2712727272727273e-06,
+      "loss": 3.582,
+      "step": 5177
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 1.2705454545454546e-06,
+      "loss": 3.1678,
+      "step": 5178
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 1.269818181818182e-06,
+      "loss": 3.756,
+      "step": 5179
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 1.2690909090909093e-06,
+      "loss": 3.1942,
+      "step": 5180
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 1.2683636363636366e-06,
+      "loss": 3.7231,
+      "step": 5181
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 1.2676363636363636e-06,
+      "loss": 3.3172,
+      "step": 5182
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 1.266909090909091e-06,
+      "loss": 3.4529,
+      "step": 5183
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 1.2661818181818183e-06,
+      "loss": 3.319,
+      "step": 5184
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 1.2654545454545456e-06,
+      "loss": 3.2874,
+      "step": 5185
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 1.2647272727272728e-06,
+      "loss": 3.5081,
+      "step": 5186
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 1.2640000000000003e-06,
+      "loss": 3.4985,
+      "step": 5187
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 1.2632727272727274e-06,
+      "loss": 3.3441,
+      "step": 5188
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 1.2625454545454546e-06,
+      "loss": 3.2591,
+      "step": 5189
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 1.2618181818181819e-06,
+      "loss": 3.2256,
+      "step": 5190
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 1.2610909090909093e-06,
+      "loss": 3.3017,
+      "step": 5191
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 1.2603636363636366e-06,
+      "loss": 3.1748,
+      "step": 5192
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 1.2596363636363636e-06,
+      "loss": 3.6058,
+      "step": 5193
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 1.2589090909090909e-06,
+      "loss": 3.1174,
+      "step": 5194
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 1.2581818181818184e-06,
+      "loss": 3.445,
+      "step": 5195
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 1.2574545454545456e-06,
+      "loss": 3.3108,
+      "step": 5196
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 1.2567272727272729e-06,
+      "loss": 3.2428,
+      "step": 5197
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 1.256e-06,
+      "loss": 3.6961,
+      "step": 5198
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 1.2552727272727274e-06,
+      "loss": 3.2268,
+      "step": 5199
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 1.2545454545454546e-06,
+      "loss": 3.1294,
+      "step": 5200
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 1.253818181818182e-06,
+      "loss": 3.3988,
+      "step": 5201
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 1.2530909090909092e-06,
+      "loss": 3.3447,
+      "step": 5202
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 1.2523636363636366e-06,
+      "loss": 3.4651,
+      "step": 5203
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 1.2516363636363637e-06,
+      "loss": 3.4268,
+      "step": 5204
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 1.250909090909091e-06,
+      "loss": 3.4348,
+      "step": 5205
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 1.2501818181818182e-06,
+      "loss": 3.3087,
+      "step": 5206
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 1.2494545454545454e-06,
+      "loss": 3.4573,
+      "step": 5207
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 1.248727272727273e-06,
+      "loss": 3.6424,
+      "step": 5208
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 1.248e-06,
+      "loss": 3.0871,
+      "step": 5209
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 1.2472727272727274e-06,
+      "loss": 3.6622,
+      "step": 5210
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 1.2465454545454547e-06,
+      "loss": 3.4109,
+      "step": 5211
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 1.245818181818182e-06,
+      "loss": 3.3489,
+      "step": 5212
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 1.2450909090909092e-06,
+      "loss": 3.5536,
+      "step": 5213
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 1.2443636363636364e-06,
+      "loss": 3.3432,
+      "step": 5214
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 1.2436363636363637e-06,
+      "loss": 3.4623,
+      "step": 5215
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 1.242909090909091e-06,
+      "loss": 3.5545,
+      "step": 5216
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 1.2421818181818184e-06,
+      "loss": 3.7993,
+      "step": 5217
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 1.2414545454545455e-06,
+      "loss": 3.2515,
+      "step": 5218
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 1.240727272727273e-06,
+      "loss": 3.2341,
+      "step": 5219
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 3.3997,
+      "step": 5220
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 1.2392727272727275e-06,
+      "loss": 3.1671,
+      "step": 5221
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 1.2385454545454547e-06,
+      "loss": 3.3383,
+      "step": 5222
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 1.237818181818182e-06,
+      "loss": 3.4108,
+      "step": 5223
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 1.2370909090909092e-06,
+      "loss": 3.2823,
+      "step": 5224
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 1.2363636363636365e-06,
+      "loss": 3.3873,
+      "step": 5225
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 1.2356363636363637e-06,
+      "loss": 3.2359,
+      "step": 5226
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 1.234909090909091e-06,
+      "loss": 3.3853,
+      "step": 5227
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 1.2341818181818183e-06,
+      "loss": 3.3909,
+      "step": 5228
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 1.2334545454545455e-06,
+      "loss": 3.4344,
+      "step": 5229
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 1.2327272727272728e-06,
+      "loss": 3.3288,
+      "step": 5230
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 1.2320000000000002e-06,
+      "loss": 3.402,
+      "step": 5231
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 1.2312727272727273e-06,
+      "loss": 3.3167,
+      "step": 5232
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 1.2305454545454547e-06,
+      "loss": 3.1499,
+      "step": 5233
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 1.2298181818181818e-06,
+      "loss": 3.3944,
+      "step": 5234
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 1.2290909090909093e-06,
+      "loss": 3.2641,
+      "step": 5235
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 1.2283636363636365e-06,
+      "loss": 3.4762,
+      "step": 5236
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 1.2276363636363638e-06,
+      "loss": 3.3075,
+      "step": 5237
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 1.226909090909091e-06,
+      "loss": 3.5089,
+      "step": 5238
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 1.2261818181818183e-06,
+      "loss": 3.2459,
+      "step": 5239
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 1.2254545454545455e-06,
+      "loss": 3.3057,
+      "step": 5240
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 1.2247272727272728e-06,
+      "loss": 3.2656,
+      "step": 5241
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 1.224e-06,
+      "loss": 3.4708,
+      "step": 5242
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 1.2232727272727273e-06,
+      "loss": 3.224,
+      "step": 5243
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 1.2225454545454546e-06,
+      "loss": 3.5672,
+      "step": 5244
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 1.221818181818182e-06,
+      "loss": 3.4337,
+      "step": 5245
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 1.221090909090909e-06,
+      "loss": 3.4062,
+      "step": 5246
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 1.2203636363636365e-06,
+      "loss": 3.3366,
+      "step": 5247
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 1.2196363636363636e-06,
+      "loss": 3.2968,
+      "step": 5248
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 1.218909090909091e-06,
+      "loss": 3.5939,
+      "step": 5249
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 1.2181818181818183e-06,
+      "loss": 3.2221,
+      "step": 5250
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 1.2174545454545456e-06,
+      "loss": 3.4533,
+      "step": 5251
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 1.2167272727272728e-06,
+      "loss": 3.7527,
+      "step": 5252
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 1.216e-06,
+      "loss": 3.4608,
+      "step": 5253
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 1.2152727272727273e-06,
+      "loss": 3.3478,
+      "step": 5254
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 1.2145454545454546e-06,
+      "loss": 3.2818,
+      "step": 5255
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 1.213818181818182e-06,
+      "loss": 3.3474,
+      "step": 5256
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 1.2130909090909091e-06,
+      "loss": 3.4362,
+      "step": 5257
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 1.2123636363636366e-06,
+      "loss": 3.6549,
+      "step": 5258
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 1.2116363636363636e-06,
+      "loss": 3.3939,
+      "step": 5259
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 1.210909090909091e-06,
+      "loss": 3.3687,
+      "step": 5260
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 1.2101818181818184e-06,
+      "loss": 3.4449,
+      "step": 5261
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 1.2094545454545456e-06,
+      "loss": 3.5555,
+      "step": 5262
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 1.2087272727272729e-06,
+      "loss": 3.3588,
+      "step": 5263
+    },
+    {
+      "epoch": 19.0,
+      "eval_accuracy": 0.21081081081081082,
+      "eval_loss": 3.4174647331237793,
+      "eval_runtime": 143.2416,
+      "eval_samples_per_second": 2.583,
+      "eval_steps_per_second": 0.649,
+      "step": 5263
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 1.2080000000000001e-06,
+      "loss": 3.3544,
+      "step": 5264
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 1.2072727272727274e-06,
+      "loss": 3.607,
+      "step": 5265
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 1.2065454545454546e-06,
+      "loss": 3.3925,
+      "step": 5266
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 1.2058181818181819e-06,
+      "loss": 3.3337,
+      "step": 5267
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 1.2050909090909091e-06,
+      "loss": 3.4865,
+      "step": 5268
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 1.2043636363636364e-06,
+      "loss": 3.3972,
+      "step": 5269
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.2036363636363639e-06,
+      "loss": 3.4579,
+      "step": 5270
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.202909090909091e-06,
+      "loss": 3.5969,
+      "step": 5271
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.2021818181818184e-06,
+      "loss": 3.2905,
+      "step": 5272
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.2014545454545454e-06,
+      "loss": 3.3693,
+      "step": 5273
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.200727272727273e-06,
+      "loss": 3.2937,
+      "step": 5274
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 3.5435,
+      "step": 5275
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 1.1992727272727274e-06,
+      "loss": 3.4311,
+      "step": 5276
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 1.1985454545454547e-06,
+      "loss": 3.2291,
+      "step": 5277
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 1.197818181818182e-06,
+      "loss": 3.5143,
+      "step": 5278
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 1.1970909090909092e-06,
+      "loss": 3.6713,
+      "step": 5279
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 1.1963636363636364e-06,
+      "loss": 3.4134,
+      "step": 5280
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 1.1956363636363637e-06,
+      "loss": 3.2769,
+      "step": 5281
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 1.194909090909091e-06,
+      "loss": 3.3037,
+      "step": 5282
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 1.1941818181818182e-06,
+      "loss": 3.6331,
+      "step": 5283
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 1.1934545454545455e-06,
+      "loss": 3.7996,
+      "step": 5284
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 1.1927272727272727e-06,
+      "loss": 3.4138,
+      "step": 5285
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 1.1920000000000002e-06,
+      "loss": 3.2258,
+      "step": 5286
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 1.1912727272727272e-06,
+      "loss": 3.627,
+      "step": 5287
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 1.1905454545454547e-06,
+      "loss": 3.3892,
+      "step": 5288
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 1.189818181818182e-06,
+      "loss": 3.693,
+      "step": 5289
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 1.1890909090909092e-06,
+      "loss": 3.5475,
+      "step": 5290
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 1.1883636363636365e-06,
+      "loss": 3.4166,
+      "step": 5291
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 1.1876363636363637e-06,
+      "loss": 3.5491,
+      "step": 5292
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 1.186909090909091e-06,
+      "loss": 3.7367,
+      "step": 5293
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 1.1861818181818182e-06,
+      "loss": 3.5884,
+      "step": 5294
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 1.1854545454545457e-06,
+      "loss": 3.1691,
+      "step": 5295
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 1.1847272727272728e-06,
+      "loss": 3.5323,
+      "step": 5296
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 1.1840000000000002e-06,
+      "loss": 3.3433,
+      "step": 5297
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 1.1832727272727273e-06,
+      "loss": 3.3159,
+      "step": 5298
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 1.1825454545454547e-06,
+      "loss": 3.3767,
+      "step": 5299
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 1.181818181818182e-06,
+      "loss": 3.3765,
+      "step": 5300
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 1.1810909090909093e-06,
+      "loss": 3.1675,
+      "step": 5301
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 1.1803636363636365e-06,
+      "loss": 3.2681,
+      "step": 5302
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 1.1796363636363638e-06,
+      "loss": 3.2149,
+      "step": 5303
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 1.178909090909091e-06,
+      "loss": 3.3081,
+      "step": 5304
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 1.1781818181818183e-06,
+      "loss": 3.3415,
+      "step": 5305
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 1.1774545454545455e-06,
+      "loss": 3.4689,
+      "step": 5306
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 1.1767272727272728e-06,
+      "loss": 3.3082,
+      "step": 5307
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 1.176e-06,
+      "loss": 3.0694,
+      "step": 5308
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 1.1752727272727273e-06,
+      "loss": 3.2767,
+      "step": 5309
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 1.1745454545454546e-06,
+      "loss": 3.2943,
+      "step": 5310
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 1.173818181818182e-06,
+      "loss": 3.3225,
+      "step": 5311
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 1.173090909090909e-06,
+      "loss": 3.3684,
+      "step": 5312
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 1.1723636363636365e-06,
+      "loss": 3.2982,
+      "step": 5313
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 1.1716363636363636e-06,
+      "loss": 3.1441,
+      "step": 5314
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 1.170909090909091e-06,
+      "loss": 3.4566,
+      "step": 5315
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 1.1701818181818183e-06,
+      "loss": 3.4856,
+      "step": 5316
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 1.1694545454545456e-06,
+      "loss": 3.4108,
+      "step": 5317
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 1.1687272727272728e-06,
+      "loss": 3.3985,
+      "step": 5318
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 1.168e-06,
+      "loss": 3.3354,
+      "step": 5319
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 1.1672727272727273e-06,
+      "loss": 3.3398,
+      "step": 5320
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 1.1665454545454546e-06,
+      "loss": 3.3318,
+      "step": 5321
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 1.1658181818181819e-06,
+      "loss": 3.3484,
+      "step": 5322
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 1.1650909090909091e-06,
+      "loss": 3.2105,
+      "step": 5323
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 1.1643636363636364e-06,
+      "loss": 3.4962,
+      "step": 5324
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 1.1636363636363638e-06,
+      "loss": 3.5112,
+      "step": 5325
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 1.162909090909091e-06,
+      "loss": 3.2747,
+      "step": 5326
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 1.1621818181818183e-06,
+      "loss": 3.3416,
+      "step": 5327
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 1.1614545454545456e-06,
+      "loss": 3.4395,
+      "step": 5328
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 1.1607272727272729e-06,
+      "loss": 3.5163,
+      "step": 5329
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 3.6527,
+      "step": 5330
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 1.1592727272727274e-06,
+      "loss": 3.5583,
+      "step": 5331
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 1.1585454545454546e-06,
+      "loss": 3.4764,
+      "step": 5332
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 1.1578181818181819e-06,
+      "loss": 3.4002,
+      "step": 5333
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 1.1570909090909091e-06,
+      "loss": 3.3668,
+      "step": 5334
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 1.1563636363636364e-06,
+      "loss": 3.7407,
+      "step": 5335
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 1.1556363636363639e-06,
+      "loss": 3.402,
+      "step": 5336
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 1.154909090909091e-06,
+      "loss": 3.3987,
+      "step": 5337
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 1.1541818181818184e-06,
+      "loss": 3.4008,
+      "step": 5338
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 1.1534545454545454e-06,
+      "loss": 3.3802,
+      "step": 5339
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 1.152727272727273e-06,
+      "loss": 3.4152,
+      "step": 5340
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 1.1520000000000002e-06,
+      "loss": 3.2549,
+      "step": 5341
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 1.1512727272727274e-06,
+      "loss": 3.3066,
+      "step": 5342
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 1.1505454545454547e-06,
+      "loss": 3.8353,
+      "step": 5343
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 1.149818181818182e-06,
+      "loss": 3.4044,
+      "step": 5344
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 1.1490909090909092e-06,
+      "loss": 3.4497,
+      "step": 5345
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 1.1483636363636364e-06,
+      "loss": 3.217,
+      "step": 5346
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 1.1476363636363637e-06,
+      "loss": 3.246,
+      "step": 5347
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 1.146909090909091e-06,
+      "loss": 3.3066,
+      "step": 5348
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 1.1461818181818182e-06,
+      "loss": 3.3817,
+      "step": 5349
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 1.1454545454545457e-06,
+      "loss": 3.505,
+      "step": 5350
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 1.1447272727272727e-06,
+      "loss": 3.3049,
+      "step": 5351
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 1.1440000000000002e-06,
+      "loss": 3.3399,
+      "step": 5352
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 1.1432727272727272e-06,
+      "loss": 3.348,
+      "step": 5353
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 1.1425454545454547e-06,
+      "loss": 3.0759,
+      "step": 5354
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 1.141818181818182e-06,
+      "loss": 3.5073,
+      "step": 5355
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 1.1410909090909092e-06,
+      "loss": 3.7451,
+      "step": 5356
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 1.1403636363636365e-06,
+      "loss": 3.4394,
+      "step": 5357
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 1.1396363636363637e-06,
+      "loss": 3.7544,
+      "step": 5358
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.138909090909091e-06,
+      "loss": 3.4341,
+      "step": 5359
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.1381818181818182e-06,
+      "loss": 3.5929,
+      "step": 5360
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.1374545454545457e-06,
+      "loss": 3.6058,
+      "step": 5361
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 1.1367272727272728e-06,
+      "loss": 3.3763,
+      "step": 5362
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 1.1360000000000002e-06,
+      "loss": 3.3292,
+      "step": 5363
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 1.1352727272727273e-06,
+      "loss": 3.5004,
+      "step": 5364
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 1.1345454545454547e-06,
+      "loss": 3.256,
+      "step": 5365
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 1.133818181818182e-06,
+      "loss": 3.4825,
+      "step": 5366
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 1.1330909090909092e-06,
+      "loss": 3.3498,
+      "step": 5367
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 1.1323636363636365e-06,
+      "loss": 3.5977,
+      "step": 5368
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 1.1316363636363638e-06,
+      "loss": 3.2525,
+      "step": 5369
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 1.130909090909091e-06,
+      "loss": 3.3253,
+      "step": 5370
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 1.1301818181818183e-06,
+      "loss": 3.5823,
+      "step": 5371
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 1.1294545454545455e-06,
+      "loss": 3.6742,
+      "step": 5372
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 1.1287272727272728e-06,
+      "loss": 3.436,
+      "step": 5373
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 1.128e-06,
+      "loss": 3.3607,
+      "step": 5374
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 1.1272727272727275e-06,
+      "loss": 3.3667,
+      "step": 5375
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 1.1265454545454546e-06,
+      "loss": 3.5685,
+      "step": 5376
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 1.125818181818182e-06,
+      "loss": 3.4623,
+      "step": 5377
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 1.125090909090909e-06,
+      "loss": 3.3574,
+      "step": 5378
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 1.1243636363636365e-06,
+      "loss": 3.2038,
+      "step": 5379
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 1.1236363636363638e-06,
+      "loss": 3.5238,
+      "step": 5380
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 1.122909090909091e-06,
+      "loss": 3.3237,
+      "step": 5381
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 1.1221818181818183e-06,
+      "loss": 3.3159,
+      "step": 5382
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 1.1214545454545456e-06,
+      "loss": 3.5273,
+      "step": 5383
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.1207272727272728e-06,
+      "loss": 3.3932,
+      "step": 5384
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.12e-06,
+      "loss": 3.4289,
+      "step": 5385
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.1192727272727273e-06,
+      "loss": 3.2095,
+      "step": 5386
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 1.1185454545454546e-06,
+      "loss": 3.3148,
+      "step": 5387
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 1.1178181818181818e-06,
+      "loss": 3.4287,
+      "step": 5388
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 1.117090909090909e-06,
+      "loss": 3.353,
+      "step": 5389
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.1163636363636364e-06,
+      "loss": 3.1227,
+      "step": 5390
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.1156363636363638e-06,
+      "loss": 3.5433,
+      "step": 5391
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.1149090909090909e-06,
+      "loss": 3.3476,
+      "step": 5392
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.1141818181818183e-06,
+      "loss": 3.5283,
+      "step": 5393
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.1134545454545454e-06,
+      "loss": 3.1756,
+      "step": 5394
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 1.1127272727272729e-06,
+      "loss": 3.5498,
+      "step": 5395
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 1.1120000000000001e-06,
+      "loss": 3.4705,
+      "step": 5396
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 1.1112727272727274e-06,
+      "loss": 3.36,
+      "step": 5397
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 1.1105454545454546e-06,
+      "loss": 3.3229,
+      "step": 5398
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 1.1098181818181819e-06,
+      "loss": 3.5143,
+      "step": 5399
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 1.1090909090909093e-06,
+      "loss": 3.5871,
+      "step": 5400
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 1.1083636363636364e-06,
+      "loss": 3.0622,
+      "step": 5401
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 1.1076363636363639e-06,
+      "loss": 3.4304,
+      "step": 5402
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 1.106909090909091e-06,
+      "loss": 3.5794,
+      "step": 5403
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 1.1061818181818184e-06,
+      "loss": 3.3592,
+      "step": 5404
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 1.1054545454545456e-06,
+      "loss": 3.4054,
+      "step": 5405
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 1.1047272727272729e-06,
+      "loss": 3.3825,
+      "step": 5406
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 1.1040000000000001e-06,
+      "loss": 3.1591,
+      "step": 5407
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 1.1032727272727274e-06,
+      "loss": 3.5184,
+      "step": 5408
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 1.1025454545454547e-06,
+      "loss": 3.5357,
+      "step": 5409
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 1.101818181818182e-06,
+      "loss": 3.5058,
+      "step": 5410
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 1.1010909090909092e-06,
+      "loss": 3.6391,
+      "step": 5411
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 1.1003636363636364e-06,
+      "loss": 3.4066,
+      "step": 5412
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 1.0996363636363637e-06,
+      "loss": 3.4263,
+      "step": 5413
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 1.098909090909091e-06,
+      "loss": 3.4539,
+      "step": 5414
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 1.0981818181818182e-06,
+      "loss": 3.2682,
+      "step": 5415
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 1.0974545454545457e-06,
+      "loss": 3.2091,
+      "step": 5416
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 1.0967272727272727e-06,
+      "loss": 3.3903,
+      "step": 5417
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 1.0960000000000002e-06,
+      "loss": 3.6207,
+      "step": 5418
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 1.0952727272727272e-06,
+      "loss": 3.3353,
+      "step": 5419
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 1.0945454545454547e-06,
+      "loss": 3.7242,
+      "step": 5420
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 1.093818181818182e-06,
+      "loss": 3.5674,
+      "step": 5421
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 1.0930909090909092e-06,
+      "loss": 3.4662,
+      "step": 5422
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 1.0923636363636365e-06,
+      "loss": 3.4693,
+      "step": 5423
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 1.0916363636363637e-06,
+      "loss": 3.2408,
+      "step": 5424
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 1.090909090909091e-06,
+      "loss": 3.2564,
+      "step": 5425
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 1.0901818181818182e-06,
+      "loss": 3.271,
+      "step": 5426
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 1.0894545454545455e-06,
+      "loss": 3.5334,
+      "step": 5427
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 1.0887272727272727e-06,
+      "loss": 3.3814,
+      "step": 5428
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 1.088e-06,
+      "loss": 3.356,
+      "step": 5429
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 1.0872727272727275e-06,
+      "loss": 3.3383,
+      "step": 5430
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 1.0865454545454545e-06,
+      "loss": 3.5331,
+      "step": 5431
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 1.085818181818182e-06,
+      "loss": 3.8314,
+      "step": 5432
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 1.085090909090909e-06,
+      "loss": 3.5897,
+      "step": 5433
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 1.0843636363636365e-06,
+      "loss": 3.4682,
+      "step": 5434
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 1.0836363636363638e-06,
+      "loss": 3.4302,
+      "step": 5435
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 1.082909090909091e-06,
+      "loss": 3.5942,
+      "step": 5436
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 1.0821818181818183e-06,
+      "loss": 3.5051,
+      "step": 5437
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 1.0814545454545455e-06,
+      "loss": 3.5542,
+      "step": 5438
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.0807272727272728e-06,
+      "loss": 3.3998,
+      "step": 5439
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.08e-06,
+      "loss": 3.301,
+      "step": 5440
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.0792727272727275e-06,
+      "loss": 3.386,
+      "step": 5441
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.0785454545454545e-06,
+      "loss": 3.6155,
+      "step": 5442
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.077818181818182e-06,
+      "loss": 3.3133,
+      "step": 5443
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.077090909090909e-06,
+      "loss": 3.2579,
+      "step": 5444
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.0763636363636365e-06,
+      "loss": 3.3245,
+      "step": 5445
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.0756363636363638e-06,
+      "loss": 3.3709,
+      "step": 5446
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.074909090909091e-06,
+      "loss": 3.2526,
+      "step": 5447
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.0741818181818183e-06,
+      "loss": 3.3875,
+      "step": 5448
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.0734545454545456e-06,
+      "loss": 3.0773,
+      "step": 5449
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.0727272727272728e-06,
+      "loss": 3.2854,
+      "step": 5450
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.072e-06,
+      "loss": 3.0834,
+      "step": 5451
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.0712727272727273e-06,
+      "loss": 3.4243,
+      "step": 5452
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.0705454545454546e-06,
+      "loss": 3.5472,
+      "step": 5453
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.0698181818181818e-06,
+      "loss": 3.6123,
+      "step": 5454
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.0690909090909093e-06,
+      "loss": 3.4614,
+      "step": 5455
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.0683636363636364e-06,
+      "loss": 3.5282,
+      "step": 5456
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.0676363636363638e-06,
+      "loss": 3.3223,
+      "step": 5457
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.0669090909090909e-06,
+      "loss": 3.3594,
+      "step": 5458
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.0661818181818183e-06,
+      "loss": 3.253,
+      "step": 5459
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.0654545454545456e-06,
+      "loss": 3.6354,
+      "step": 5460
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.0647272727272728e-06,
+      "loss": 3.1994,
+      "step": 5461
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.064e-06,
+      "loss": 3.1677,
+      "step": 5462
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.0632727272727274e-06,
+      "loss": 3.3127,
+      "step": 5463
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.0625454545454546e-06,
+      "loss": 3.1474,
+      "step": 5464
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.0618181818181819e-06,
+      "loss": 3.3822,
+      "step": 5465
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.0610909090909091e-06,
+      "loss": 3.4954,
+      "step": 5466
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.0603636363636364e-06,
+      "loss": 3.2426,
+      "step": 5467
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.0596363636363636e-06,
+      "loss": 3.2866,
+      "step": 5468
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.058909090909091e-06,
+      "loss": 3.6492,
+      "step": 5469
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.0581818181818184e-06,
+      "loss": 3.3674,
+      "step": 5470
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.0574545454545456e-06,
+      "loss": 3.6752,
+      "step": 5471
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.0567272727272729e-06,
+      "loss": 3.4003,
+      "step": 5472
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.0560000000000001e-06,
+      "loss": 3.4458,
+      "step": 5473
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.0552727272727274e-06,
+      "loss": 3.1883,
+      "step": 5474
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.0545454545454547e-06,
+      "loss": 3.3357,
+      "step": 5475
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.053818181818182e-06,
+      "loss": 3.6382,
+      "step": 5476
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.0530909090909092e-06,
+      "loss": 3.244,
+      "step": 5477
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.0523636363636364e-06,
+      "loss": 3.1743,
+      "step": 5478
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.0516363636363637e-06,
+      "loss": 3.1813,
+      "step": 5479
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.0509090909090911e-06,
+      "loss": 3.5343,
+      "step": 5480
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.0501818181818182e-06,
+      "loss": 3.2996,
+      "step": 5481
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.0494545454545457e-06,
+      "loss": 3.3019,
+      "step": 5482
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.0487272727272727e-06,
+      "loss": 3.2519,
+      "step": 5483
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.0480000000000002e-06,
+      "loss": 3.393,
+      "step": 5484
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.0472727272727274e-06,
+      "loss": 3.411,
+      "step": 5485
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.0465454545454547e-06,
+      "loss": 3.2368,
+      "step": 5486
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.045818181818182e-06,
+      "loss": 3.3935,
+      "step": 5487
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.0450909090909092e-06,
+      "loss": 3.3795,
+      "step": 5488
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.0443636363636365e-06,
+      "loss": 3.3587,
+      "step": 5489
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.0436363636363637e-06,
+      "loss": 3.3247,
+      "step": 5490
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.042909090909091e-06,
+      "loss": 3.3641,
+      "step": 5491
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 1.0421818181818182e-06,
+      "loss": 3.3481,
+      "step": 5492
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 1.0414545454545455e-06,
+      "loss": 3.8769,
+      "step": 5493
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 1.0407272727272727e-06,
+      "loss": 3.325,
+      "step": 5494
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 1.04e-06,
+      "loss": 3.2733,
+      "step": 5495
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 1.0392727272727275e-06,
+      "loss": 3.5845,
+      "step": 5496
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 1.0385454545454545e-06,
+      "loss": 3.3392,
+      "step": 5497
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 1.037818181818182e-06,
+      "loss": 3.4549,
+      "step": 5498
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 1.037090909090909e-06,
+      "loss": 3.51,
+      "step": 5499
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 1.0363636363636365e-06,
+      "loss": 3.7554,
+      "step": 5500
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 1.0356363636363637e-06,
+      "loss": 3.3082,
+      "step": 5501
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 1.034909090909091e-06,
+      "loss": 3.3925,
+      "step": 5502
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 1.0341818181818183e-06,
+      "loss": 3.5731,
+      "step": 5503
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 1.0334545454545455e-06,
+      "loss": 3.2174,
+      "step": 5504
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 1.0327272727272728e-06,
+      "loss": 3.4613,
+      "step": 5505
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 1.032e-06,
+      "loss": 3.3237,
+      "step": 5506
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 1.0312727272727275e-06,
+      "loss": 3.6523,
+      "step": 5507
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 1.0305454545454545e-06,
+      "loss": 3.1847,
+      "step": 5508
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 1.029818181818182e-06,
+      "loss": 3.6947,
+      "step": 5509
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 1.0290909090909093e-06,
+      "loss": 3.7483,
+      "step": 5510
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 1.0283636363636365e-06,
+      "loss": 3.529,
+      "step": 5511
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 1.0276363636363638e-06,
+      "loss": 3.4103,
+      "step": 5512
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 1.026909090909091e-06,
+      "loss": 3.4465,
+      "step": 5513
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 1.0261818181818183e-06,
+      "loss": 3.487,
+      "step": 5514
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 1.0254545454545456e-06,
+      "loss": 3.3922,
+      "step": 5515
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 1.0247272727272728e-06,
+      "loss": 3.6129,
+      "step": 5516
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 1.024e-06,
+      "loss": 3.4742,
+      "step": 5517
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 1.0232727272727273e-06,
+      "loss": 3.4141,
+      "step": 5518
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 1.0225454545454546e-06,
+      "loss": 3.5908,
+      "step": 5519
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 1.0218181818181818e-06,
+      "loss": 3.2353,
+      "step": 5520
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 1.0210909090909093e-06,
+      "loss": 3.4467,
+      "step": 5521
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 1.0203636363636363e-06,
+      "loss": 3.0493,
+      "step": 5522
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 1.0196363636363638e-06,
+      "loss": 3.2527,
+      "step": 5523
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 1.0189090909090909e-06,
+      "loss": 3.1891,
+      "step": 5524
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 1.0181818181818183e-06,
+      "loss": 3.4989,
+      "step": 5525
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 1.0174545454545456e-06,
+      "loss": 3.2206,
+      "step": 5526
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 1.0167272727272728e-06,
+      "loss": 3.5316,
+      "step": 5527
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 1.016e-06,
+      "loss": 3.3688,
+      "step": 5528
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 1.0152727272727274e-06,
+      "loss": 3.5389,
+      "step": 5529
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 1.0145454545454546e-06,
+      "loss": 3.212,
+      "step": 5530
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 1.0138181818181819e-06,
+      "loss": 3.3226,
+      "step": 5531
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 1.0130909090909091e-06,
+      "loss": 3.4477,
+      "step": 5532
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 1.0123636363636364e-06,
+      "loss": 3.578,
+      "step": 5533
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.0116363636363636e-06,
+      "loss": 3.2611,
+      "step": 5534
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.0109090909090911e-06,
+      "loss": 3.2462,
+      "step": 5535
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 1.0101818181818182e-06,
+      "loss": 3.502,
+      "step": 5536
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 1.0094545454545456e-06,
+      "loss": 3.3125,
+      "step": 5537
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 1.0087272727272727e-06,
+      "loss": 3.725,
+      "step": 5538
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.0080000000000001e-06,
+      "loss": 3.3451,
+      "step": 5539
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.0072727272727274e-06,
+      "loss": 3.4145,
+      "step": 5540
+    },
+    {
+      "epoch": 20.0,
+      "eval_accuracy": 0.20270270270270271,
+      "eval_loss": 3.4048352241516113,
+      "eval_runtime": 144.5448,
+      "eval_samples_per_second": 2.56,
+      "eval_steps_per_second": 0.643,
+      "step": 5540
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.0065454545454546e-06,
+      "loss": 3.6323,
+      "step": 5541
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.005818181818182e-06,
+      "loss": 3.6997,
+      "step": 5542
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.0050909090909092e-06,
+      "loss": 3.1405,
+      "step": 5543
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.0043636363636364e-06,
+      "loss": 3.3931,
+      "step": 5544
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.0036363636363637e-06,
+      "loss": 3.5387,
+      "step": 5545
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.0029090909090911e-06,
+      "loss": 3.381,
+      "step": 5546
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.0021818181818182e-06,
+      "loss": 3.4071,
+      "step": 5547
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.0014545454545457e-06,
+      "loss": 3.1399,
+      "step": 5548
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.0007272727272727e-06,
+      "loss": 3.4195,
+      "step": 5549
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.4757,
+      "step": 5550
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 9.992727272727274e-07,
+      "loss": 3.3168,
+      "step": 5551
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 9.985454545454547e-07,
+      "loss": 3.3936,
+      "step": 5552
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 9.97818181818182e-07,
+      "loss": 3.1907,
+      "step": 5553
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 9.970909090909092e-07,
+      "loss": 3.2665,
+      "step": 5554
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 9.963636363636364e-07,
+      "loss": 3.095,
+      "step": 5555
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 9.956363636363637e-07,
+      "loss": 3.2804,
+      "step": 5556
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 9.94909090909091e-07,
+      "loss": 3.6954,
+      "step": 5557
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 9.941818181818182e-07,
+      "loss": 3.4343,
+      "step": 5558
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 9.934545454545455e-07,
+      "loss": 3.5469,
+      "step": 5559
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 9.92727272727273e-07,
+      "loss": 3.3605,
+      "step": 5560
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 9.92e-07,
+      "loss": 3.2799,
+      "step": 5561
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 9.912727272727275e-07,
+      "loss": 3.5219,
+      "step": 5562
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 9.905454545454545e-07,
+      "loss": 3.2326,
+      "step": 5563
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 9.89818181818182e-07,
+      "loss": 3.4832,
+      "step": 5564
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 9.890909090909092e-07,
+      "loss": 3.28,
+      "step": 5565
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 9.883636363636365e-07,
+      "loss": 3.395,
+      "step": 5566
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 9.876363636363637e-07,
+      "loss": 3.3736,
+      "step": 5567
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 9.86909090909091e-07,
+      "loss": 3.4216,
+      "step": 5568
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 9.861818181818183e-07,
+      "loss": 3.2556,
+      "step": 5569
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 9.854545454545455e-07,
+      "loss": 3.5952,
+      "step": 5570
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 9.847272727272728e-07,
+      "loss": 3.2826,
+      "step": 5571
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 9.84e-07,
+      "loss": 3.3686,
+      "step": 5572
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 9.832727272727273e-07,
+      "loss": 3.3595,
+      "step": 5573
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 9.825454545454545e-07,
+      "loss": 3.231,
+      "step": 5574
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 9.818181818181818e-07,
+      "loss": 3.5868,
+      "step": 5575
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 9.810909090909093e-07,
+      "loss": 3.325,
+      "step": 5576
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 9.803636363636363e-07,
+      "loss": 3.5424,
+      "step": 5577
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 9.796363636363638e-07,
+      "loss": 3.3755,
+      "step": 5578
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 9.789090909090908e-07,
+      "loss": 3.5572,
+      "step": 5579
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 9.781818181818183e-07,
+      "loss": 3.2167,
+      "step": 5580
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 9.774545454545455e-07,
+      "loss": 3.3839,
+      "step": 5581
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 9.767272727272728e-07,
+      "loss": 3.1308,
+      "step": 5582
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 9.76e-07,
+      "loss": 3.5494,
+      "step": 5583
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 9.752727272727273e-07,
+      "loss": 3.4219,
+      "step": 5584
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 9.745454545454548e-07,
+      "loss": 3.5421,
+      "step": 5585
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 9.738181818181818e-07,
+      "loss": 3.6821,
+      "step": 5586
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 9.730909090909093e-07,
+      "loss": 3.1766,
+      "step": 5587
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 9.723636363636363e-07,
+      "loss": 3.197,
+      "step": 5588
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 9.716363636363638e-07,
+      "loss": 3.4164,
+      "step": 5589
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 9.70909090909091e-07,
+      "loss": 3.5707,
+      "step": 5590
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 9.701818181818183e-07,
+      "loss": 3.4027,
+      "step": 5591
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 9.694545454545456e-07,
+      "loss": 3.6226,
+      "step": 5592
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 9.687272727272728e-07,
+      "loss": 3.759,
+      "step": 5593
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 9.68e-07,
+      "loss": 3.2983,
+      "step": 5594
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 9.672727272727273e-07,
+      "loss": 3.2805,
+      "step": 5595
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 9.665454545454546e-07,
+      "loss": 3.6281,
+      "step": 5596
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 9.658181818181819e-07,
+      "loss": 3.406,
+      "step": 5597
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 9.650909090909091e-07,
+      "loss": 3.2667,
+      "step": 5598
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 9.643636363636364e-07,
+      "loss": 3.3519,
+      "step": 5599
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 9.636363636363636e-07,
+      "loss": 3.2406,
+      "step": 5600
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 9.62909090909091e-07,
+      "loss": 3.3502,
+      "step": 5601
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 9.621818181818181e-07,
+      "loss": 3.3507,
+      "step": 5602
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 9.614545454545456e-07,
+      "loss": 3.5356,
+      "step": 5603
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 9.607272727272727e-07,
+      "loss": 3.3034,
+      "step": 5604
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 3.3189,
+      "step": 5605
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 9.592727272727274e-07,
+      "loss": 3.5182,
+      "step": 5606
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 9.585454545454546e-07,
+      "loss": 3.4201,
+      "step": 5607
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 9.57818181818182e-07,
+      "loss": 3.213,
+      "step": 5608
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 9.570909090909092e-07,
+      "loss": 3.6211,
+      "step": 5609
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 9.563636363636364e-07,
+      "loss": 3.8035,
+      "step": 5610
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 9.556363636363637e-07,
+      "loss": 3.2266,
+      "step": 5611
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 9.54909090909091e-07,
+      "loss": 3.3014,
+      "step": 5612
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 9.541818181818182e-07,
+      "loss": 3.2019,
+      "step": 5613
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 9.534545454545454e-07,
+      "loss": 3.3786,
+      "step": 5614
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 9.527272727272728e-07,
+      "loss": 3.3602,
+      "step": 5615
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 9.520000000000002e-07,
+      "loss": 3.3333,
+      "step": 5616
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 9.512727272727273e-07,
+      "loss": 3.2058,
+      "step": 5617
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 9.505454545454547e-07,
+      "loss": 3.561,
+      "step": 5618
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 9.498181818181819e-07,
+      "loss": 3.5938,
+      "step": 5619
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 9.490909090909092e-07,
+      "loss": 3.1965,
+      "step": 5620
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 9.483636363636364e-07,
+      "loss": 3.388,
+      "step": 5621
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 9.476363636363638e-07,
+      "loss": 3.2937,
+      "step": 5622
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 9.46909090909091e-07,
+      "loss": 3.4143,
+      "step": 5623
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 9.461818181818183e-07,
+      "loss": 3.3142,
+      "step": 5624
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 9.454545454545455e-07,
+      "loss": 3.3987,
+      "step": 5625
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 9.447272727272728e-07,
+      "loss": 3.5575,
+      "step": 5626
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 9.440000000000001e-07,
+      "loss": 3.2479,
+      "step": 5627
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 9.432727272727273e-07,
+      "loss": 3.4828,
+      "step": 5628
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 9.425454545454546e-07,
+      "loss": 3.2894,
+      "step": 5629
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 9.41818181818182e-07,
+      "loss": 3.2774,
+      "step": 5630
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 9.410909090909091e-07,
+      "loss": 3.3598,
+      "step": 5631
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 9.403636363636365e-07,
+      "loss": 3.4922,
+      "step": 5632
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 9.396363636363636e-07,
+      "loss": 3.3575,
+      "step": 5633
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 9.38909090909091e-07,
+      "loss": 3.4224,
+      "step": 5634
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 9.381818181818182e-07,
+      "loss": 3.3167,
+      "step": 5635
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 9.374545454545456e-07,
+      "loss": 3.5856,
+      "step": 5636
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 9.367272727272728e-07,
+      "loss": 3.4424,
+      "step": 5637
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 9.360000000000001e-07,
+      "loss": 3.9199,
+      "step": 5638
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 9.352727272727273e-07,
+      "loss": 3.3696,
+      "step": 5639
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 9.345454545454546e-07,
+      "loss": 3.7126,
+      "step": 5640
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 9.338181818181819e-07,
+      "loss": 3.3669,
+      "step": 5641
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 9.330909090909091e-07,
+      "loss": 3.3372,
+      "step": 5642
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 9.323636363636364e-07,
+      "loss": 3.245,
+      "step": 5643
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 9.316363636363638e-07,
+      "loss": 3.4579,
+      "step": 5644
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 9.309090909090909e-07,
+      "loss": 3.326,
+      "step": 5645
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 9.301818181818183e-07,
+      "loss": 3.2775,
+      "step": 5646
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 9.294545454545454e-07,
+      "loss": 3.2151,
+      "step": 5647
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 9.287272727272728e-07,
+      "loss": 3.3354,
+      "step": 5648
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 9.28e-07,
+      "loss": 3.2385,
+      "step": 5649
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 9.272727272727273e-07,
+      "loss": 3.3408,
+      "step": 5650
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 9.265454545454546e-07,
+      "loss": 3.5721,
+      "step": 5651
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 9.258181818181819e-07,
+      "loss": 3.4394,
+      "step": 5652
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 9.250909090909092e-07,
+      "loss": 3.1203,
+      "step": 5653
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 9.243636363636364e-07,
+      "loss": 3.167,
+      "step": 5654
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 9.236363636363638e-07,
+      "loss": 3.4797,
+      "step": 5655
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 9.22909090909091e-07,
+      "loss": 3.2547,
+      "step": 5656
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 9.221818181818183e-07,
+      "loss": 3.2225,
+      "step": 5657
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 9.214545454545455e-07,
+      "loss": 3.5424,
+      "step": 5658
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 9.207272727272728e-07,
+      "loss": 3.7151,
+      "step": 5659
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 3.9241,
+      "step": 5660
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 9.192727272727274e-07,
+      "loss": 3.4047,
+      "step": 5661
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 9.185454545454546e-07,
+      "loss": 3.198,
+      "step": 5662
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 9.17818181818182e-07,
+      "loss": 3.748,
+      "step": 5663
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 9.170909090909091e-07,
+      "loss": 3.3484,
+      "step": 5664
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 9.163636363636365e-07,
+      "loss": 3.4103,
+      "step": 5665
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 9.156363636363637e-07,
+      "loss": 3.3605,
+      "step": 5666
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 9.14909090909091e-07,
+      "loss": 3.4323,
+      "step": 5667
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 9.141818181818182e-07,
+      "loss": 3.527,
+      "step": 5668
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 9.134545454545456e-07,
+      "loss": 3.1285,
+      "step": 5669
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 9.127272727272728e-07,
+      "loss": 3.525,
+      "step": 5670
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 9.120000000000001e-07,
+      "loss": 3.3243,
+      "step": 5671
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 9.112727272727273e-07,
+      "loss": 3.3809,
+      "step": 5672
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 9.105454545454546e-07,
+      "loss": 3.4295,
+      "step": 5673
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 9.098181818181819e-07,
+      "loss": 3.4551,
+      "step": 5674
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 3.5208,
+      "step": 5675
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 9.083636363636364e-07,
+      "loss": 3.2798,
+      "step": 5676
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 9.076363636363638e-07,
+      "loss": 3.7447,
+      "step": 5677
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 9.069090909090909e-07,
+      "loss": 3.3456,
+      "step": 5678
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 9.061818181818183e-07,
+      "loss": 3.354,
+      "step": 5679
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 9.054545454545454e-07,
+      "loss": 3.2902,
+      "step": 5680
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 9.047272727272728e-07,
+      "loss": 3.1926,
+      "step": 5681
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 9.04e-07,
+      "loss": 3.3501,
+      "step": 5682
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 9.032727272727273e-07,
+      "loss": 3.2148,
+      "step": 5683
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 9.025454545454546e-07,
+      "loss": 3.1456,
+      "step": 5684
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 9.018181818181819e-07,
+      "loss": 3.2532,
+      "step": 5685
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 9.010909090909091e-07,
+      "loss": 3.2121,
+      "step": 5686
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 9.003636363636364e-07,
+      "loss": 3.3149,
+      "step": 5687
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 8.996363636363637e-07,
+      "loss": 3.3487,
+      "step": 5688
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 8.98909090909091e-07,
+      "loss": 3.524,
+      "step": 5689
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 8.981818181818183e-07,
+      "loss": 3.1755,
+      "step": 5690
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 8.974545454545456e-07,
+      "loss": 3.292,
+      "step": 5691
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 8.967272727272728e-07,
+      "loss": 3.454,
+      "step": 5692
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 8.960000000000001e-07,
+      "loss": 3.1781,
+      "step": 5693
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 8.952727272727274e-07,
+      "loss": 3.2608,
+      "step": 5694
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 8.945454545454546e-07,
+      "loss": 3.137,
+      "step": 5695
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 8.93818181818182e-07,
+      "loss": 3.2359,
+      "step": 5696
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 8.930909090909091e-07,
+      "loss": 3.3062,
+      "step": 5697
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 8.923636363636365e-07,
+      "loss": 3.7409,
+      "step": 5698
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 8.916363636363637e-07,
+      "loss": 3.6721,
+      "step": 5699
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 8.90909090909091e-07,
+      "loss": 3.2385,
+      "step": 5700
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 8.901818181818182e-07,
+      "loss": 3.1915,
+      "step": 5701
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 8.894545454545456e-07,
+      "loss": 3.2556,
+      "step": 5702
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 8.887272727272728e-07,
+      "loss": 3.2093,
+      "step": 5703
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 8.880000000000001e-07,
+      "loss": 3.2855,
+      "step": 5704
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 8.872727272727273e-07,
+      "loss": 3.6213,
+      "step": 5705
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 8.865454545454546e-07,
+      "loss": 3.2449,
+      "step": 5706
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 8.858181818181819e-07,
+      "loss": 3.2752,
+      "step": 5707
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 8.850909090909091e-07,
+      "loss": 3.1291,
+      "step": 5708
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 8.843636363636364e-07,
+      "loss": 3.4475,
+      "step": 5709
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 8.836363636363638e-07,
+      "loss": 3.5221,
+      "step": 5710
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 8.829090909090909e-07,
+      "loss": 3.6801,
+      "step": 5711
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 8.821818181818183e-07,
+      "loss": 3.6716,
+      "step": 5712
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 8.814545454545455e-07,
+      "loss": 3.2866,
+      "step": 5713
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 8.807272727272728e-07,
+      "loss": 3.2784,
+      "step": 5714
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 8.8e-07,
+      "loss": 3.5282,
+      "step": 5715
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 8.792727272727274e-07,
+      "loss": 3.4406,
+      "step": 5716
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 8.785454545454546e-07,
+      "loss": 3.3691,
+      "step": 5717
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 8.778181818181819e-07,
+      "loss": 3.5901,
+      "step": 5718
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 8.770909090909091e-07,
+      "loss": 3.2937,
+      "step": 5719
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 8.763636363636364e-07,
+      "loss": 3.8213,
+      "step": 5720
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 8.756363636363637e-07,
+      "loss": 3.3359,
+      "step": 5721
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 8.749090909090909e-07,
+      "loss": 3.4479,
+      "step": 5722
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 8.741818181818182e-07,
+      "loss": 3.4519,
+      "step": 5723
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 8.734545454545456e-07,
+      "loss": 3.2406,
+      "step": 5724
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 8.727272727272728e-07,
+      "loss": 3.1988,
+      "step": 5725
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 8.720000000000001e-07,
+      "loss": 3.3526,
+      "step": 5726
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 8.712727272727274e-07,
+      "loss": 3.3245,
+      "step": 5727
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 8.705454545454546e-07,
+      "loss": 3.23,
+      "step": 5728
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 8.69818181818182e-07,
+      "loss": 3.5933,
+      "step": 5729
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 8.690909090909091e-07,
+      "loss": 3.1273,
+      "step": 5730
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 8.683636363636365e-07,
+      "loss": 3.3772,
+      "step": 5731
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 8.676363636363637e-07,
+      "loss": 3.3419,
+      "step": 5732
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 8.66909090909091e-07,
+      "loss": 3.2075,
+      "step": 5733
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 8.661818181818182e-07,
+      "loss": 3.3675,
+      "step": 5734
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 8.654545454545456e-07,
+      "loss": 3.2232,
+      "step": 5735
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 8.647272727272728e-07,
+      "loss": 3.2532,
+      "step": 5736
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 8.640000000000001e-07,
+      "loss": 3.8254,
+      "step": 5737
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 8.632727272727274e-07,
+      "loss": 3.6889,
+      "step": 5738
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 8.625454545454546e-07,
+      "loss": 3.5056,
+      "step": 5739
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 8.618181818181819e-07,
+      "loss": 3.3071,
+      "step": 5740
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 8.610909090909092e-07,
+      "loss": 3.2442,
+      "step": 5741
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 8.603636363636364e-07,
+      "loss": 3.5015,
+      "step": 5742
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 8.596363636363638e-07,
+      "loss": 3.2274,
+      "step": 5743
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 8.589090909090909e-07,
+      "loss": 3.3677,
+      "step": 5744
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 8.581818181818183e-07,
+      "loss": 3.4984,
+      "step": 5745
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 8.574545454545455e-07,
+      "loss": 3.2641,
+      "step": 5746
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 8.567272727272728e-07,
+      "loss": 3.2493,
+      "step": 5747
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 8.56e-07,
+      "loss": 3.3735,
+      "step": 5748
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 8.552727272727274e-07,
+      "loss": 3.1373,
+      "step": 5749
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 8.545454545454546e-07,
+      "loss": 3.2505,
+      "step": 5750
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 8.538181818181819e-07,
+      "loss": 3.1072,
+      "step": 5751
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 8.530909090909091e-07,
+      "loss": 3.5927,
+      "step": 5752
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 8.523636363636364e-07,
+      "loss": 3.3231,
+      "step": 5753
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 8.516363636363637e-07,
+      "loss": 3.7682,
+      "step": 5754
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 8.509090909090909e-07,
+      "loss": 3.2903,
+      "step": 5755
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 8.501818181818182e-07,
+      "loss": 3.1637,
+      "step": 5756
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 8.494545454545456e-07,
+      "loss": 3.2177,
+      "step": 5757
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 8.487272727272727e-07,
+      "loss": 3.9201,
+      "step": 5758
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 8.480000000000001e-07,
+      "loss": 3.3316,
+      "step": 5759
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 8.472727272727272e-07,
+      "loss": 3.358,
+      "step": 5760
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 8.465454545454546e-07,
+      "loss": 3.1527,
+      "step": 5761
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 8.45818181818182e-07,
+      "loss": 3.2086,
+      "step": 5762
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 8.450909090909092e-07,
+      "loss": 3.1986,
+      "step": 5763
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 8.443636363636365e-07,
+      "loss": 3.3616,
+      "step": 5764
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 8.436363636363637e-07,
+      "loss": 3.4102,
+      "step": 5765
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 8.429090909090911e-07,
+      "loss": 3.258,
+      "step": 5766
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 8.421818181818182e-07,
+      "loss": 3.5663,
+      "step": 5767
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 8.414545454545456e-07,
+      "loss": 3.3789,
+      "step": 5768
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 8.407272727272727e-07,
+      "loss": 3.4817,
+      "step": 5769
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 3.2947,
+      "step": 5770
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 8.392727272727274e-07,
+      "loss": 3.4457,
+      "step": 5771
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 8.385454545454546e-07,
+      "loss": 3.4085,
+      "step": 5772
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 8.378181818181819e-07,
+      "loss": 3.356,
+      "step": 5773
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 8.370909090909092e-07,
+      "loss": 3.1818,
+      "step": 5774
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 8.363636363636364e-07,
+      "loss": 3.4351,
+      "step": 5775
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 8.356363636363638e-07,
+      "loss": 3.2333,
+      "step": 5776
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 8.349090909090909e-07,
+      "loss": 3.5961,
+      "step": 5777
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 8.341818181818183e-07,
+      "loss": 3.4353,
+      "step": 5778
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 8.334545454545455e-07,
+      "loss": 3.5793,
+      "step": 5779
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 8.327272727272728e-07,
+      "loss": 3.3073,
+      "step": 5780
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 8.32e-07,
+      "loss": 3.5914,
+      "step": 5781
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 8.312727272727274e-07,
+      "loss": 3.2644,
+      "step": 5782
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 8.305454545454546e-07,
+      "loss": 3.6585,
+      "step": 5783
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 8.298181818181819e-07,
+      "loss": 3.7772,
+      "step": 5784
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 8.290909090909091e-07,
+      "loss": 3.8071,
+      "step": 5785
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 8.283636363636364e-07,
+      "loss": 3.3038,
+      "step": 5786
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 8.276363636363637e-07,
+      "loss": 3.2037,
+      "step": 5787
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 8.26909090909091e-07,
+      "loss": 3.218,
+      "step": 5788
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 8.261818181818182e-07,
+      "loss": 3.8786,
+      "step": 5789
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 8.254545454545456e-07,
+      "loss": 3.2834,
+      "step": 5790
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 8.247272727272727e-07,
+      "loss": 3.229,
+      "step": 5791
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 8.240000000000001e-07,
+      "loss": 3.1903,
+      "step": 5792
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 8.232727272727273e-07,
+      "loss": 3.5039,
+      "step": 5793
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 8.225454545454546e-07,
+      "loss": 3.4098,
+      "step": 5794
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 8.218181818181818e-07,
+      "loss": 3.469,
+      "step": 5795
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 8.210909090909092e-07,
+      "loss": 3.6058,
+      "step": 5796
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 8.203636363636364e-07,
+      "loss": 3.2767,
+      "step": 5797
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 8.196363636363637e-07,
+      "loss": 3.2714,
+      "step": 5798
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 8.189090909090911e-07,
+      "loss": 3.4036,
+      "step": 5799
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 8.181818181818182e-07,
+      "loss": 3.3282,
+      "step": 5800
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 8.174545454545456e-07,
+      "loss": 3.5692,
+      "step": 5801
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 8.167272727272727e-07,
+      "loss": 3.2239,
+      "step": 5802
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 8.160000000000001e-07,
+      "loss": 3.5219,
+      "step": 5803
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 8.152727272727274e-07,
+      "loss": 3.2101,
+      "step": 5804
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 8.145454545454546e-07,
+      "loss": 3.5286,
+      "step": 5805
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 8.138181818181819e-07,
+      "loss": 3.1969,
+      "step": 5806
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 8.130909090909092e-07,
+      "loss": 3.3922,
+      "step": 5807
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 8.123636363636364e-07,
+      "loss": 3.0769,
+      "step": 5808
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 8.116363636363638e-07,
+      "loss": 3.2816,
+      "step": 5809
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 8.109090909090909e-07,
+      "loss": 3.2338,
+      "step": 5810
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 8.101818181818183e-07,
+      "loss": 3.254,
+      "step": 5811
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 8.094545454545455e-07,
+      "loss": 3.309,
+      "step": 5812
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 8.087272727272729e-07,
+      "loss": 3.3022,
+      "step": 5813
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 8.08e-07,
+      "loss": 3.3519,
+      "step": 5814
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 8.072727272727274e-07,
+      "loss": 3.405,
+      "step": 5815
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 8.065454545454545e-07,
+      "loss": 3.2304,
+      "step": 5816
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 8.058181818181819e-07,
+      "loss": 3.2267,
+      "step": 5817
+    },
+    {
+      "epoch": 21.0,
+      "eval_accuracy": 0.20540540540540542,
+      "eval_loss": 3.3848252296447754,
+      "eval_runtime": 143.4395,
+      "eval_samples_per_second": 2.579,
+      "eval_steps_per_second": 0.648,
+      "step": 5817
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 8.050909090909092e-07,
+      "loss": 3.2545,
+      "step": 5818
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 8.043636363636364e-07,
+      "loss": 3.4988,
+      "step": 5819
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 8.036363636363637e-07,
+      "loss": 3.2567,
+      "step": 5820
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 8.02909090909091e-07,
+      "loss": 3.2852,
+      "step": 5821
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.021818181818182e-07,
+      "loss": 3.1297,
+      "step": 5822
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.014545454545456e-07,
+      "loss": 3.3144,
+      "step": 5823
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.007272727272727e-07,
+      "loss": 3.6496,
+      "step": 5824
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.3384,
+      "step": 5825
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.992727272727273e-07,
+      "loss": 3.3501,
+      "step": 5826
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 7.985454545454546e-07,
+      "loss": 3.3501,
+      "step": 5827
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 7.978181818181818e-07,
+      "loss": 3.4933,
+      "step": 5828
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 7.970909090909092e-07,
+      "loss": 3.3476,
+      "step": 5829
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 7.963636363636364e-07,
+      "loss": 3.4544,
+      "step": 5830
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 7.956363636363637e-07,
+      "loss": 3.2943,
+      "step": 5831
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 7.949090909090909e-07,
+      "loss": 3.646,
+      "step": 5832
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 7.941818181818182e-07,
+      "loss": 3.7213,
+      "step": 5833
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 7.934545454545456e-07,
+      "loss": 3.5404,
+      "step": 5834
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 7.927272727272727e-07,
+      "loss": 3.2216,
+      "step": 5835
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 7.920000000000001e-07,
+      "loss": 3.6984,
+      "step": 5836
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 7.912727272727274e-07,
+      "loss": 3.3566,
+      "step": 5837
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 7.905454545454547e-07,
+      "loss": 3.4647,
+      "step": 5838
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 7.898181818181819e-07,
+      "loss": 3.718,
+      "step": 5839
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 7.890909090909092e-07,
+      "loss": 3.51,
+      "step": 5840
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 7.883636363636364e-07,
+      "loss": 3.0971,
+      "step": 5841
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 7.876363636363637e-07,
+      "loss": 3.4531,
+      "step": 5842
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 7.86909090909091e-07,
+      "loss": 3.2772,
+      "step": 5843
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 7.861818181818183e-07,
+      "loss": 3.1467,
+      "step": 5844
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 7.854545454545455e-07,
+      "loss": 3.2328,
+      "step": 5845
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 7.847272727272729e-07,
+      "loss": 3.316,
+      "step": 5846
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 7.84e-07,
+      "loss": 3.234,
+      "step": 5847
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 7.832727272727274e-07,
+      "loss": 3.2452,
+      "step": 5848
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 7.825454545454545e-07,
+      "loss": 3.1676,
+      "step": 5849
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 7.818181818181819e-07,
+      "loss": 3.0948,
+      "step": 5850
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 7.810909090909092e-07,
+      "loss": 3.2312,
+      "step": 5851
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 7.803636363636364e-07,
+      "loss": 3.2966,
+      "step": 5852
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 7.796363636363637e-07,
+      "loss": 3.5756,
+      "step": 5853
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 7.78909090909091e-07,
+      "loss": 3.2646,
+      "step": 5854
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 7.781818181818182e-07,
+      "loss": 3.6072,
+      "step": 5855
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 7.774545454545456e-07,
+      "loss": 3.4464,
+      "step": 5856
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 7.767272727272727e-07,
+      "loss": 3.3003,
+      "step": 5857
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 7.760000000000001e-07,
+      "loss": 3.2465,
+      "step": 5858
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 7.752727272727273e-07,
+      "loss": 3.3955,
+      "step": 5859
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 7.745454545454546e-07,
+      "loss": 3.663,
+      "step": 5860
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 7.738181818181818e-07,
+      "loss": 4.0391,
+      "step": 5861
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 7.730909090909092e-07,
+      "loss": 3.6193,
+      "step": 5862
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 7.723636363636363e-07,
+      "loss": 3.4309,
+      "step": 5863
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 7.716363636363637e-07,
+      "loss": 3.4816,
+      "step": 5864
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 7.709090909090909e-07,
+      "loss": 3.6089,
+      "step": 5865
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 7.701818181818182e-07,
+      "loss": 3.2585,
+      "step": 5866
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 7.694545454545455e-07,
+      "loss": 3.3248,
+      "step": 5867
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 7.687272727272728e-07,
+      "loss": 3.5309,
+      "step": 5868
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 7.68e-07,
+      "loss": 3.5256,
+      "step": 5869
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 7.672727272727274e-07,
+      "loss": 3.5718,
+      "step": 5870
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 7.665454545454547e-07,
+      "loss": 3.5424,
+      "step": 5871
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 7.658181818181819e-07,
+      "loss": 3.525,
+      "step": 5872
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 7.650909090909092e-07,
+      "loss": 3.4831,
+      "step": 5873
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 7.643636363636364e-07,
+      "loss": 3.2317,
+      "step": 5874
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 7.636363636363637e-07,
+      "loss": 3.707,
+      "step": 5875
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 7.62909090909091e-07,
+      "loss": 3.0505,
+      "step": 5876
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 7.621818181818183e-07,
+      "loss": 3.2321,
+      "step": 5877
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 7.614545454545455e-07,
+      "loss": 3.6331,
+      "step": 5878
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 7.607272727272729e-07,
+      "loss": 3.4638,
+      "step": 5879
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 7.6e-07,
+      "loss": 3.4331,
+      "step": 5880
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 7.592727272727274e-07,
+      "loss": 3.5871,
+      "step": 5881
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 7.585454545454545e-07,
+      "loss": 3.3158,
+      "step": 5882
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 7.578181818181819e-07,
+      "loss": 3.2518,
+      "step": 5883
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 7.570909090909092e-07,
+      "loss": 3.4966,
+      "step": 5884
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 7.563636363636364e-07,
+      "loss": 3.6,
+      "step": 5885
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 7.556363636363637e-07,
+      "loss": 3.4651,
+      "step": 5886
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 7.54909090909091e-07,
+      "loss": 3.2994,
+      "step": 5887
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 7.541818181818182e-07,
+      "loss": 3.512,
+      "step": 5888
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 7.534545454545456e-07,
+      "loss": 3.1108,
+      "step": 5889
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 7.527272727272727e-07,
+      "loss": 3.2589,
+      "step": 5890
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 7.520000000000001e-07,
+      "loss": 3.6006,
+      "step": 5891
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 7.512727272727273e-07,
+      "loss": 3.2344,
+      "step": 5892
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 7.505454545454547e-07,
+      "loss": 3.0048,
+      "step": 5893
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 7.498181818181818e-07,
+      "loss": 3.3446,
+      "step": 5894
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 7.490909090909092e-07,
+      "loss": 3.5036,
+      "step": 5895
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 7.483636363636363e-07,
+      "loss": 3.5747,
+      "step": 5896
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 7.476363636363637e-07,
+      "loss": 3.4536,
+      "step": 5897
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 7.46909090909091e-07,
+      "loss": 3.3434,
+      "step": 5898
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 7.461818181818182e-07,
+      "loss": 3.4119,
+      "step": 5899
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 7.454545454545455e-07,
+      "loss": 3.6163,
+      "step": 5900
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 7.447272727272728e-07,
+      "loss": 3.1705,
+      "step": 5901
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 7.44e-07,
+      "loss": 3.5182,
+      "step": 5902
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 7.432727272727274e-07,
+      "loss": 3.4299,
+      "step": 5903
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 7.425454545454545e-07,
+      "loss": 3.4691,
+      "step": 5904
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 7.418181818181819e-07,
+      "loss": 3.3522,
+      "step": 5905
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 7.410909090909091e-07,
+      "loss": 3.5151,
+      "step": 5906
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 7.403636363636364e-07,
+      "loss": 3.4946,
+      "step": 5907
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 7.396363636363637e-07,
+      "loss": 3.602,
+      "step": 5908
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 7.38909090909091e-07,
+      "loss": 3.2052,
+      "step": 5909
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 7.381818181818183e-07,
+      "loss": 3.2548,
+      "step": 5910
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 7.374545454545455e-07,
+      "loss": 3.2715,
+      "step": 5911
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 7.367272727272729e-07,
+      "loss": 3.3652,
+      "step": 5912
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 7.36e-07,
+      "loss": 3.4281,
+      "step": 5913
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 7.352727272727274e-07,
+      "loss": 3.406,
+      "step": 5914
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 7.345454545454545e-07,
+      "loss": 3.654,
+      "step": 5915
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 7.338181818181819e-07,
+      "loss": 3.4257,
+      "step": 5916
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 7.330909090909092e-07,
+      "loss": 3.2836,
+      "step": 5917
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 7.323636363636365e-07,
+      "loss": 3.4855,
+      "step": 5918
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 7.316363636363637e-07,
+      "loss": 3.6636,
+      "step": 5919
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 7.30909090909091e-07,
+      "loss": 3.3048,
+      "step": 5920
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 7.301818181818182e-07,
+      "loss": 3.7545,
+      "step": 5921
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 7.294545454545455e-07,
+      "loss": 3.3692,
+      "step": 5922
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 7.287272727272728e-07,
+      "loss": 3.2634,
+      "step": 5923
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 7.280000000000001e-07,
+      "loss": 3.2232,
+      "step": 5924
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 7.272727272727273e-07,
+      "loss": 3.2162,
+      "step": 5925
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 7.265454545454547e-07,
+      "loss": 3.1663,
+      "step": 5926
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 7.258181818181818e-07,
+      "loss": 3.3778,
+      "step": 5927
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 7.250909090909092e-07,
+      "loss": 3.317,
+      "step": 5928
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 7.243636363636363e-07,
+      "loss": 3.3373,
+      "step": 5929
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 7.236363636363637e-07,
+      "loss": 3.344,
+      "step": 5930
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 7.22909090909091e-07,
+      "loss": 3.4191,
+      "step": 5931
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 7.221818181818182e-07,
+      "loss": 3.3849,
+      "step": 5932
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 7.214545454545455e-07,
+      "loss": 3.2727,
+      "step": 5933
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 7.207272727272728e-07,
+      "loss": 3.2335,
+      "step": 5934
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 7.2e-07,
+      "loss": 3.2945,
+      "step": 5935
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 7.192727272727274e-07,
+      "loss": 3.2652,
+      "step": 5936
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 7.185454545454545e-07,
+      "loss": 3.0463,
+      "step": 5937
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 7.178181818181819e-07,
+      "loss": 3.7317,
+      "step": 5938
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 7.170909090909091e-07,
+      "loss": 3.206,
+      "step": 5939
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 7.163636363636364e-07,
+      "loss": 3.304,
+      "step": 5940
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 7.156363636363636e-07,
+      "loss": 3.3261,
+      "step": 5941
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 7.14909090909091e-07,
+      "loss": 3.1663,
+      "step": 5942
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 7.141818181818184e-07,
+      "loss": 3.202,
+      "step": 5943
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 7.134545454545455e-07,
+      "loss": 3.2561,
+      "step": 5944
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 7.127272727272729e-07,
+      "loss": 3.5579,
+      "step": 5945
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 7.12e-07,
+      "loss": 3.1602,
+      "step": 5946
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 7.112727272727274e-07,
+      "loss": 3.5998,
+      "step": 5947
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 7.105454545454546e-07,
+      "loss": 3.5596,
+      "step": 5948
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 7.098181818181819e-07,
+      "loss": 3.2854,
+      "step": 5949
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 7.090909090909092e-07,
+      "loss": 3.1886,
+      "step": 5950
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 7.083636363636365e-07,
+      "loss": 3.4219,
+      "step": 5951
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 7.076363636363637e-07,
+      "loss": 3.418,
+      "step": 5952
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 7.06909090909091e-07,
+      "loss": 3.383,
+      "step": 5953
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 7.061818181818182e-07,
+      "loss": 3.6106,
+      "step": 5954
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 7.054545454545455e-07,
+      "loss": 3.2205,
+      "step": 5955
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 7.047272727272728e-07,
+      "loss": 3.51,
+      "step": 5956
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 7.040000000000001e-07,
+      "loss": 3.2468,
+      "step": 5957
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 7.032727272727273e-07,
+      "loss": 3.4836,
+      "step": 5958
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 7.025454545454547e-07,
+      "loss": 3.4863,
+      "step": 5959
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 7.018181818181818e-07,
+      "loss": 3.34,
+      "step": 5960
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 7.010909090909092e-07,
+      "loss": 3.4464,
+      "step": 5961
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 7.003636363636363e-07,
+      "loss": 3.4679,
+      "step": 5962
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 6.996363636363637e-07,
+      "loss": 3.2926,
+      "step": 5963
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 6.98909090909091e-07,
+      "loss": 3.2228,
+      "step": 5964
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 6.981818181818182e-07,
+      "loss": 3.7567,
+      "step": 5965
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 6.974545454545455e-07,
+      "loss": 3.3856,
+      "step": 5966
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 6.967272727272728e-07,
+      "loss": 3.647,
+      "step": 5967
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 6.96e-07,
+      "loss": 3.2556,
+      "step": 5968
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 6.952727272727273e-07,
+      "loss": 3.3425,
+      "step": 5969
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 6.945454545454545e-07,
+      "loss": 3.5246,
+      "step": 5970
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 6.938181818181819e-07,
+      "loss": 3.3862,
+      "step": 5971
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 6.930909090909091e-07,
+      "loss": 3.4172,
+      "step": 5972
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 6.923636363636365e-07,
+      "loss": 3.5814,
+      "step": 5973
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 6.916363636363636e-07,
+      "loss": 3.8412,
+      "step": 5974
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 6.90909090909091e-07,
+      "loss": 3.5902,
+      "step": 5975
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 6.901818181818181e-07,
+      "loss": 3.4718,
+      "step": 5976
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 6.894545454545455e-07,
+      "loss": 3.4383,
+      "step": 5977
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 6.887272727272728e-07,
+      "loss": 3.3092,
+      "step": 5978
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 6.88e-07,
+      "loss": 3.4266,
+      "step": 5979
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 6.872727272727274e-07,
+      "loss": 3.283,
+      "step": 5980
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 6.865454545454546e-07,
+      "loss": 3.3608,
+      "step": 5981
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 6.858181818181819e-07,
+      "loss": 3.2951,
+      "step": 5982
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 6.850909090909092e-07,
+      "loss": 3.3514,
+      "step": 5983
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 6.843636363636365e-07,
+      "loss": 3.4155,
+      "step": 5984
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 6.836363636363637e-07,
+      "loss": 3.5333,
+      "step": 5985
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 6.82909090909091e-07,
+      "loss": 3.2849,
+      "step": 5986
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 6.821818181818182e-07,
+      "loss": 3.5649,
+      "step": 5987
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 6.814545454545455e-07,
+      "loss": 3.2521,
+      "step": 5988
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 6.807272727272728e-07,
+      "loss": 3.3283,
+      "step": 5989
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 3.3482,
+      "step": 5990
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 6.792727272727273e-07,
+      "loss": 3.3928,
+      "step": 5991
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 6.785454545454547e-07,
+      "loss": 3.3672,
+      "step": 5992
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 6.778181818181818e-07,
+      "loss": 3.4413,
+      "step": 5993
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 6.770909090909092e-07,
+      "loss": 3.9623,
+      "step": 5994
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 6.763636363636363e-07,
+      "loss": 3.318,
+      "step": 5995
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 6.756363636363637e-07,
+      "loss": 3.5176,
+      "step": 5996
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 6.74909090909091e-07,
+      "loss": 3.2904,
+      "step": 5997
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 6.741818181818183e-07,
+      "loss": 3.1927,
+      "step": 5998
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 6.734545454545455e-07,
+      "loss": 3.2271,
+      "step": 5999
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 6.727272727272728e-07,
+      "loss": 3.128,
+      "step": 6000
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 6.72e-07,
+      "loss": 3.5709,
+      "step": 6001
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 6.712727272727273e-07,
+      "loss": 3.1728,
+      "step": 6002
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 6.705454545454546e-07,
+      "loss": 3.3597,
+      "step": 6003
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 6.698181818181819e-07,
+      "loss": 3.1951,
+      "step": 6004
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 6.690909090909091e-07,
+      "loss": 3.3795,
+      "step": 6005
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 6.683636363636365e-07,
+      "loss": 3.401,
+      "step": 6006
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 6.676363636363636e-07,
+      "loss": 3.4101,
+      "step": 6007
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 6.66909090909091e-07,
+      "loss": 3.1792,
+      "step": 6008
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 6.661818181818181e-07,
+      "loss": 3.2962,
+      "step": 6009
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 6.654545454545455e-07,
+      "loss": 3.7715,
+      "step": 6010
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 6.647272727272728e-07,
+      "loss": 3.2983,
+      "step": 6011
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 6.64e-07,
+      "loss": 3.3473,
+      "step": 6012
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 6.632727272727273e-07,
+      "loss": 3.1796,
+      "step": 6013
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 6.625454545454546e-07,
+      "loss": 3.3589,
+      "step": 6014
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 6.618181818181818e-07,
+      "loss": 3.53,
+      "step": 6015
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 6.610909090909091e-07,
+      "loss": 3.2465,
+      "step": 6016
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 6.603636363636365e-07,
+      "loss": 3.176,
+      "step": 6017
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 6.596363636363637e-07,
+      "loss": 3.1916,
+      "step": 6018
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 6.58909090909091e-07,
+      "loss": 3.9283,
+      "step": 6019
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 6.581818181818182e-07,
+      "loss": 3.636,
+      "step": 6020
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 6.574545454545455e-07,
+      "loss": 3.3791,
+      "step": 6021
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 6.567272727272728e-07,
+      "loss": 3.4042,
+      "step": 6022
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 6.560000000000002e-07,
+      "loss": 3.694,
+      "step": 6023
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 6.552727272727273e-07,
+      "loss": 3.5324,
+      "step": 6024
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 6.545454545454547e-07,
+      "loss": 3.5546,
+      "step": 6025
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 6.538181818181818e-07,
+      "loss": 3.5757,
+      "step": 6026
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 6.530909090909092e-07,
+      "loss": 3.4561,
+      "step": 6027
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 6.523636363636364e-07,
+      "loss": 3.3069,
+      "step": 6028
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 6.516363636363637e-07,
+      "loss": 3.1525,
+      "step": 6029
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 6.50909090909091e-07,
+      "loss": 3.5063,
+      "step": 6030
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 6.501818181818183e-07,
+      "loss": 3.1303,
+      "step": 6031
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 6.494545454545455e-07,
+      "loss": 3.6217,
+      "step": 6032
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 6.487272727272728e-07,
+      "loss": 3.3593,
+      "step": 6033
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 6.48e-07,
+      "loss": 3.371,
+      "step": 6034
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 6.472727272727273e-07,
+      "loss": 3.1584,
+      "step": 6035
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 6.465454545454546e-07,
+      "loss": 3.6928,
+      "step": 6036
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 6.458181818181819e-07,
+      "loss": 3.4236,
+      "step": 6037
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 6.450909090909091e-07,
+      "loss": 3.5854,
+      "step": 6038
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 6.443636363636365e-07,
+      "loss": 3.5988,
+      "step": 6039
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 6.436363636363636e-07,
+      "loss": 3.5056,
+      "step": 6040
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 6.42909090909091e-07,
+      "loss": 3.7155,
+      "step": 6041
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 6.421818181818181e-07,
+      "loss": 3.2943,
+      "step": 6042
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 6.414545454545455e-07,
+      "loss": 3.2811,
+      "step": 6043
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 6.407272727272728e-07,
+      "loss": 3.322,
+      "step": 6044
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 6.4e-07,
+      "loss": 3.3339,
+      "step": 6045
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 6.392727272727273e-07,
+      "loss": 3.5013,
+      "step": 6046
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 6.385454545454546e-07,
+      "loss": 3.6224,
+      "step": 6047
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 6.378181818181818e-07,
+      "loss": 3.5508,
+      "step": 6048
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 6.370909090909091e-07,
+      "loss": 3.4117,
+      "step": 6049
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 6.363636363636364e-07,
+      "loss": 3.4197,
+      "step": 6050
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 6.356363636363637e-07,
+      "loss": 3.2767,
+      "step": 6051
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 6.34909090909091e-07,
+      "loss": 3.1249,
+      "step": 6052
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 6.341818181818183e-07,
+      "loss": 3.2219,
+      "step": 6053
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 6.334545454545455e-07,
+      "loss": 3.3525,
+      "step": 6054
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 6.327272727272728e-07,
+      "loss": 3.2319,
+      "step": 6055
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 6.320000000000002e-07,
+      "loss": 3.4685,
+      "step": 6056
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 6.312727272727273e-07,
+      "loss": 3.3122,
+      "step": 6057
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 6.305454545454547e-07,
+      "loss": 3.4593,
+      "step": 6058
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 6.298181818181818e-07,
+      "loss": 3.2187,
+      "step": 6059
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 6.290909090909092e-07,
+      "loss": 3.2168,
+      "step": 6060
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 6.283636363636364e-07,
+      "loss": 3.3409,
+      "step": 6061
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 6.276363636363637e-07,
+      "loss": 3.4607,
+      "step": 6062
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 6.26909090909091e-07,
+      "loss": 3.2565,
+      "step": 6063
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 6.261818181818183e-07,
+      "loss": 3.4125,
+      "step": 6064
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 6.254545454545455e-07,
+      "loss": 3.1268,
+      "step": 6065
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 6.247272727272727e-07,
+      "loss": 3.4526,
+      "step": 6066
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 6.24e-07,
+      "loss": 3.3002,
+      "step": 6067
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 6.232727272727273e-07,
+      "loss": 3.5729,
+      "step": 6068
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 6.225454545454546e-07,
+      "loss": 3.1794,
+      "step": 6069
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 6.218181818181819e-07,
+      "loss": 3.1745,
+      "step": 6070
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 6.210909090909092e-07,
+      "loss": 3.1268,
+      "step": 6071
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 6.203636363636365e-07,
+      "loss": 3.3885,
+      "step": 6072
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 6.196363636363637e-07,
+      "loss": 3.2215,
+      "step": 6073
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 6.18909090909091e-07,
+      "loss": 3.207,
+      "step": 6074
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 6.181818181818182e-07,
+      "loss": 3.3255,
+      "step": 6075
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 6.174545454545455e-07,
+      "loss": 3.4255,
+      "step": 6076
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 6.167272727272728e-07,
+      "loss": 3.5098,
+      "step": 6077
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 6.160000000000001e-07,
+      "loss": 3.5105,
+      "step": 6078
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 6.152727272727274e-07,
+      "loss": 3.5332,
+      "step": 6079
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 6.145454545454546e-07,
+      "loss": 3.1114,
+      "step": 6080
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 6.138181818181819e-07,
+      "loss": 3.1881,
+      "step": 6081
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 6.130909090909091e-07,
+      "loss": 3.2152,
+      "step": 6082
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 6.123636363636364e-07,
+      "loss": 3.261,
+      "step": 6083
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 6.116363636363637e-07,
+      "loss": 3.6738,
+      "step": 6084
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 6.10909090909091e-07,
+      "loss": 3.307,
+      "step": 6085
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 6.101818181818183e-07,
+      "loss": 3.3603,
+      "step": 6086
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 6.094545454545455e-07,
+      "loss": 3.6886,
+      "step": 6087
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 6.087272727272728e-07,
+      "loss": 3.2241,
+      "step": 6088
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 6.08e-07,
+      "loss": 3.2916,
+      "step": 6089
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 6.072727272727273e-07,
+      "loss": 3.3415,
+      "step": 6090
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 6.065454545454546e-07,
+      "loss": 3.4029,
+      "step": 6091
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 6.058181818181818e-07,
+      "loss": 3.0965,
+      "step": 6092
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.050909090909092e-07,
+      "loss": 3.3282,
+      "step": 6093
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.043636363636364e-07,
+      "loss": 3.062,
+      "step": 6094
+    },
+    {
+      "epoch": 22.0,
+      "eval_accuracy": 0.1864864864864865,
+      "eval_loss": 3.382753849029541,
+      "eval_runtime": 144.0717,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 0.646,
+      "step": 6094
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.036363636363637e-07,
+      "loss": 3.4644,
+      "step": 6095
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 6.029090909090909e-07,
+      "loss": 3.2889,
+      "step": 6096
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 6.021818181818182e-07,
+      "loss": 3.5349,
+      "step": 6097
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 6.014545454545455e-07,
+      "loss": 3.1954,
+      "step": 6098
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 6.007272727272727e-07,
+      "loss": 3.3048,
+      "step": 6099
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.3321,
+      "step": 6100
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 5.992727272727273e-07,
+      "loss": 3.3865,
+      "step": 6101
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 5.985454545454546e-07,
+      "loss": 3.2114,
+      "step": 6102
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 5.978181818181818e-07,
+      "loss": 3.4415,
+      "step": 6103
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 5.970909090909091e-07,
+      "loss": 3.5022,
+      "step": 6104
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 5.963636363636364e-07,
+      "loss": 3.6518,
+      "step": 6105
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 5.956363636363636e-07,
+      "loss": 3.5798,
+      "step": 6106
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 5.94909090909091e-07,
+      "loss": 3.3787,
+      "step": 6107
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 5.941818181818182e-07,
+      "loss": 3.6335,
+      "step": 6108
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 5.934545454545455e-07,
+      "loss": 3.5049,
+      "step": 6109
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 5.927272727272729e-07,
+      "loss": 3.3559,
+      "step": 6110
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 5.920000000000001e-07,
+      "loss": 3.2156,
+      "step": 6111
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 5.912727272727274e-07,
+      "loss": 3.4815,
+      "step": 6112
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 5.905454545454546e-07,
+      "loss": 3.5385,
+      "step": 6113
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 5.898181818181819e-07,
+      "loss": 3.2533,
+      "step": 6114
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 5.890909090909091e-07,
+      "loss": 3.1458,
+      "step": 6115
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 5.883636363636364e-07,
+      "loss": 3.4191,
+      "step": 6116
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 5.876363636363637e-07,
+      "loss": 3.1888,
+      "step": 6117
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 5.86909090909091e-07,
+      "loss": 3.3039,
+      "step": 6118
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 5.861818181818183e-07,
+      "loss": 3.2553,
+      "step": 6119
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 5.854545454545455e-07,
+      "loss": 3.0455,
+      "step": 6120
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 5.847272727272728e-07,
+      "loss": 3.4231,
+      "step": 6121
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 5.84e-07,
+      "loss": 3.6821,
+      "step": 6122
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 5.832727272727273e-07,
+      "loss": 3.2609,
+      "step": 6123
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 5.825454545454546e-07,
+      "loss": 3.3311,
+      "step": 6124
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 5.818181818181819e-07,
+      "loss": 3.3061,
+      "step": 6125
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 5.810909090909092e-07,
+      "loss": 3.4656,
+      "step": 6126
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 5.803636363636364e-07,
+      "loss": 3.2739,
+      "step": 6127
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 5.796363636363637e-07,
+      "loss": 3.3691,
+      "step": 6128
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 5.789090909090909e-07,
+      "loss": 3.5695,
+      "step": 6129
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 5.781818181818182e-07,
+      "loss": 3.8596,
+      "step": 6130
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 5.774545454545455e-07,
+      "loss": 3.1613,
+      "step": 6131
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 5.767272727272727e-07,
+      "loss": 3.3528,
+      "step": 6132
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 5.760000000000001e-07,
+      "loss": 3.3659,
+      "step": 6133
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 5.752727272727273e-07,
+      "loss": 3.7736,
+      "step": 6134
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 5.745454545454546e-07,
+      "loss": 3.2709,
+      "step": 6135
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 5.738181818181818e-07,
+      "loss": 3.5461,
+      "step": 6136
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 5.730909090909091e-07,
+      "loss": 3.7592,
+      "step": 6137
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 5.723636363636364e-07,
+      "loss": 3.5109,
+      "step": 6138
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 5.716363636363636e-07,
+      "loss": 3.5864,
+      "step": 6139
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 5.70909090909091e-07,
+      "loss": 3.3665,
+      "step": 6140
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 5.701818181818182e-07,
+      "loss": 3.4431,
+      "step": 6141
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 5.694545454545455e-07,
+      "loss": 3.3881,
+      "step": 6142
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 5.687272727272729e-07,
+      "loss": 3.2075,
+      "step": 6143
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 5.680000000000001e-07,
+      "loss": 3.0195,
+      "step": 6144
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 5.672727272727274e-07,
+      "loss": 3.1734,
+      "step": 6145
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 5.665454545454546e-07,
+      "loss": 3.5406,
+      "step": 6146
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 5.658181818181819e-07,
+      "loss": 3.7938,
+      "step": 6147
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 5.650909090909091e-07,
+      "loss": 3.5005,
+      "step": 6148
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 5.643636363636364e-07,
+      "loss": 3.3808,
+      "step": 6149
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 5.636363636363638e-07,
+      "loss": 3.4895,
+      "step": 6150
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 5.62909090909091e-07,
+      "loss": 3.3257,
+      "step": 6151
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 5.621818181818183e-07,
+      "loss": 3.6786,
+      "step": 6152
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 5.614545454545455e-07,
+      "loss": 3.1674,
+      "step": 6153
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 5.607272727272728e-07,
+      "loss": 3.765,
+      "step": 6154
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 5.6e-07,
+      "loss": 3.2871,
+      "step": 6155
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 5.592727272727273e-07,
+      "loss": 3.3383,
+      "step": 6156
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 5.585454545454546e-07,
+      "loss": 3.3024,
+      "step": 6157
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 5.578181818181819e-07,
+      "loss": 3.3747,
+      "step": 6158
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 5.570909090909092e-07,
+      "loss": 3.1736,
+      "step": 6159
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 5.563636363636364e-07,
+      "loss": 3.5491,
+      "step": 6160
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 5.556363636363637e-07,
+      "loss": 3.464,
+      "step": 6161
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 5.549090909090909e-07,
+      "loss": 3.3661,
+      "step": 6162
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 5.541818181818182e-07,
+      "loss": 3.3708,
+      "step": 6163
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 5.534545454545455e-07,
+      "loss": 3.5078,
+      "step": 6164
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 5.527272727272728e-07,
+      "loss": 3.3816,
+      "step": 6165
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 5.520000000000001e-07,
+      "loss": 3.5047,
+      "step": 6166
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 5.512727272727273e-07,
+      "loss": 3.4465,
+      "step": 6167
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 5.505454545454546e-07,
+      "loss": 3.1818,
+      "step": 6168
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 5.498181818181818e-07,
+      "loss": 3.2752,
+      "step": 6169
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 5.490909090909091e-07,
+      "loss": 3.5911,
+      "step": 6170
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 5.483636363636364e-07,
+      "loss": 3.4237,
+      "step": 6171
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 5.476363636363636e-07,
+      "loss": 3.2613,
+      "step": 6172
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 5.46909090909091e-07,
+      "loss": 3.4924,
+      "step": 6173
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 5.461818181818182e-07,
+      "loss": 3.1817,
+      "step": 6174
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 5.454545454545455e-07,
+      "loss": 3.3799,
+      "step": 6175
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 5.447272727272727e-07,
+      "loss": 3.3075,
+      "step": 6176
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 5.44e-07,
+      "loss": 3.1532,
+      "step": 6177
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 5.432727272727273e-07,
+      "loss": 3.2095,
+      "step": 6178
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 5.425454545454545e-07,
+      "loss": 3.2851,
+      "step": 6179
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 5.418181818181819e-07,
+      "loss": 3.4678,
+      "step": 6180
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 5.410909090909091e-07,
+      "loss": 3.268,
+      "step": 6181
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 5.403636363636364e-07,
+      "loss": 3.881,
+      "step": 6182
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 5.396363636363638e-07,
+      "loss": 3.1847,
+      "step": 6183
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 5.38909090909091e-07,
+      "loss": 3.4221,
+      "step": 6184
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 5.381818181818183e-07,
+      "loss": 3.5145,
+      "step": 6185
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 5.374545454545455e-07,
+      "loss": 3.4474,
+      "step": 6186
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 5.367272727272728e-07,
+      "loss": 3.3712,
+      "step": 6187
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 5.36e-07,
+      "loss": 3.2724,
+      "step": 6188
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 5.352727272727273e-07,
+      "loss": 3.3939,
+      "step": 6189
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 5.345454545454547e-07,
+      "loss": 3.5536,
+      "step": 6190
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 5.338181818181819e-07,
+      "loss": 3.2481,
+      "step": 6191
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 5.330909090909092e-07,
+      "loss": 3.3079,
+      "step": 6192
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 5.323636363636364e-07,
+      "loss": 3.2809,
+      "step": 6193
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 5.316363636363637e-07,
+      "loss": 3.3072,
+      "step": 6194
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 5.309090909090909e-07,
+      "loss": 3.5126,
+      "step": 6195
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 5.301818181818182e-07,
+      "loss": 3.1236,
+      "step": 6196
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 5.294545454545455e-07,
+      "loss": 3.3318,
+      "step": 6197
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 5.287272727272728e-07,
+      "loss": 3.2424,
+      "step": 6198
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 5.280000000000001e-07,
+      "loss": 3.1286,
+      "step": 6199
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 5.272727272727273e-07,
+      "loss": 3.6125,
+      "step": 6200
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 5.265454545454546e-07,
+      "loss": 3.28,
+      "step": 6201
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 5.258181818181818e-07,
+      "loss": 3.6364,
+      "step": 6202
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 5.250909090909091e-07,
+      "loss": 3.423,
+      "step": 6203
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 5.243636363636364e-07,
+      "loss": 3.3974,
+      "step": 6204
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 5.236363636363637e-07,
+      "loss": 3.3369,
+      "step": 6205
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 5.22909090909091e-07,
+      "loss": 3.4544,
+      "step": 6206
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 5.221818181818182e-07,
+      "loss": 3.1285,
+      "step": 6207
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 5.214545454545455e-07,
+      "loss": 3.2621,
+      "step": 6208
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 5.207272727272727e-07,
+      "loss": 3.3781,
+      "step": 6209
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 5.2e-07,
+      "loss": 3.394,
+      "step": 6210
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 5.192727272727273e-07,
+      "loss": 3.6367,
+      "step": 6211
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 5.185454545454545e-07,
+      "loss": 3.3371,
+      "step": 6212
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 5.178181818181819e-07,
+      "loss": 3.3636,
+      "step": 6213
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 5.170909090909091e-07,
+      "loss": 3.3584,
+      "step": 6214
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 5.163636363636364e-07,
+      "loss": 3.1952,
+      "step": 6215
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 5.156363636363637e-07,
+      "loss": 3.2571,
+      "step": 6216
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 5.14909090909091e-07,
+      "loss": 3.1662,
+      "step": 6217
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 5.141818181818183e-07,
+      "loss": 3.3996,
+      "step": 6218
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 5.134545454545455e-07,
+      "loss": 3.2763,
+      "step": 6219
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 5.127272727272728e-07,
+      "loss": 3.4797,
+      "step": 6220
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 5.12e-07,
+      "loss": 3.2174,
+      "step": 6221
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 5.112727272727273e-07,
+      "loss": 3.4336,
+      "step": 6222
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 5.105454545454547e-07,
+      "loss": 3.3616,
+      "step": 6223
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 5.098181818181819e-07,
+      "loss": 3.5659,
+      "step": 6224
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 5.090909090909092e-07,
+      "loss": 3.3206,
+      "step": 6225
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 5.083636363636364e-07,
+      "loss": 3.3027,
+      "step": 6226
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 5.076363636363637e-07,
+      "loss": 3.2439,
+      "step": 6227
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 5.069090909090909e-07,
+      "loss": 3.504,
+      "step": 6228
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 5.061818181818182e-07,
+      "loss": 3.0454,
+      "step": 6229
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 5.054545454545456e-07,
+      "loss": 3.0613,
+      "step": 6230
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 5.047272727272728e-07,
+      "loss": 3.508,
+      "step": 6231
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 5.040000000000001e-07,
+      "loss": 3.4022,
+      "step": 6232
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 5.032727272727273e-07,
+      "loss": 3.4548,
+      "step": 6233
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 5.025454545454546e-07,
+      "loss": 3.5602,
+      "step": 6234
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 5.018181818181818e-07,
+      "loss": 3.2065,
+      "step": 6235
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 5.010909090909091e-07,
+      "loss": 3.1644,
+      "step": 6236
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 5.003636363636363e-07,
+      "loss": 3.2835,
+      "step": 6237
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 4.996363636363637e-07,
+      "loss": 3.4009,
+      "step": 6238
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 4.98909090909091e-07,
+      "loss": 3.4939,
+      "step": 6239
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 4.981818181818182e-07,
+      "loss": 3.4482,
+      "step": 6240
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 4.974545454545455e-07,
+      "loss": 3.5794,
+      "step": 6241
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 4.967272727272727e-07,
+      "loss": 3.3435,
+      "step": 6242
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 4.96e-07,
+      "loss": 3.5538,
+      "step": 6243
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 4.952727272727273e-07,
+      "loss": 3.5586,
+      "step": 6244
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 4.945454545454546e-07,
+      "loss": 3.4402,
+      "step": 6245
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 4.938181818181819e-07,
+      "loss": 3.1185,
+      "step": 6246
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 4.930909090909091e-07,
+      "loss": 3.2838,
+      "step": 6247
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 4.923636363636364e-07,
+      "loss": 3.6258,
+      "step": 6248
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 4.916363636363636e-07,
+      "loss": 3.3567,
+      "step": 6249
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 4.909090909090909e-07,
+      "loss": 3.196,
+      "step": 6250
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 4.901818181818182e-07,
+      "loss": 3.3643,
+      "step": 6251
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 4.894545454545454e-07,
+      "loss": 3.2812,
+      "step": 6252
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 4.887272727272728e-07,
+      "loss": 3.3948,
+      "step": 6253
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 4.88e-07,
+      "loss": 3.2378,
+      "step": 6254
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 4.872727272727274e-07,
+      "loss": 3.4858,
+      "step": 6255
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 4.865454545454546e-07,
+      "loss": 3.1821,
+      "step": 6256
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 4.858181818181819e-07,
+      "loss": 3.3238,
+      "step": 6257
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 4.850909090909092e-07,
+      "loss": 3.6708,
+      "step": 6258
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 4.843636363636364e-07,
+      "loss": 3.6561,
+      "step": 6259
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 4.836363636363637e-07,
+      "loss": 3.3523,
+      "step": 6260
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 4.829090909090909e-07,
+      "loss": 3.2665,
+      "step": 6261
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 4.821818181818182e-07,
+      "loss": 3.3527,
+      "step": 6262
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 4.814545454545455e-07,
+      "loss": 3.2662,
+      "step": 6263
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 4.807272727272728e-07,
+      "loss": 3.466,
+      "step": 6264
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 3.3341,
+      "step": 6265
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 4.792727272727273e-07,
+      "loss": 3.6449,
+      "step": 6266
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 4.785454545454546e-07,
+      "loss": 3.2748,
+      "step": 6267
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 4.778181818181818e-07,
+      "loss": 3.3676,
+      "step": 6268
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 4.770909090909091e-07,
+      "loss": 3.5381,
+      "step": 6269
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 4.763636363636364e-07,
+      "loss": 3.3583,
+      "step": 6270
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 4.7563636363636366e-07,
+      "loss": 3.1405,
+      "step": 6271
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 4.7490909090909096e-07,
+      "loss": 3.2031,
+      "step": 6272
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 4.741818181818182e-07,
+      "loss": 3.1666,
+      "step": 6273
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 4.734545454545455e-07,
+      "loss": 3.0758,
+      "step": 6274
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 4.7272727272727273e-07,
+      "loss": 3.2156,
+      "step": 6275
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 4.7200000000000004e-07,
+      "loss": 3.8352,
+      "step": 6276
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 4.712727272727273e-07,
+      "loss": 3.7451,
+      "step": 6277
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 4.7054545454545456e-07,
+      "loss": 3.6001,
+      "step": 6278
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 4.698181818181818e-07,
+      "loss": 3.256,
+      "step": 6279
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 4.690909090909091e-07,
+      "loss": 3.2009,
+      "step": 6280
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 4.683636363636364e-07,
+      "loss": 3.0822,
+      "step": 6281
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 4.6763636363636364e-07,
+      "loss": 3.3057,
+      "step": 6282
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 4.6690909090909095e-07,
+      "loss": 3.4034,
+      "step": 6283
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 4.661818181818182e-07,
+      "loss": 3.3575,
+      "step": 6284
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 4.6545454545454546e-07,
+      "loss": 3.4716,
+      "step": 6285
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 4.647272727272727e-07,
+      "loss": 3.3025,
+      "step": 6286
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 4.64e-07,
+      "loss": 3.3547,
+      "step": 6287
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 4.632727272727273e-07,
+      "loss": 3.3472,
+      "step": 6288
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 4.625454545454546e-07,
+      "loss": 3.5551,
+      "step": 6289
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 4.618181818181819e-07,
+      "loss": 3.344,
+      "step": 6290
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 4.6109090909090916e-07,
+      "loss": 3.3529,
+      "step": 6291
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 4.603636363636364e-07,
+      "loss": 3.2722,
+      "step": 6292
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 4.596363636363637e-07,
+      "loss": 3.6478,
+      "step": 6293
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 4.58909090909091e-07,
+      "loss": 3.5549,
+      "step": 6294
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 4.5818181818181824e-07,
+      "loss": 3.5863,
+      "step": 6295
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 4.574545454545455e-07,
+      "loss": 3.322,
+      "step": 6296
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 4.567272727272728e-07,
+      "loss": 3.1463,
+      "step": 6297
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 4.5600000000000006e-07,
+      "loss": 3.4564,
+      "step": 6298
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 4.552727272727273e-07,
+      "loss": 3.2704,
+      "step": 6299
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 4.5454545454545457e-07,
+      "loss": 3.1729,
+      "step": 6300
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 4.538181818181819e-07,
+      "loss": 3.2825,
+      "step": 6301
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 4.5309090909090914e-07,
+      "loss": 3.2862,
+      "step": 6302
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 4.523636363636364e-07,
+      "loss": 3.5544,
+      "step": 6303
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 4.5163636363636365e-07,
+      "loss": 3.1116,
+      "step": 6304
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 4.5090909090909096e-07,
+      "loss": 3.0908,
+      "step": 6305
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 4.501818181818182e-07,
+      "loss": 3.2542,
+      "step": 6306
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 4.494545454545455e-07,
+      "loss": 3.2542,
+      "step": 6307
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 4.487272727272728e-07,
+      "loss": 3.3542,
+      "step": 6308
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 4.4800000000000004e-07,
+      "loss": 3.2732,
+      "step": 6309
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 4.472727272727273e-07,
+      "loss": 3.6003,
+      "step": 6310
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 4.4654545454545455e-07,
+      "loss": 3.0824,
+      "step": 6311
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 4.4581818181818186e-07,
+      "loss": 3.5872,
+      "step": 6312
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 4.450909090909091e-07,
+      "loss": 3.5114,
+      "step": 6313
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 4.443636363636364e-07,
+      "loss": 3.6612,
+      "step": 6314
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 4.4363636363636363e-07,
+      "loss": 3.0935,
+      "step": 6315
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 4.4290909090909094e-07,
+      "loss": 3.7296,
+      "step": 6316
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 4.421818181818182e-07,
+      "loss": 3.6095,
+      "step": 6317
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 4.4145454545454546e-07,
+      "loss": 3.8273,
+      "step": 6318
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 4.4072727272727277e-07,
+      "loss": 3.2025,
+      "step": 6319
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 4.4e-07,
+      "loss": 3.6576,
+      "step": 6320
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 4.392727272727273e-07,
+      "loss": 3.2946,
+      "step": 6321
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 4.3854545454545454e-07,
+      "loss": 3.5854,
+      "step": 6322
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 4.3781818181818185e-07,
+      "loss": 3.1165,
+      "step": 6323
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 4.370909090909091e-07,
+      "loss": 3.4157,
+      "step": 6324
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 4.363636363636364e-07,
+      "loss": 3.3918,
+      "step": 6325
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 4.356363636363637e-07,
+      "loss": 3.4641,
+      "step": 6326
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 4.34909090909091e-07,
+      "loss": 3.4433,
+      "step": 6327
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 4.3418181818181823e-07,
+      "loss": 3.4086,
+      "step": 6328
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 4.334545454545455e-07,
+      "loss": 3.3186,
+      "step": 6329
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 4.327272727272728e-07,
+      "loss": 3.3408,
+      "step": 6330
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 4.3200000000000006e-07,
+      "loss": 3.2459,
+      "step": 6331
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 4.312727272727273e-07,
+      "loss": 3.2554,
+      "step": 6332
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 4.305454545454546e-07,
+      "loss": 3.3326,
+      "step": 6333
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 4.298181818181819e-07,
+      "loss": 3.4836,
+      "step": 6334
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 4.2909090909090914e-07,
+      "loss": 3.459,
+      "step": 6335
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 4.283636363636364e-07,
+      "loss": 3.2913,
+      "step": 6336
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 4.276363636363637e-07,
+      "loss": 3.452,
+      "step": 6337
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 4.2690909090909096e-07,
+      "loss": 3.2529,
+      "step": 6338
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 4.261818181818182e-07,
+      "loss": 3.4272,
+      "step": 6339
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 4.2545454545454547e-07,
+      "loss": 3.1649,
+      "step": 6340
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 4.247272727272728e-07,
+      "loss": 3.4308,
+      "step": 6341
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 4.2400000000000004e-07,
+      "loss": 3.33,
+      "step": 6342
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 4.232727272727273e-07,
+      "loss": 3.2667,
+      "step": 6343
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 4.225454545454546e-07,
+      "loss": 3.6182,
+      "step": 6344
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 4.2181818181818186e-07,
+      "loss": 3.1882,
+      "step": 6345
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 4.210909090909091e-07,
+      "loss": 3.2747,
+      "step": 6346
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 4.203636363636364e-07,
+      "loss": 3.243,
+      "step": 6347
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 4.196363636363637e-07,
+      "loss": 3.6486,
+      "step": 6348
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 4.1890909090909094e-07,
+      "loss": 3.4034,
+      "step": 6349
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 4.181818181818182e-07,
+      "loss": 3.2698,
+      "step": 6350
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 4.1745454545454545e-07,
+      "loss": 3.2548,
+      "step": 6351
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 4.1672727272727276e-07,
+      "loss": 3.075,
+      "step": 6352
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 4.16e-07,
+      "loss": 3.3923,
+      "step": 6353
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 4.152727272727273e-07,
+      "loss": 3.4254,
+      "step": 6354
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 4.1454545454545453e-07,
+      "loss": 3.5,
+      "step": 6355
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 4.1381818181818184e-07,
+      "loss": 3.1944,
+      "step": 6356
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 4.130909090909091e-07,
+      "loss": 3.234,
+      "step": 6357
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 4.1236363636363636e-07,
+      "loss": 3.0484,
+      "step": 6358
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 4.1163636363636366e-07,
+      "loss": 3.1867,
+      "step": 6359
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 4.109090909090909e-07,
+      "loss": 3.1433,
+      "step": 6360
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 4.101818181818182e-07,
+      "loss": 3.1891,
+      "step": 6361
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 4.0945454545454554e-07,
+      "loss": 3.7759,
+      "step": 6362
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 4.087272727272728e-07,
+      "loss": 3.1389,
+      "step": 6363
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 4.0800000000000005e-07,
+      "loss": 3.1251,
+      "step": 6364
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 4.072727272727273e-07,
+      "loss": 3.3186,
+      "step": 6365
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 4.065454545454546e-07,
+      "loss": 3.3388,
+      "step": 6366
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 4.058181818181819e-07,
+      "loss": 3.8441,
+      "step": 6367
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 4.0509090909090913e-07,
+      "loss": 3.4323,
+      "step": 6368
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 4.0436363636363644e-07,
+      "loss": 3.441,
+      "step": 6369
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 4.036363636363637e-07,
+      "loss": 3.128,
+      "step": 6370
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 4.0290909090909096e-07,
+      "loss": 3.1515,
+      "step": 6371
+    },
+    {
+      "epoch": 23.0,
+      "eval_accuracy": 0.1945945945945946,
+      "eval_loss": 3.3746345043182373,
+      "eval_runtime": 144.5925,
+      "eval_samples_per_second": 2.559,
+      "eval_steps_per_second": 0.643,
+      "step": 6371
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 4.021818181818182e-07,
+      "loss": 3.3883,
+      "step": 6372
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 4.014545454545455e-07,
+      "loss": 3.2027,
+      "step": 6373
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 4.007272727272728e-07,
+      "loss": 3.1194,
+      "step": 6374
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.5314,
+      "step": 6375
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 3.992727272727273e-07,
+      "loss": 3.3178,
+      "step": 6376
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 3.985454545454546e-07,
+      "loss": 3.7032,
+      "step": 6377
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 3.9781818181818186e-07,
+      "loss": 3.2646,
+      "step": 6378
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 3.970909090909091e-07,
+      "loss": 3.6458,
+      "step": 6379
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 3.9636363636363637e-07,
+      "loss": 3.3,
+      "step": 6380
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 3.956363636363637e-07,
+      "loss": 3.5858,
+      "step": 6381
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 3.9490909090909094e-07,
+      "loss": 3.2519,
+      "step": 6382
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 3.941818181818182e-07,
+      "loss": 3.364,
+      "step": 6383
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 3.934545454545455e-07,
+      "loss": 3.3222,
+      "step": 6384
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 3.9272727272727276e-07,
+      "loss": 3.2094,
+      "step": 6385
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 3.92e-07,
+      "loss": 3.2172,
+      "step": 6386
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 3.9127272727272727e-07,
+      "loss": 3.1982,
+      "step": 6387
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 3.905454545454546e-07,
+      "loss": 3.4341,
+      "step": 6388
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 3.8981818181818184e-07,
+      "loss": 3.428,
+      "step": 6389
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 3.890909090909091e-07,
+      "loss": 3.436,
+      "step": 6390
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 3.8836363636363635e-07,
+      "loss": 3.0598,
+      "step": 6391
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 3.8763636363636366e-07,
+      "loss": 3.6623,
+      "step": 6392
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 3.869090909090909e-07,
+      "loss": 3.2758,
+      "step": 6393
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 3.861818181818182e-07,
+      "loss": 3.1947,
+      "step": 6394
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 3.8545454545454543e-07,
+      "loss": 3.4755,
+      "step": 6395
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 3.8472727272727274e-07,
+      "loss": 3.1711,
+      "step": 6396
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 3.84e-07,
+      "loss": 3.3512,
+      "step": 6397
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 3.8327272727272736e-07,
+      "loss": 3.1254,
+      "step": 6398
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 3.825454545454546e-07,
+      "loss": 3.4225,
+      "step": 6399
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 3.8181818181818187e-07,
+      "loss": 3.6648,
+      "step": 6400
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 3.8109090909090913e-07,
+      "loss": 3.3412,
+      "step": 6401
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 3.8036363636363644e-07,
+      "loss": 3.1213,
+      "step": 6402
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 3.796363636363637e-07,
+      "loss": 3.3234,
+      "step": 6403
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 3.7890909090909095e-07,
+      "loss": 3.2847,
+      "step": 6404
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 3.781818181818182e-07,
+      "loss": 3.3863,
+      "step": 6405
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 3.774545454545455e-07,
+      "loss": 3.3234,
+      "step": 6406
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 3.767272727272728e-07,
+      "loss": 3.6359,
+      "step": 6407
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 3.7600000000000003e-07,
+      "loss": 3.4896,
+      "step": 6408
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 3.7527272727272734e-07,
+      "loss": 3.2225,
+      "step": 6409
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 3.745454545454546e-07,
+      "loss": 3.466,
+      "step": 6410
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 3.7381818181818185e-07,
+      "loss": 3.1519,
+      "step": 6411
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 3.730909090909091e-07,
+      "loss": 3.2855,
+      "step": 6412
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 3.723636363636364e-07,
+      "loss": 3.3215,
+      "step": 6413
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 3.716363636363637e-07,
+      "loss": 3.2768,
+      "step": 6414
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 3.7090909090909093e-07,
+      "loss": 3.5646,
+      "step": 6415
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 3.701818181818182e-07,
+      "loss": 3.225,
+      "step": 6416
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 3.694545454545455e-07,
+      "loss": 3.2351,
+      "step": 6417
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 3.6872727272727276e-07,
+      "loss": 3.464,
+      "step": 6418
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 3.68e-07,
+      "loss": 3.5101,
+      "step": 6419
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 3.6727272727272727e-07,
+      "loss": 3.5801,
+      "step": 6420
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 3.665454545454546e-07,
+      "loss": 3.2454,
+      "step": 6421
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 3.6581818181818184e-07,
+      "loss": 3.3792,
+      "step": 6422
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 3.650909090909091e-07,
+      "loss": 3.2132,
+      "step": 6423
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 3.643636363636364e-07,
+      "loss": 3.3314,
+      "step": 6424
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 3.6363636363636366e-07,
+      "loss": 3.1625,
+      "step": 6425
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 3.629090909090909e-07,
+      "loss": 3.5395,
+      "step": 6426
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 3.6218181818181817e-07,
+      "loss": 3.7273,
+      "step": 6427
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 3.614545454545455e-07,
+      "loss": 3.1363,
+      "step": 6428
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 3.6072727272727274e-07,
+      "loss": 3.4337,
+      "step": 6429
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 3.6e-07,
+      "loss": 3.3492,
+      "step": 6430
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 3.5927272727272725e-07,
+      "loss": 3.4152,
+      "step": 6431
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 3.5854545454545456e-07,
+      "loss": 3.4416,
+      "step": 6432
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 3.578181818181818e-07,
+      "loss": 3.5987,
+      "step": 6433
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 3.570909090909092e-07,
+      "loss": 3.5502,
+      "step": 6434
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 3.5636363636363644e-07,
+      "loss": 3.5395,
+      "step": 6435
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 3.556363636363637e-07,
+      "loss": 3.3858,
+      "step": 6436
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 3.5490909090909095e-07,
+      "loss": 3.1871,
+      "step": 6437
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 3.5418181818181826e-07,
+      "loss": 3.7073,
+      "step": 6438
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 3.534545454545455e-07,
+      "loss": 3.1926,
+      "step": 6439
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 3.5272727272727277e-07,
+      "loss": 3.4176,
+      "step": 6440
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 3.5200000000000003e-07,
+      "loss": 3.2474,
+      "step": 6441
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 3.5127272727272734e-07,
+      "loss": 3.2979,
+      "step": 6442
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 3.505454545454546e-07,
+      "loss": 3.1939,
+      "step": 6443
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 3.4981818181818185e-07,
+      "loss": 3.6518,
+      "step": 6444
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 3.490909090909091e-07,
+      "loss": 3.8755,
+      "step": 6445
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 3.483636363636364e-07,
+      "loss": 3.2172,
+      "step": 6446
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 3.476363636363637e-07,
+      "loss": 3.4293,
+      "step": 6447
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 3.4690909090909093e-07,
+      "loss": 3.3223,
+      "step": 6448
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 3.4618181818181824e-07,
+      "loss": 3.1648,
+      "step": 6449
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 3.454545454545455e-07,
+      "loss": 3.2536,
+      "step": 6450
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 3.4472727272727275e-07,
+      "loss": 3.3989,
+      "step": 6451
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 3.44e-07,
+      "loss": 3.2023,
+      "step": 6452
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 3.432727272727273e-07,
+      "loss": 3.2987,
+      "step": 6453
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 3.425454545454546e-07,
+      "loss": 3.0787,
+      "step": 6454
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 3.4181818181818183e-07,
+      "loss": 3.325,
+      "step": 6455
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 3.410909090909091e-07,
+      "loss": 3.4704,
+      "step": 6456
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 3.403636363636364e-07,
+      "loss": 3.3322,
+      "step": 6457
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 3.3963636363636366e-07,
+      "loss": 3.5605,
+      "step": 6458
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 3.389090909090909e-07,
+      "loss": 3.4903,
+      "step": 6459
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 3.3818181818181817e-07,
+      "loss": 3.5697,
+      "step": 6460
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 3.374545454545455e-07,
+      "loss": 3.1154,
+      "step": 6461
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 3.3672727272727273e-07,
+      "loss": 3.3245,
+      "step": 6462
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 3.36e-07,
+      "loss": 3.0738,
+      "step": 6463
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 3.352727272727273e-07,
+      "loss": 3.5892,
+      "step": 6464
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 3.3454545454545456e-07,
+      "loss": 3.2286,
+      "step": 6465
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 3.338181818181818e-07,
+      "loss": 3.6021,
+      "step": 6466
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 3.3309090909090907e-07,
+      "loss": 3.5014,
+      "step": 6467
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 3.323636363636364e-07,
+      "loss": 3.5059,
+      "step": 6468
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 3.3163636363636364e-07,
+      "loss": 3.3215,
+      "step": 6469
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 3.309090909090909e-07,
+      "loss": 3.3032,
+      "step": 6470
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 3.3018181818181826e-07,
+      "loss": 3.0453,
+      "step": 6471
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 3.294545454545455e-07,
+      "loss": 3.0918,
+      "step": 6472
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 3.2872727272727277e-07,
+      "loss": 3.2786,
+      "step": 6473
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 3.280000000000001e-07,
+      "loss": 3.3247,
+      "step": 6474
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 3.2727272727272733e-07,
+      "loss": 3.232,
+      "step": 6475
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 3.265454545454546e-07,
+      "loss": 3.2733,
+      "step": 6476
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 3.2581818181818185e-07,
+      "loss": 3.4696,
+      "step": 6477
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 3.2509090909090916e-07,
+      "loss": 3.5608,
+      "step": 6478
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 3.243636363636364e-07,
+      "loss": 3.1592,
+      "step": 6479
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 3.2363636363636367e-07,
+      "loss": 3.3946,
+      "step": 6480
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 3.2290909090909093e-07,
+      "loss": 3.3746,
+      "step": 6481
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 3.2218181818181824e-07,
+      "loss": 3.2288,
+      "step": 6482
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 3.214545454545455e-07,
+      "loss": 3.2286,
+      "step": 6483
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 3.2072727272727275e-07,
+      "loss": 3.2115,
+      "step": 6484
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 3.2e-07,
+      "loss": 3.3614,
+      "step": 6485
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 3.192727272727273e-07,
+      "loss": 3.2061,
+      "step": 6486
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 3.1854545454545457e-07,
+      "loss": 3.6129,
+      "step": 6487
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 3.1781818181818183e-07,
+      "loss": 3.5696,
+      "step": 6488
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 3.1709090909090914e-07,
+      "loss": 3.2743,
+      "step": 6489
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 3.163636363636364e-07,
+      "loss": 3.1926,
+      "step": 6490
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 3.1563636363636365e-07,
+      "loss": 3.5052,
+      "step": 6491
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 3.149090909090909e-07,
+      "loss": 3.3703,
+      "step": 6492
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 3.141818181818182e-07,
+      "loss": 3.2787,
+      "step": 6493
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 3.134545454545455e-07,
+      "loss": 3.3566,
+      "step": 6494
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 3.1272727272727273e-07,
+      "loss": 3.4483,
+      "step": 6495
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 3.12e-07,
+      "loss": 3.3021,
+      "step": 6496
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 3.112727272727273e-07,
+      "loss": 3.3812,
+      "step": 6497
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 3.105454545454546e-07,
+      "loss": 3.4549,
+      "step": 6498
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 3.0981818181818186e-07,
+      "loss": 3.7728,
+      "step": 6499
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 3.090909090909091e-07,
+      "loss": 3.3612,
+      "step": 6500
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 3.083636363636364e-07,
+      "loss": 3.2233,
+      "step": 6501
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 3.076363636363637e-07,
+      "loss": 3.3217,
+      "step": 6502
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 3.0690909090909094e-07,
+      "loss": 3.2915,
+      "step": 6503
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 3.061818181818182e-07,
+      "loss": 3.2234,
+      "step": 6504
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 3.054545454545455e-07,
+      "loss": 3.5068,
+      "step": 6505
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 3.0472727272727277e-07,
+      "loss": 3.3016,
+      "step": 6506
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 3.04e-07,
+      "loss": 3.2167,
+      "step": 6507
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 3.032727272727273e-07,
+      "loss": 3.4193,
+      "step": 6508
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 3.025454545454546e-07,
+      "loss": 3.2821,
+      "step": 6509
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 3.0181818181818184e-07,
+      "loss": 3.2113,
+      "step": 6510
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 3.010909090909091e-07,
+      "loss": 3.8151,
+      "step": 6511
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 3.0036363636363636e-07,
+      "loss": 3.2222,
+      "step": 6512
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 2.9963636363636367e-07,
+      "loss": 3.2131,
+      "step": 6513
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 2.989090909090909e-07,
+      "loss": 3.7898,
+      "step": 6514
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 2.981818181818182e-07,
+      "loss": 3.2746,
+      "step": 6515
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 2.974545454545455e-07,
+      "loss": 3.1067,
+      "step": 6516
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 2.9672727272727275e-07,
+      "loss": 3.2296,
+      "step": 6517
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 2.9600000000000006e-07,
+      "loss": 3.3588,
+      "step": 6518
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 2.952727272727273e-07,
+      "loss": 3.2318,
+      "step": 6519
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 2.9454545454545457e-07,
+      "loss": 3.1798,
+      "step": 6520
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 2.938181818181818e-07,
+      "loss": 3.6125,
+      "step": 6521
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 2.9309090909090914e-07,
+      "loss": 3.2866,
+      "step": 6522
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 2.923636363636364e-07,
+      "loss": 3.4342,
+      "step": 6523
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 2.9163636363636365e-07,
+      "loss": 3.4279,
+      "step": 6524
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 2.9090909090909096e-07,
+      "loss": 3.5044,
+      "step": 6525
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 2.901818181818182e-07,
+      "loss": 3.2085,
+      "step": 6526
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 2.8945454545454547e-07,
+      "loss": 3.5592,
+      "step": 6527
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 2.8872727272727273e-07,
+      "loss": 3.41,
+      "step": 6528
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 2.8800000000000004e-07,
+      "loss": 3.0398,
+      "step": 6529
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 2.872727272727273e-07,
+      "loss": 3.3284,
+      "step": 6530
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 2.8654545454545455e-07,
+      "loss": 3.5595,
+      "step": 6531
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 2.858181818181818e-07,
+      "loss": 3.2239,
+      "step": 6532
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 2.850909090909091e-07,
+      "loss": 3.5336,
+      "step": 6533
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 2.8436363636363643e-07,
+      "loss": 3.6174,
+      "step": 6534
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 2.836363636363637e-07,
+      "loss": 3.1937,
+      "step": 6535
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 2.8290909090909094e-07,
+      "loss": 3.4618,
+      "step": 6536
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 2.821818181818182e-07,
+      "loss": 3.5276,
+      "step": 6537
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 2.814545454545455e-07,
+      "loss": 3.247,
+      "step": 6538
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 2.8072727272727276e-07,
+      "loss": 3.4614,
+      "step": 6539
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 2.8e-07,
+      "loss": 3.0733,
+      "step": 6540
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 2.792727272727273e-07,
+      "loss": 3.4341,
+      "step": 6541
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 2.785454545454546e-07,
+      "loss": 3.4097,
+      "step": 6542
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 2.7781818181818184e-07,
+      "loss": 3.3273,
+      "step": 6543
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 2.770909090909091e-07,
+      "loss": 3.2603,
+      "step": 6544
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 2.763636363636364e-07,
+      "loss": 3.1702,
+      "step": 6545
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 2.7563636363636366e-07,
+      "loss": 3.44,
+      "step": 6546
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 2.749090909090909e-07,
+      "loss": 3.6411,
+      "step": 6547
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 2.741818181818182e-07,
+      "loss": 3.4639,
+      "step": 6548
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 2.734545454545455e-07,
+      "loss": 3.4268,
+      "step": 6549
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 2.7272727272727274e-07,
+      "loss": 3.2201,
+      "step": 6550
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 2.72e-07,
+      "loss": 3.1612,
+      "step": 6551
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 2.7127272727272726e-07,
+      "loss": 3.5254,
+      "step": 6552
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 2.7054545454545457e-07,
+      "loss": 3.3175,
+      "step": 6553
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 2.698181818181819e-07,
+      "loss": 3.3916,
+      "step": 6554
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 2.6909090909090913e-07,
+      "loss": 3.5022,
+      "step": 6555
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 2.683636363636364e-07,
+      "loss": 3.3637,
+      "step": 6556
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 2.6763636363636365e-07,
+      "loss": 3.7644,
+      "step": 6557
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 2.6690909090909096e-07,
+      "loss": 3.672,
+      "step": 6558
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 2.661818181818182e-07,
+      "loss": 3.3125,
+      "step": 6559
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 2.6545454545454547e-07,
+      "loss": 3.3107,
+      "step": 6560
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 2.647272727272727e-07,
+      "loss": 3.2213,
+      "step": 6561
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 2.6400000000000003e-07,
+      "loss": 3.1988,
+      "step": 6562
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 2.632727272727273e-07,
+      "loss": 3.4923,
+      "step": 6563
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 2.6254545454545455e-07,
+      "loss": 3.3196,
+      "step": 6564
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 2.6181818181818186e-07,
+      "loss": 3.1485,
+      "step": 6565
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 2.610909090909091e-07,
+      "loss": 3.8345,
+      "step": 6566
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 2.6036363636363637e-07,
+      "loss": 3.459,
+      "step": 6567
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 2.5963636363636363e-07,
+      "loss": 3.4613,
+      "step": 6568
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 2.5890909090909094e-07,
+      "loss": 3.177,
+      "step": 6569
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 2.581818181818182e-07,
+      "loss": 3.6667,
+      "step": 6570
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 2.574545454545455e-07,
+      "loss": 3.5397,
+      "step": 6571
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 2.5672727272727276e-07,
+      "loss": 3.1389,
+      "step": 6572
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 2.56e-07,
+      "loss": 3.229,
+      "step": 6573
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 2.552727272727273e-07,
+      "loss": 3.2531,
+      "step": 6574
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 2.545454545454546e-07,
+      "loss": 3.4569,
+      "step": 6575
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 2.5381818181818184e-07,
+      "loss": 3.5554,
+      "step": 6576
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 2.530909090909091e-07,
+      "loss": 3.1473,
+      "step": 6577
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 2.523636363636364e-07,
+      "loss": 3.4394,
+      "step": 6578
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 2.5163636363636366e-07,
+      "loss": 3.3599,
+      "step": 6579
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 2.509090909090909e-07,
+      "loss": 3.2774,
+      "step": 6580
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 2.501818181818182e-07,
+      "loss": 3.3736,
+      "step": 6581
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 2.494545454545455e-07,
+      "loss": 3.4906,
+      "step": 6582
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 2.4872727272727274e-07,
+      "loss": 3.4984,
+      "step": 6583
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 2.48e-07,
+      "loss": 3.233,
+      "step": 6584
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 2.472727272727273e-07,
+      "loss": 3.3389,
+      "step": 6585
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 2.4654545454545456e-07,
+      "loss": 3.4606,
+      "step": 6586
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 2.458181818181818e-07,
+      "loss": 3.4767,
+      "step": 6587
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 2.450909090909091e-07,
+      "loss": 3.2702,
+      "step": 6588
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 2.443636363636364e-07,
+      "loss": 3.4438,
+      "step": 6589
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 2.436363636363637e-07,
+      "loss": 3.231,
+      "step": 6590
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 2.4290909090909095e-07,
+      "loss": 3.3543,
+      "step": 6591
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 2.421818181818182e-07,
+      "loss": 3.1279,
+      "step": 6592
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 2.4145454545454547e-07,
+      "loss": 3.2847,
+      "step": 6593
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 2.407272727272728e-07,
+      "loss": 3.2022,
+      "step": 6594
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 3.2184,
+      "step": 6595
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 2.392727272727273e-07,
+      "loss": 3.5801,
+      "step": 6596
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 2.3854545454545454e-07,
+      "loss": 3.2269,
+      "step": 6597
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 2.3781818181818183e-07,
+      "loss": 3.2497,
+      "step": 6598
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 2.370909090909091e-07,
+      "loss": 3.4035,
+      "step": 6599
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 2.3636363636363637e-07,
+      "loss": 3.3157,
+      "step": 6600
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 2.3563636363636365e-07,
+      "loss": 3.0049,
+      "step": 6601
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 2.349090909090909e-07,
+      "loss": 3.4258,
+      "step": 6602
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 2.341818181818182e-07,
+      "loss": 3.6956,
+      "step": 6603
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 2.3345454545454547e-07,
+      "loss": 3.2903,
+      "step": 6604
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 2.3272727272727273e-07,
+      "loss": 3.5685,
+      "step": 6605
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 2.32e-07,
+      "loss": 3.356,
+      "step": 6606
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 2.312727272727273e-07,
+      "loss": 3.3182,
+      "step": 6607
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 2.3054545454545458e-07,
+      "loss": 3.8546,
+      "step": 6608
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 2.2981818181818186e-07,
+      "loss": 3.3866,
+      "step": 6609
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 2.2909090909090912e-07,
+      "loss": 3.2643,
+      "step": 6610
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 2.283636363636364e-07,
+      "loss": 3.5128,
+      "step": 6611
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 2.2763636363636366e-07,
+      "loss": 3.4435,
+      "step": 6612
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 2.2690909090909094e-07,
+      "loss": 3.3256,
+      "step": 6613
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 2.261818181818182e-07,
+      "loss": 3.3485,
+      "step": 6614
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 2.2545454545454548e-07,
+      "loss": 3.5898,
+      "step": 6615
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 2.2472727272727274e-07,
+      "loss": 3.4839,
+      "step": 6616
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 2.2400000000000002e-07,
+      "loss": 3.3915,
+      "step": 6617
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 2.2327272727272728e-07,
+      "loss": 3.5603,
+      "step": 6618
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 2.2254545454545456e-07,
+      "loss": 3.4085,
+      "step": 6619
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 2.2181818181818182e-07,
+      "loss": 3.2774,
+      "step": 6620
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 2.210909090909091e-07,
+      "loss": 3.3013,
+      "step": 6621
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 2.2036363636363638e-07,
+      "loss": 3.4924,
+      "step": 6622
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 2.1963636363636364e-07,
+      "loss": 3.0605,
+      "step": 6623
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 2.1890909090909092e-07,
+      "loss": 3.5405,
+      "step": 6624
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 2.181818181818182e-07,
+      "loss": 3.1341,
+      "step": 6625
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 2.174545454545455e-07,
+      "loss": 3.626,
+      "step": 6626
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 2.1672727272727275e-07,
+      "loss": 3.3995,
+      "step": 6627
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 2.1600000000000003e-07,
+      "loss": 3.3225,
+      "step": 6628
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 2.152727272727273e-07,
+      "loss": 3.18,
+      "step": 6629
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 2.1454545454545457e-07,
+      "loss": 3.3884,
+      "step": 6630
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 2.1381818181818185e-07,
+      "loss": 3.4583,
+      "step": 6631
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 2.130909090909091e-07,
+      "loss": 3.4207,
+      "step": 6632
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 2.123636363636364e-07,
+      "loss": 3.1595,
+      "step": 6633
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 2.1163636363636365e-07,
+      "loss": 3.1853,
+      "step": 6634
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 2.1090909090909093e-07,
+      "loss": 3.2987,
+      "step": 6635
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 2.101818181818182e-07,
+      "loss": 3.2716,
+      "step": 6636
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 2.0945454545454547e-07,
+      "loss": 3.4413,
+      "step": 6637
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 2.0872727272727273e-07,
+      "loss": 3.3272,
+      "step": 6638
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 2.08e-07,
+      "loss": 3.2025,
+      "step": 6639
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 2.0727272727272727e-07,
+      "loss": 3.5371,
+      "step": 6640
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 2.0654545454545455e-07,
+      "loss": 3.3651,
+      "step": 6641
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 2.0581818181818183e-07,
+      "loss": 3.2301,
+      "step": 6642
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 2.050909090909091e-07,
+      "loss": 3.5502,
+      "step": 6643
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 2.043636363636364e-07,
+      "loss": 3.284,
+      "step": 6644
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 2.0363636363636366e-07,
+      "loss": 3.4659,
+      "step": 6645
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 2.0290909090909094e-07,
+      "loss": 3.7022,
+      "step": 6646
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 2.0218181818181822e-07,
+      "loss": 3.4003,
+      "step": 6647
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 2.0145454545454548e-07,
+      "loss": 3.264,
+      "step": 6648
+    },
+    {
+      "epoch": 24.0,
+      "eval_accuracy": 0.1864864864864865,
+      "eval_loss": 3.373474359512329,
+      "eval_runtime": 143.7017,
+      "eval_samples_per_second": 2.575,
+      "eval_steps_per_second": 0.647,
+      "step": 6648
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 2.0072727272727276e-07,
+      "loss": 3.2082,
+      "step": 6649
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.512,
+      "step": 6650
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.992727272727273e-07,
+      "loss": 3.1715,
+      "step": 6651
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.9854545454545456e-07,
+      "loss": 3.9258,
+      "step": 6652
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.9781818181818184e-07,
+      "loss": 3.6674,
+      "step": 6653
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.970909090909091e-07,
+      "loss": 3.2933,
+      "step": 6654
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.9636363636363638e-07,
+      "loss": 3.0996,
+      "step": 6655
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.9563636363636364e-07,
+      "loss": 3.183,
+      "step": 6656
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.9490909090909092e-07,
+      "loss": 3.2185,
+      "step": 6657
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.9418181818181818e-07,
+      "loss": 3.3966,
+      "step": 6658
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.9345454545454546e-07,
+      "loss": 3.4824,
+      "step": 6659
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.9272727272727272e-07,
+      "loss": 3.76,
+      "step": 6660
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.92e-07,
+      "loss": 3.1397,
+      "step": 6661
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.912727272727273e-07,
+      "loss": 3.2339,
+      "step": 6662
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.9054545454545456e-07,
+      "loss": 3.0968,
+      "step": 6663
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.8981818181818185e-07,
+      "loss": 3.5954,
+      "step": 6664
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.890909090909091e-07,
+      "loss": 3.2165,
+      "step": 6665
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.883636363636364e-07,
+      "loss": 3.2715,
+      "step": 6666
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.8763636363636367e-07,
+      "loss": 3.4556,
+      "step": 6667
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.8690909090909093e-07,
+      "loss": 3.6036,
+      "step": 6668
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.861818181818182e-07,
+      "loss": 3.2541,
+      "step": 6669
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.8545454545454547e-07,
+      "loss": 3.1147,
+      "step": 6670
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.8472727272727275e-07,
+      "loss": 3.2745,
+      "step": 6671
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.84e-07,
+      "loss": 3.2748,
+      "step": 6672
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.832727272727273e-07,
+      "loss": 3.4674,
+      "step": 6673
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.8254545454545455e-07,
+      "loss": 3.0888,
+      "step": 6674
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.8181818181818183e-07,
+      "loss": 3.4938,
+      "step": 6675
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.8109090909090909e-07,
+      "loss": 3.5023,
+      "step": 6676
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.8036363636363637e-07,
+      "loss": 3.1508,
+      "step": 6677
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.7963636363636363e-07,
+      "loss": 3.4621,
+      "step": 6678
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.789090909090909e-07,
+      "loss": 3.578,
+      "step": 6679
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.7818181818181822e-07,
+      "loss": 3.3147,
+      "step": 6680
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.7745454545454547e-07,
+      "loss": 3.2938,
+      "step": 6681
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.7672727272727276e-07,
+      "loss": 3.6293,
+      "step": 6682
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.7600000000000001e-07,
+      "loss": 3.4298,
+      "step": 6683
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.752727272727273e-07,
+      "loss": 3.6139,
+      "step": 6684
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.7454545454545455e-07,
+      "loss": 3.3329,
+      "step": 6685
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.7381818181818184e-07,
+      "loss": 3.2261,
+      "step": 6686
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.7309090909090912e-07,
+      "loss": 3.5783,
+      "step": 6687
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.7236363636363638e-07,
+      "loss": 3.1012,
+      "step": 6688
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.7163636363636366e-07,
+      "loss": 3.5224,
+      "step": 6689
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.7090909090909092e-07,
+      "loss": 3.5694,
+      "step": 6690
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.701818181818182e-07,
+      "loss": 3.4125,
+      "step": 6691
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.6945454545454546e-07,
+      "loss": 3.4211,
+      "step": 6692
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.6872727272727274e-07,
+      "loss": 3.3928,
+      "step": 6693
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.68e-07,
+      "loss": 3.1666,
+      "step": 6694
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.6727272727272728e-07,
+      "loss": 3.2952,
+      "step": 6695
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.6654545454545454e-07,
+      "loss": 3.1539,
+      "step": 6696
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.6581818181818182e-07,
+      "loss": 3.5078,
+      "step": 6697
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.6509090909090913e-07,
+      "loss": 3.3589,
+      "step": 6698
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.6436363636363638e-07,
+      "loss": 3.2361,
+      "step": 6699
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.6363636363636367e-07,
+      "loss": 3.1635,
+      "step": 6700
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.6290909090909092e-07,
+      "loss": 3.3774,
+      "step": 6701
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.621818181818182e-07,
+      "loss": 3.6863,
+      "step": 6702
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.6145454545454546e-07,
+      "loss": 3.1284,
+      "step": 6703
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.6072727272727275e-07,
+      "loss": 3.4847,
+      "step": 6704
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.6e-07,
+      "loss": 3.2707,
+      "step": 6705
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.5927272727272729e-07,
+      "loss": 3.1869,
+      "step": 6706
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.5854545454545457e-07,
+      "loss": 3.2891,
+      "step": 6707
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.5781818181818183e-07,
+      "loss": 3.2566,
+      "step": 6708
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.570909090909091e-07,
+      "loss": 3.2065,
+      "step": 6709
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.5636363636363637e-07,
+      "loss": 3.6344,
+      "step": 6710
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.5563636363636365e-07,
+      "loss": 3.2869,
+      "step": 6711
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.5490909090909093e-07,
+      "loss": 3.3321,
+      "step": 6712
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.541818181818182e-07,
+      "loss": 3.583,
+      "step": 6713
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.5345454545454547e-07,
+      "loss": 3.388,
+      "step": 6714
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.5272727272727275e-07,
+      "loss": 3.3919,
+      "step": 6715
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.52e-07,
+      "loss": 3.2326,
+      "step": 6716
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.512727272727273e-07,
+      "loss": 3.493,
+      "step": 6717
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.5054545454545455e-07,
+      "loss": 3.1418,
+      "step": 6718
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4981818181818183e-07,
+      "loss": 3.3791,
+      "step": 6719
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.490909090909091e-07,
+      "loss": 3.4749,
+      "step": 6720
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4836363636363637e-07,
+      "loss": 3.1695,
+      "step": 6721
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4763636363636366e-07,
+      "loss": 3.3087,
+      "step": 6722
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.469090909090909e-07,
+      "loss": 3.5057,
+      "step": 6723
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.461818181818182e-07,
+      "loss": 3.1673,
+      "step": 6724
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4545454545454548e-07,
+      "loss": 3.1498,
+      "step": 6725
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4472727272727274e-07,
+      "loss": 3.2617,
+      "step": 6726
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4400000000000002e-07,
+      "loss": 3.0483,
+      "step": 6727
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4327272727272728e-07,
+      "loss": 3.3675,
+      "step": 6728
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4254545454545456e-07,
+      "loss": 3.0827,
+      "step": 6729
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4181818181818184e-07,
+      "loss": 3.2473,
+      "step": 6730
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.410909090909091e-07,
+      "loss": 3.1246,
+      "step": 6731
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4036363636363638e-07,
+      "loss": 3.4311,
+      "step": 6732
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.3963636363636364e-07,
+      "loss": 3.6749,
+      "step": 6733
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.3890909090909092e-07,
+      "loss": 3.3617,
+      "step": 6734
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.381818181818182e-07,
+      "loss": 3.4743,
+      "step": 6735
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.3745454545454546e-07,
+      "loss": 3.3684,
+      "step": 6736
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.3672727272727274e-07,
+      "loss": 3.3905,
+      "step": 6737
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.36e-07,
+      "loss": 3.2047,
+      "step": 6738
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.3527272727272728e-07,
+      "loss": 3.3331,
+      "step": 6739
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.3454545454545457e-07,
+      "loss": 3.2904,
+      "step": 6740
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.3381818181818182e-07,
+      "loss": 3.4024,
+      "step": 6741
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.330909090909091e-07,
+      "loss": 3.3996,
+      "step": 6742
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.3236363636363636e-07,
+      "loss": 3.3375,
+      "step": 6743
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.3163636363636365e-07,
+      "loss": 3.1454,
+      "step": 6744
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.3090909090909093e-07,
+      "loss": 3.573,
+      "step": 6745
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.3018181818181819e-07,
+      "loss": 3.3146,
+      "step": 6746
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.2945454545454547e-07,
+      "loss": 3.5324,
+      "step": 6747
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.2872727272727275e-07,
+      "loss": 3.7437,
+      "step": 6748
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.28e-07,
+      "loss": 3.2373,
+      "step": 6749
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.272727272727273e-07,
+      "loss": 3.2934,
+      "step": 6750
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.2654545454545455e-07,
+      "loss": 3.1669,
+      "step": 6751
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.2581818181818183e-07,
+      "loss": 3.2527,
+      "step": 6752
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.250909090909091e-07,
+      "loss": 3.7402,
+      "step": 6753
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.2436363636363637e-07,
+      "loss": 3.7375,
+      "step": 6754
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.2363636363636365e-07,
+      "loss": 3.3192,
+      "step": 6755
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.229090909090909e-07,
+      "loss": 3.3617,
+      "step": 6756
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.221818181818182e-07,
+      "loss": 3.3998,
+      "step": 6757
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.2145454545454548e-07,
+      "loss": 3.3414,
+      "step": 6758
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.2072727272727273e-07,
+      "loss": 3.3784,
+      "step": 6759
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 3.1626,
+      "step": 6760
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.1927272727272727e-07,
+      "loss": 3.3729,
+      "step": 6761
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.1854545454545456e-07,
+      "loss": 3.4494,
+      "step": 6762
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.1781818181818183e-07,
+      "loss": 3.4183,
+      "step": 6763
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.170909090909091e-07,
+      "loss": 3.1939,
+      "step": 6764
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.1636363636363636e-07,
+      "loss": 3.3712,
+      "step": 6765
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.1563636363636365e-07,
+      "loss": 3.2086,
+      "step": 6766
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.1490909090909093e-07,
+      "loss": 3.523,
+      "step": 6767
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.141818181818182e-07,
+      "loss": 3.4928,
+      "step": 6768
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.1345454545454547e-07,
+      "loss": 3.2939,
+      "step": 6769
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.1272727272727274e-07,
+      "loss": 3.1034,
+      "step": 6770
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.1200000000000001e-07,
+      "loss": 3.0739,
+      "step": 6771
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.1127272727272728e-07,
+      "loss": 3.5017,
+      "step": 6772
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.1054545454545455e-07,
+      "loss": 3.3635,
+      "step": 6773
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.0981818181818182e-07,
+      "loss": 3.3635,
+      "step": 6774
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.090909090909091e-07,
+      "loss": 3.2187,
+      "step": 6775
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.0836363636363637e-07,
+      "loss": 3.2005,
+      "step": 6776
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.0763636363636366e-07,
+      "loss": 3.1687,
+      "step": 6777
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.0690909090909093e-07,
+      "loss": 3.379,
+      "step": 6778
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.061818181818182e-07,
+      "loss": 3.3145,
+      "step": 6779
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.0545454545454547e-07,
+      "loss": 3.4359,
+      "step": 6780
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.0472727272727273e-07,
+      "loss": 3.3413,
+      "step": 6781
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.04e-07,
+      "loss": 3.1121,
+      "step": 6782
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.0327272727272727e-07,
+      "loss": 3.4082,
+      "step": 6783
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.0254545454545454e-07,
+      "loss": 3.611,
+      "step": 6784
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.0181818181818183e-07,
+      "loss": 3.4682,
+      "step": 6785
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.0109090909090911e-07,
+      "loss": 3.5872,
+      "step": 6786
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.0036363636363638e-07,
+      "loss": 3.6837,
+      "step": 6787
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 9.963636363636365e-08,
+      "loss": 3.4929,
+      "step": 6788
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 9.890909090909092e-08,
+      "loss": 3.3406,
+      "step": 6789
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 9.818181818181819e-08,
+      "loss": 3.2879,
+      "step": 6790
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 9.745454545454546e-08,
+      "loss": 3.1972,
+      "step": 6791
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 9.672727272727273e-08,
+      "loss": 3.1469,
+      "step": 6792
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 9.6e-08,
+      "loss": 3.2763,
+      "step": 6793
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 9.527272727272728e-08,
+      "loss": 3.3523,
+      "step": 6794
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 9.454545454545455e-08,
+      "loss": 3.5321,
+      "step": 6795
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 9.381818181818184e-08,
+      "loss": 3.2664,
+      "step": 6796
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 9.30909090909091e-08,
+      "loss": 3.5289,
+      "step": 6797
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 9.236363636363637e-08,
+      "loss": 3.4555,
+      "step": 6798
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 9.163636363636364e-08,
+      "loss": 3.4443,
+      "step": 6799
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 9.090909090909091e-08,
+      "loss": 3.333,
+      "step": 6800
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 9.018181818181818e-08,
+      "loss": 3.3459,
+      "step": 6801
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 8.945454545454545e-08,
+      "loss": 3.2237,
+      "step": 6802
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 8.872727272727274e-08,
+      "loss": 3.2026,
+      "step": 6803
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 8.800000000000001e-08,
+      "loss": 3.3145,
+      "step": 6804
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 8.727272727272728e-08,
+      "loss": 3.4572,
+      "step": 6805
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 8.654545454545456e-08,
+      "loss": 3.3647,
+      "step": 6806
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 8.581818181818183e-08,
+      "loss": 3.6535,
+      "step": 6807
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 8.50909090909091e-08,
+      "loss": 3.2911,
+      "step": 6808
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 8.436363636363637e-08,
+      "loss": 3.1443,
+      "step": 6809
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 8.363636363636364e-08,
+      "loss": 3.2921,
+      "step": 6810
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 8.290909090909091e-08,
+      "loss": 3.2017,
+      "step": 6811
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 8.218181818181819e-08,
+      "loss": 3.3599,
+      "step": 6812
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 8.145454545454546e-08,
+      "loss": 3.3763,
+      "step": 6813
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 8.072727272727273e-08,
+      "loss": 3.4394,
+      "step": 6814
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 8e-08,
+      "loss": 3.6766,
+      "step": 6815
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 7.927272727272728e-08,
+      "loss": 3.3927,
+      "step": 6816
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 7.854545454545455e-08,
+      "loss": 3.5376,
+      "step": 6817
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 7.781818181818182e-08,
+      "loss": 3.413,
+      "step": 6818
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 7.70909090909091e-08,
+      "loss": 3.2861,
+      "step": 6819
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 7.636363636363638e-08,
+      "loss": 3.4798,
+      "step": 6820
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 7.563636363636365e-08,
+      "loss": 3.2258,
+      "step": 6821
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 7.490909090909092e-08,
+      "loss": 3.1916,
+      "step": 6822
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 7.418181818181819e-08,
+      "loss": 3.2183,
+      "step": 6823
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 7.345454545454546e-08,
+      "loss": 3.1842,
+      "step": 6824
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 7.272727272727274e-08,
+      "loss": 3.3563,
+      "step": 6825
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 7.200000000000001e-08,
+      "loss": 3.3116,
+      "step": 6826
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 7.127272727272728e-08,
+      "loss": 3.3577,
+      "step": 6827
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 7.054545454545455e-08,
+      "loss": 3.4399,
+      "step": 6828
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 6.981818181818182e-08,
+      "loss": 3.4708,
+      "step": 6829
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 6.90909090909091e-08,
+      "loss": 3.2811,
+      "step": 6830
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 6.836363636363637e-08,
+      "loss": 3.2779,
+      "step": 6831
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 6.763636363636364e-08,
+      "loss": 3.8085,
+      "step": 6832
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 6.690909090909091e-08,
+      "loss": 3.1397,
+      "step": 6833
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 6.618181818181818e-08,
+      "loss": 3.3003,
+      "step": 6834
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 6.545454545454546e-08,
+      "loss": 3.492,
+      "step": 6835
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 6.472727272727273e-08,
+      "loss": 3.4712,
+      "step": 6836
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 6.4e-08,
+      "loss": 3.731,
+      "step": 6837
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 6.327272727272727e-08,
+      "loss": 3.5439,
+      "step": 6838
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 6.254545454545454e-08,
+      "loss": 3.2146,
+      "step": 6839
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 6.181818181818183e-08,
+      "loss": 3.231,
+      "step": 6840
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 6.10909090909091e-08,
+      "loss": 3.1649,
+      "step": 6841
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 6.036363636363637e-08,
+      "loss": 3.2106,
+      "step": 6842
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 5.963636363636364e-08,
+      "loss": 3.5501,
+      "step": 6843
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 5.890909090909091e-08,
+      "loss": 3.4595,
+      "step": 6844
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 5.818181818181818e-08,
+      "loss": 3.2498,
+      "step": 6845
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 5.7454545454545465e-08,
+      "loss": 3.3573,
+      "step": 6846
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 5.6727272727272735e-08,
+      "loss": 3.5953,
+      "step": 6847
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 5.6000000000000005e-08,
+      "loss": 3.2448,
+      "step": 6848
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 5.5272727272727275e-08,
+      "loss": 3.8319,
+      "step": 6849
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 5.454545454545455e-08,
+      "loss": 3.6117,
+      "step": 6850
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 5.381818181818183e-08,
+      "loss": 3.3135,
+      "step": 6851
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 5.30909090909091e-08,
+      "loss": 3.1617,
+      "step": 6852
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 5.236363636363637e-08,
+      "loss": 3.4451,
+      "step": 6853
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 5.163636363636364e-08,
+      "loss": 3.4549,
+      "step": 6854
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 5.0909090909090914e-08,
+      "loss": 3.4443,
+      "step": 6855
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 5.018181818181819e-08,
+      "loss": 3.2179,
+      "step": 6856
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 4.945454545454546e-08,
+      "loss": 3.6667,
+      "step": 6857
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 4.872727272727273e-08,
+      "loss": 3.6518,
+      "step": 6858
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 4.8e-08,
+      "loss": 3.3965,
+      "step": 6859
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 4.7272727272727276e-08,
+      "loss": 3.17,
+      "step": 6860
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 4.654545454545455e-08,
+      "loss": 3.3428,
+      "step": 6861
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 4.581818181818182e-08,
+      "loss": 3.1785,
+      "step": 6862
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 4.509090909090909e-08,
+      "loss": 3.2245,
+      "step": 6863
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 4.436363636363637e-08,
+      "loss": 3.5343,
+      "step": 6864
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 4.363636363636364e-08,
+      "loss": 3.0777,
+      "step": 6865
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 4.2909090909090915e-08,
+      "loss": 3.5459,
+      "step": 6866
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 4.2181818181818185e-08,
+      "loss": 3.2976,
+      "step": 6867
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 4.1454545454545455e-08,
+      "loss": 3.0854,
+      "step": 6868
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 4.072727272727273e-08,
+      "loss": 3.3115,
+      "step": 6869
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 4e-08,
+      "loss": 3.2922,
+      "step": 6870
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 3.927272727272728e-08,
+      "loss": 3.4325,
+      "step": 6871
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 3.854545454545455e-08,
+      "loss": 3.6025,
+      "step": 6872
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 3.7818181818181824e-08,
+      "loss": 3.1181,
+      "step": 6873
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 3.7090909090909093e-08,
+      "loss": 3.5567,
+      "step": 6874
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 3.636363636363637e-08,
+      "loss": 3.4422,
+      "step": 6875
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 3.563636363636364e-08,
+      "loss": 3.5501,
+      "step": 6876
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 3.490909090909091e-08,
+      "loss": 3.6811,
+      "step": 6877
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 3.4181818181818186e-08,
+      "loss": 3.3271,
+      "step": 6878
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 3.3454545454545456e-08,
+      "loss": 3.3017,
+      "step": 6879
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 3.272727272727273e-08,
+      "loss": 3.2683,
+      "step": 6880
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 3.2e-08,
+      "loss": 3.2626,
+      "step": 6881
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 3.127272727272727e-08,
+      "loss": 3.2507,
+      "step": 6882
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 3.054545454545455e-08,
+      "loss": 3.422,
+      "step": 6883
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 2.981818181818182e-08,
+      "loss": 3.5287,
+      "step": 6884
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 2.909090909090909e-08,
+      "loss": 3.207,
+      "step": 6885
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 2.8363636363636368e-08,
+      "loss": 3.3684,
+      "step": 6886
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 2.7636363636363637e-08,
+      "loss": 3.4454,
+      "step": 6887
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 2.6909090909090914e-08,
+      "loss": 3.0616,
+      "step": 6888
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 2.6181818181818184e-08,
+      "loss": 3.1544,
+      "step": 6889
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 2.5454545454545457e-08,
+      "loss": 3.3203,
+      "step": 6890
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 2.472727272727273e-08,
+      "loss": 3.3055,
+      "step": 6891
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 2.4e-08,
+      "loss": 3.0864,
+      "step": 6892
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 2.3272727272727276e-08,
+      "loss": 3.267,
+      "step": 6893
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 2.2545454545454546e-08,
+      "loss": 3.1752,
+      "step": 6894
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 2.181818181818182e-08,
+      "loss": 3.4119,
+      "step": 6895
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 2.1090909090909092e-08,
+      "loss": 3.4851,
+      "step": 6896
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 2.0363636363636366e-08,
+      "loss": 3.411,
+      "step": 6897
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.963636363636364e-08,
+      "loss": 3.3851,
+      "step": 6898
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.8909090909090912e-08,
+      "loss": 3.3375,
+      "step": 6899
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.8181818181818185e-08,
+      "loss": 3.4371,
+      "step": 6900
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.7454545454545455e-08,
+      "loss": 3.5182,
+      "step": 6901
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.6727272727272728e-08,
+      "loss": 3.3126,
+      "step": 6902
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.6e-08,
+      "loss": 3.4975,
+      "step": 6903
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.5272727272727274e-08,
+      "loss": 3.0901,
+      "step": 6904
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.4545454545454546e-08,
+      "loss": 3.5045,
+      "step": 6905
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.3818181818181819e-08,
+      "loss": 3.2495,
+      "step": 6906
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.3090909090909092e-08,
+      "loss": 3.2809,
+      "step": 6907
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2363636363636365e-08,
+      "loss": 3.3231,
+      "step": 6908
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.1636363636363638e-08,
+      "loss": 3.3292,
+      "step": 6909
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.090909090909091e-08,
+      "loss": 3.3082,
+      "step": 6910
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.0181818181818183e-08,
+      "loss": 3.554,
+      "step": 6911
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 9.454545454545456e-09,
+      "loss": 3.2617,
+      "step": 6912
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 8.727272727272727e-09,
+      "loss": 3.2052,
+      "step": 6913
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 8e-09,
+      "loss": 3.4303,
+      "step": 6914
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 7.272727272727273e-09,
+      "loss": 3.2194,
+      "step": 6915
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 6.545454545454546e-09,
+      "loss": 3.1008,
+      "step": 6916
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 5.818181818181819e-09,
+      "loss": 3.3002,
+      "step": 6917
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 5.090909090909091e-09,
+      "loss": 3.2115,
+      "step": 6918
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 4.363636363636364e-09,
+      "loss": 3.2151,
+      "step": 6919
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 3.6363636363636364e-09,
+      "loss": 3.4662,
+      "step": 6920
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 2.9090909090909095e-09,
+      "loss": 3.2976,
+      "step": 6921
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 2.181818181818182e-09,
+      "loss": 3.2364,
+      "step": 6922
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.4545454545454548e-09,
+      "loss": 3.4714,
+      "step": 6923
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 7.272727272727274e-10,
+      "loss": 3.4002,
+      "step": 6924
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 0.0,
+      "loss": 3.489,
+      "step": 6925
+    },
+    {
+      "epoch": 25.0,
+      "eval_accuracy": 0.1918918918918919,
+      "eval_loss": 3.3698325157165527,
+      "eval_runtime": 146.1445,
+      "eval_samples_per_second": 2.532,
+      "eval_steps_per_second": 0.636,
+      "step": 6925
+    }
+  ],
+  "max_steps": 6925,
+  "num_train_epochs": 25,
+  "total_flos": 7.519967946034781e+18,
+  "trial_name": null,
+  "trial_params": null
+}