diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17586 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998860918100011,
+  "eval_steps": 500,
+  "global_step": 2926,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2727272727272729e-07,
+      "loss": 2.0384,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5454545454545457e-07,
+      "loss": 1.9808,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.818181818181818e-07,
+      "loss": 2.0641,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 2.0398,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1363636363636364e-06,
+      "loss": 2.0087,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3636363636363636e-06,
+      "loss": 1.9463,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.590909090909091e-06,
+      "loss": 1.9657,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 1.9382,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0454545454545457e-06,
+      "loss": 1.9093,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2727272727272728e-06,
+      "loss": 1.8782,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.7172,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 1.765,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.954545454545455e-06,
+      "loss": 1.78,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.181818181818182e-06,
+      "loss": 1.5682,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.409090909090909e-06,
+      "loss": 1.5741,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 1.5239,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.863636363636364e-06,
+      "loss": 1.4729,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0909090909090915e-06,
+      "loss": 1.4706,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3181818181818185e-06,
+      "loss": 1.3792,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 1.4571,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.772727272727273e-06,
+      "loss": 1.3786,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.3865,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2272727272727274e-06,
+      "loss": 1.3841,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 1.357,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.681818181818183e-06,
+      "loss": 1.3138,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.90909090909091e-06,
+      "loss": 1.2225,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.136363636363637e-06,
+      "loss": 1.2303,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 1.2904,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.590909090909091e-06,
+      "loss": 1.234,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.818181818181818e-06,
+      "loss": 1.2952,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.045454545454546e-06,
+      "loss": 1.221,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.1881,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.1982,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.727272727272727e-06,
+      "loss": 1.2229,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.954545454545455e-06,
+      "loss": 1.1583,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 1.265,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.40909090909091e-06,
+      "loss": 1.2044,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.636363636363637e-06,
+      "loss": 1.0977,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.863636363636365e-06,
+      "loss": 1.1869,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 1.1431,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.318181818181819e-06,
+      "loss": 1.1416,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.545454545454547e-06,
+      "loss": 1.1601,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.772727272727273e-06,
+      "loss": 1.0484,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.1585,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0227272727272729e-05,
+      "loss": 1.1564,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0454545454545455e-05,
+      "loss": 1.0976,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0681818181818183e-05,
+      "loss": 1.1285,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 1.0729,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1136363636363637e-05,
+      "loss": 1.1063,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1363636363636366e-05,
+      "loss": 1.0705,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1590909090909093e-05,
+      "loss": 1.0517,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.181818181818182e-05,
+      "loss": 1.1073,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2045454545454547e-05,
+      "loss": 1.1203,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2272727272727274e-05,
+      "loss": 1.0799,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 1.1048,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 1.0251,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2954545454545455e-05,
+      "loss": 1.0732,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3181818181818183e-05,
+      "loss": 1.0894,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3409090909090909e-05,
+      "loss": 1.0005,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 1.0208,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3863636363636366e-05,
+      "loss": 1.0453,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4090909090909092e-05,
+      "loss": 0.9864,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.431818181818182e-05,
+      "loss": 0.9973,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 1.0441,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4772727272727274e-05,
+      "loss": 1.015,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.9982,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5227272727272728e-05,
+      "loss": 1.0866,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 1.0523,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5681818181818182e-05,
+      "loss": 1.0862,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.590909090909091e-05,
+      "loss": 0.9596,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6136363636363638e-05,
+      "loss": 1.0444,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 0.9777,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6590909090909094e-05,
+      "loss": 0.9614,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.681818181818182e-05,
+      "loss": 0.937,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7045454545454546e-05,
+      "loss": 0.987,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 1.022,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 1.0674,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.772727272727273e-05,
+      "loss": 1.0234,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7954545454545454e-05,
+      "loss": 0.9846,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 1.0427,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.840909090909091e-05,
+      "loss": 1.0164,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8636363636363638e-05,
+      "loss": 0.9688,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8863636363636366e-05,
+      "loss": 0.9969,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9090909090909094e-05,
+      "loss": 0.9967,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.931818181818182e-05,
+      "loss": 0.9329,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9545454545454546e-05,
+      "loss": 0.9836,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9772727272727274e-05,
+      "loss": 0.9954,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9314,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999993873041723e-05,
+      "loss": 0.9709,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997549217439e-05,
+      "loss": 1.0212,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999944857420527e-05,
+      "loss": 0.8797,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999901968817678e-05,
+      "loss": 0.9783,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999984682641839e-05,
+      "loss": 0.9863,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999779430290247e-05,
+      "loss": 1.014,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999699780515824e-05,
+      "loss": 1.0583,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999607877192728e-05,
+      "loss": 0.9298,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999503720433575e-05,
+      "loss": 0.9585,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999387310366e-05,
+      "loss": 0.989,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999258647132645e-05,
+      "loss": 0.9579,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999911773089118e-05,
+      "loss": 0.9727,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999896456181428e-05,
+      "loss": 0.9635,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999879914008964e-05,
+      "loss": 0.9625,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999862146591996e-05,
+      "loss": 0.9535,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998431539522965e-05,
+      "loss": 1.0251,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998229361131386e-05,
+      "loss": 1.0123,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998014930992976e-05,
+      "loss": 0.9188,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999778824937049e-05,
+      "loss": 0.9269,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997549316541708e-05,
+      "loss": 1.0006,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997298132799408e-05,
+      "loss": 0.9419,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997034698451396e-05,
+      "loss": 1.0012,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996759013820477e-05,
+      "loss": 0.9149,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996471079244477e-05,
+      "loss": 0.8806,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996170895076227e-05,
+      "loss": 0.8832,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999585846168357e-05,
+      "loss": 1.0051,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999553377944936e-05,
+      "loss": 1.0054,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995196848771456e-05,
+      "loss": 0.9595,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994847670062735e-05,
+      "loss": 0.9778,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994486243751076e-05,
+      "loss": 0.9917,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994112570279367e-05,
+      "loss": 0.954,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993726650105508e-05,
+      "loss": 1.0081,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993328483702393e-05,
+      "loss": 0.9705,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992918071557942e-05,
+      "loss": 0.944,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992495414175067e-05,
+      "loss": 0.9865,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999206051207169e-05,
+      "loss": 0.9704,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991613365780732e-05,
+      "loss": 0.9706,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991153975850125e-05,
+      "loss": 0.9797,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990682342842805e-05,
+      "loss": 1.0005,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99901984673367e-05,
+      "loss": 0.9085,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989702349924752e-05,
+      "loss": 1.0409,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99891939912149e-05,
+      "loss": 0.9481,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988673391830082e-05,
+      "loss": 0.9593,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998814055240823e-05,
+      "loss": 0.9672,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987595473602292e-05,
+      "loss": 0.9166,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998703815608019e-05,
+      "loss": 0.8925,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998646860052487e-05,
+      "loss": 0.9343,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985886807634246e-05,
+      "loss": 1.0221,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985292778121255e-05,
+      "loss": 0.9465,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998468651271381e-05,
+      "loss": 0.9414,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984068012154824e-05,
+      "loss": 0.934,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983437277202205e-05,
+      "loss": 0.9771,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982794308628843e-05,
+      "loss": 1.0008,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982139107222634e-05,
+      "loss": 0.972,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998147167378645e-05,
+      "loss": 0.89,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998079200913817e-05,
+      "loss": 0.9381,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980100114110637e-05,
+      "loss": 0.9865,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99793959895517e-05,
+      "loss": 0.963,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997867963632418e-05,
+      "loss": 1.0012,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99779510553059e-05,
+      "loss": 0.9355,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997721024738965e-05,
+      "loss": 0.9177,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976457213483213e-05,
+      "loss": 1.0243,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975691954509347e-05,
+      "loss": 0.9634,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99749144714058e-05,
+      "loss": 0.9342,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974124765125293e-05,
+      "loss": 0.9278,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973322836635517e-05,
+      "loss": 0.9841,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972508686919157e-05,
+      "loss": 0.9854,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971682316973862e-05,
+      "loss": 0.943,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997084372781226e-05,
+      "loss": 0.9283,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996999292046195e-05,
+      "loss": 0.9159,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969129895965503e-05,
+      "loss": 0.8565,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968254655380465e-05,
+      "loss": 0.9854,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967367199779345e-05,
+      "loss": 0.8912,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996646753024963e-05,
+      "loss": 0.9252,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996555564789376e-05,
+      "loss": 0.933,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964631553829152e-05,
+      "loss": 0.9564,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963695249188185e-05,
+      "loss": 0.9372,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996274673511819e-05,
+      "loss": 0.9521,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996178601278148e-05,
+      "loss": 0.9708,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996081308335531e-05,
+      "loss": 0.9385,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99598279480319e-05,
+      "loss": 0.8797,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995883060801843e-05,
+      "loss": 0.9236,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995782106453702e-05,
+      "loss": 0.9563,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956799318824776e-05,
+      "loss": 0.9383,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995576537213372e-05,
+      "loss": 0.9827,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954719225730847e-05,
+      "loss": 0.8861,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99536608808981e-05,
+      "loss": 0.9346,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995259033893236e-05,
+      "loss": 0.9159,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951507601145462e-05,
+      "loss": 0.9,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995041266886419e-05,
+      "loss": 0.884,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949305543430256e-05,
+      "loss": 0.9429,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948186226200323e-05,
+      "loss": 0.9195,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947054718545996e-05,
+      "loss": 0.8827,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945911021853818e-05,
+      "loss": 0.9584,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994475513752526e-05,
+      "loss": 0.8501,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994358706697674e-05,
+      "loss": 0.9669,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994240681163959e-05,
+      "loss": 0.9586,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99412143729601e-05,
+      "loss": 0.8879,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940009752399462e-05,
+      "loss": 0.9821,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993879295143381e-05,
+      "loss": 0.8782,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993756397155421e-05,
+      "loss": 0.9111,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936322814266634e-05,
+      "loss": 0.9129,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993506948109199e-05,
+      "loss": 0.9361,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933803973566104e-05,
+      "loss": 0.887,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932526293239713e-05,
+      "loss": 0.9274,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931236441678485e-05,
+      "loss": 0.9801,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992993442046298e-05,
+      "loss": 0.8702,
+      "step": 195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928620231188694e-05,
+      "loss": 0.9188,
+      "step": 196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992729387546602e-05,
+      "loss": 0.9215,
+      "step": 197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925955354920265e-05,
+      "loss": 0.93,
+      "step": 198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992460467119164e-05,
+      "loss": 0.9368,
+      "step": 199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923241825935262e-05,
+      "loss": 0.936,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992186682082115e-05,
+      "loss": 0.9961,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992047965753422e-05,
+      "loss": 0.9102,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99190803377743e-05,
+      "loss": 0.9504,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991766886325609e-05,
+      "loss": 0.9132,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991624523570922e-05,
+      "loss": 0.9175,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914809456878166e-05,
+      "loss": 0.9898,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991336152852234e-05,
+      "loss": 0.9545,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911901452416012e-05,
+      "loss": 0.9228,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910429230348348e-05,
+      "loss": 0.8327,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908944864123398e-05,
+      "loss": 0.9023,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907448355560094e-05,
+      "loss": 0.9625,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990593970649224e-05,
+      "loss": 0.8923,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990441891876852e-05,
+      "loss": 0.8692,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902885994252506e-05,
+      "loss": 0.8666,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990134093482262e-05,
+      "loss": 0.954,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899783742372173e-05,
+      "loss": 0.8169,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 0.8913,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989663296605713e-05,
+      "loss": 0.93,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895039386053465e-05,
+      "loss": 0.8953,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893433680751105e-05,
+      "loss": 0.9645,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989181585211766e-05,
+      "loss": 0.9325,
+      "step": 221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989018590213561e-05,
+      "loss": 0.846,
+      "step": 222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888543832802277e-05,
+      "loss": 0.9409,
+      "step": 223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886889646129842e-05,
+      "loss": 0.9461,
+      "step": 224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988522334414533e-05,
+      "loss": 0.9166,
+      "step": 225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883544928890612e-05,
+      "loss": 0.8939,
+      "step": 226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881854402422407e-05,
+      "loss": 0.9071,
+      "step": 227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880151766812275e-05,
+      "loss": 0.8732,
+      "step": 228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878437024146603e-05,
+      "loss": 0.9589,
+      "step": 229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987671017652663e-05,
+      "loss": 0.8847,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874971226068417e-05,
+      "loss": 0.9034,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873220174902857e-05,
+      "loss": 0.8783,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987145702517568e-05,
+      "loss": 0.8753,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869681779047432e-05,
+      "loss": 0.9102,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986789443869348e-05,
+      "loss": 0.9542,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866095006304025e-05,
+      "loss": 0.8722,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986428348408407e-05,
+      "loss": 0.7937,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862459874253438e-05,
+      "loss": 0.9194,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860624179046773e-05,
+      "loss": 0.8346,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858776400713515e-05,
+      "loss": 0.8719,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985691654151791e-05,
+      "loss": 0.8737,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985504460373903e-05,
+      "loss": 0.9275,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985316058967072e-05,
+      "loss": 0.8995,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851264501621635e-05,
+      "loss": 0.9074,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984935634191523e-05,
+      "loss": 0.9528,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847436112889747e-05,
+      "loss": 0.869,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984550381689822e-05,
+      "loss": 0.9168,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843559456308462e-05,
+      "loss": 0.9656,
+      "step": 248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984160303350308e-05,
+      "loss": 0.9102,
+      "step": 249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983963455087946e-05,
+      "loss": 0.8912,
+      "step": 250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837654010849766e-05,
+      "loss": 0.9645,
+      "step": 251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983566141584093e-05,
+      "loss": 0.9157,
+      "step": 252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983365676829466e-05,
+      "loss": 0.9452,
+      "step": 253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831640070667445e-05,
+      "loss": 0.8863,
+      "step": 254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829611325430517e-05,
+      "loss": 0.8654,
+      "step": 255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982757053506989e-05,
+      "loss": 0.9302,
+      "step": 256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825517702086333e-05,
+      "loss": 0.8368,
+      "step": 257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823452828995363e-05,
+      "loss": 0.8697,
+      "step": 258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821375918327268e-05,
+      "loss": 0.8916,
+      "step": 259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819286972627066e-05,
+      "loss": 0.9065,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817185994454545e-05,
+      "loss": 0.866,
+      "step": 261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981507298638422e-05,
+      "loss": 0.8785,
+      "step": 262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981294795100535e-05,
+      "loss": 0.871,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810810890921943e-05,
+      "loss": 0.8612,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808661808752735e-05,
+      "loss": 0.918,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806500707131183e-05,
+      "loss": 0.8145,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804327588705493e-05,
+      "loss": 0.892,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980214245613858e-05,
+      "loss": 0.9852,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799945312108093e-05,
+      "loss": 0.9329,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979773615930639e-05,
+      "loss": 0.8559,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979551500044055e-05,
+      "loss": 0.8422,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979328183823236e-05,
+      "loss": 0.8935,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979103667541832e-05,
+      "loss": 0.8872,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788779514749635e-05,
+      "loss": 0.9392,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786510358992213e-05,
+      "loss": 0.8755,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784229210926652e-05,
+      "loss": 0.9803,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978193607334826e-05,
+      "loss": 0.8424,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779630949067015e-05,
+      "loss": 0.868,
+      "step": 278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977731384090761e-05,
+      "loss": 0.8995,
+      "step": 279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977498475170941e-05,
+      "loss": 0.8865,
+      "step": 280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772643684326455e-05,
+      "loss": 0.8931,
+      "step": 281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977029064162747e-05,
+      "loss": 0.928,
+      "step": 282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767925626495857e-05,
+      "loss": 0.8655,
+      "step": 283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765548641829685e-05,
+      "loss": 0.8932,
+      "step": 284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763159690541688e-05,
+      "loss": 0.8581,
+      "step": 285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760758775559275e-05,
+      "loss": 0.9342,
+      "step": 286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758345899824496e-05,
+      "loss": 0.9181,
+      "step": 287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975592106629408e-05,
+      "loss": 0.9566,
+      "step": 288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975348427793939e-05,
+      "loss": 0.9,
+      "step": 289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975103553774645e-05,
+      "loss": 0.8554,
+      "step": 290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974857484871592e-05,
+      "loss": 0.8111,
+      "step": 291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746102213863113e-05,
+      "loss": 0.8801,
+      "step": 292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743617636217978e-05,
+      "loss": 0.8768,
+      "step": 293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974112111882509e-05,
+      "loss": 0.9366,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973861266474366e-05,
+      "loss": 0.8377,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736092277047537e-05,
+      "loss": 0.912,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733559958825167e-05,
+      "loss": 0.9013,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731015713179643e-05,
+      "loss": 0.8847,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972845954322866e-05,
+      "loss": 0.8733,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972589145210453e-05,
+      "loss": 0.8338,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723311442954163e-05,
+      "loss": 0.9337,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720719518939083e-05,
+      "loss": 0.8895,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718115683235418e-05,
+      "loss": 0.8905,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715499939033883e-05,
+      "loss": 0.9025,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712872289539784e-05,
+      "loss": 0.9314,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971023273797303e-05,
+      "loss": 0.8899,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707581287568094e-05,
+      "loss": 0.8369,
+      "step": 307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704917941574053e-05,
+      "loss": 0.9263,
+      "step": 308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702242703254547e-05,
+      "loss": 0.9185,
+      "step": 309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969955557588778e-05,
+      "loss": 0.921,
+      "step": 310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969685656276655e-05,
+      "loss": 0.8036,
+      "step": 311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694145667198194e-05,
+      "loss": 0.8663,
+      "step": 312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691422892504626e-05,
+      "loss": 0.9314,
+      "step": 313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968868824202231e-05,
+      "loss": 0.9025,
+      "step": 314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685941719102267e-05,
+      "loss": 0.8636,
+      "step": 315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968318332711006e-05,
+      "loss": 0.8998,
+      "step": 316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680413069425796e-05,
+      "loss": 0.8608,
+      "step": 317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677630949444135e-05,
+      "loss": 0.9541,
+      "step": 318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674836970574253e-05,
+      "loss": 0.9129,
+      "step": 319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672031136239878e-05,
+      "loss": 0.886,
+      "step": 320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669213449879253e-05,
+      "loss": 0.9631,
+      "step": 321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966638391494514e-05,
+      "loss": 0.81,
+      "step": 322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663542534904837e-05,
+      "loss": 0.9285,
+      "step": 323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660689313240146e-05,
+      "loss": 0.8813,
+      "step": 324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657824253447378e-05,
+      "loss": 0.8607,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654947359037353e-05,
+      "loss": 0.9258,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652058633535396e-05,
+      "loss": 0.9322,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649158080481327e-05,
+      "loss": 0.9068,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646245703429456e-05,
+      "loss": 0.9086,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643321505948588e-05,
+      "loss": 0.9614,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964038549162201e-05,
+      "loss": 0.9354,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637437664047492e-05,
+      "loss": 0.756,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634478026837273e-05,
+      "loss": 0.9134,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963150658361807e-05,
+      "loss": 0.9544,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628523338031066e-05,
+      "loss": 0.9213,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96255282937319e-05,
+      "loss": 0.8557,
+      "step": 336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962252145439068e-05,
+      "loss": 0.8752,
+      "step": 337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961950282369196e-05,
+      "loss": 0.8853,
+      "step": 338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616472405334746e-05,
+      "loss": 0.7768,
+      "step": 339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613430203032486e-05,
+      "loss": 0.8704,
+      "step": 340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610376220513067e-05,
+      "loss": 0.839,
+      "step": 341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960731046151882e-05,
+      "loss": 0.9249,
+      "step": 342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604232929806493e-05,
+      "loss": 0.8506,
+      "step": 343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601143629147273e-05,
+      "loss": 0.9052,
+      "step": 344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959804256332676e-05,
+      "loss": 0.8853,
+      "step": 345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.8859,
+      "step": 346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591805151416353e-05,
+      "loss": 0.8892,
+      "step": 347
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958866881296973e-05,
+      "loss": 0.9101,
+      "step": 348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585520724648354e-05,
+      "loss": 0.8213,
+      "step": 349
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958236089030986e-05,
+      "loss": 0.858,
+      "step": 350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957918931382629e-05,
+      "loss": 0.8411,
+      "step": 351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957600599908406e-05,
+      "loss": 0.9637,
+      "step": 352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572810949983982e-05,
+      "loss": 0.8584,
+      "step": 353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569604170441244e-05,
+      "loss": 0.9051,
+      "step": 354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95663856643854e-05,
+      "loss": 0.9219,
+      "step": 355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563155435760385e-05,
+      "loss": 0.9514,
+      "step": 356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559913488524495e-05,
+      "loss": 0.8558,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955665982665038e-05,
+      "loss": 0.8456,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553394454125058e-05,
+      "loss": 0.8237,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550117374949886e-05,
+      "loss": 0.8383,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546828593140565e-05,
+      "loss": 0.8701,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543528112727146e-05,
+      "loss": 0.9196,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954021593775401e-05,
+      "loss": 0.9355,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536892072279863e-05,
+      "loss": 0.7971,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953355652037775e-05,
+      "loss": 0.8576,
+      "step": 365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530209286135024e-05,
+      "loss": 0.8624,
+      "step": 366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526850373653356e-05,
+      "loss": 0.8603,
+      "step": 367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523479787048733e-05,
+      "loss": 0.89,
+      "step": 368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520097530451444e-05,
+      "loss": 0.8617,
+      "step": 369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516703608006074e-05,
+      "loss": 0.8538,
+      "step": 370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513298023871516e-05,
+      "loss": 0.8663,
+      "step": 371
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950988078222093e-05,
+      "loss": 0.9497,
+      "step": 372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506451887241787e-05,
+      "loss": 0.9645,
+      "step": 373
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503011343135828e-05,
+      "loss": 0.923,
+      "step": 374
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499559154119058e-05,
+      "loss": 0.8251,
+      "step": 375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949609532442176e-05,
+      "loss": 0.9155,
+      "step": 376
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492619858288493e-05,
+      "loss": 0.7902,
+      "step": 377
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489132759978052e-05,
+      "loss": 0.8078,
+      "step": 378
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485634033763507e-05,
+      "loss": 0.8653,
+      "step": 379
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948212368393217e-05,
+      "loss": 0.864,
+      "step": 380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947860171478558e-05,
+      "loss": 0.8249,
+      "step": 381
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475068130639543e-05,
+      "loss": 0.8722,
+      "step": 382
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947152293582408e-05,
+      "loss": 0.8593,
+      "step": 383
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467966134683437e-05,
+      "loss": 0.951,
+      "step": 384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464397731576093e-05,
+      "loss": 0.9033,
+      "step": 385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460817730874743e-05,
+      "loss": 0.9258,
+      "step": 386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457226136966286e-05,
+      "loss": 0.9457,
+      "step": 387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945362295425183e-05,
+      "loss": 0.8389,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450008187146685e-05,
+      "loss": 0.8491,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944638184008036e-05,
+      "loss": 0.81,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944274391749655e-05,
+      "loss": 0.9384,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943909442385313e-05,
+      "loss": 0.8354,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435433363622165e-05,
+      "loss": 0.894,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431760741289886e-05,
+      "loss": 0.8104,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428076561356692e-05,
+      "loss": 0.8558,
+      "step": 395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424380828337146e-05,
+      "loss": 0.9383,
+      "step": 396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942067354675997e-05,
+      "loss": 0.8935,
+      "step": 397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416954721168037e-05,
+      "loss": 0.8159,
+      "step": 398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941322435611836e-05,
+      "loss": 0.8986,
+      "step": 399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409482456182105e-05,
+      "loss": 0.8609,
+      "step": 400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405729025944557e-05,
+      "loss": 0.8879,
+      "step": 401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401964070005145e-05,
+      "loss": 0.8839,
+      "step": 402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939818759297741e-05,
+      "loss": 0.9235,
+      "step": 403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394399599489014e-05,
+      "loss": 0.9136,
+      "step": 404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939060009418174e-05,
+      "loss": 0.8831,
+      "step": 405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386789081711465e-05,
+      "loss": 0.8535,
+      "step": 406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938296656674817e-05,
+      "loss": 0.9121,
+      "step": 407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379132553975936e-05,
+      "loss": 0.8972,
+      "step": 408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375287048092927e-05,
+      "loss": 0.942,
+      "step": 409
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93714300538114e-05,
+      "loss": 0.8756,
+      "step": 410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936756157585768e-05,
+      "loss": 0.8991,
+      "step": 411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363681618972166e-05,
+      "loss": 0.9492,
+      "step": 412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359790187909326e-05,
+      "loss": 0.8695,
+      "step": 413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935588728743769e-05,
+      "loss": 0.8847,
+      "step": 414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351972922339835e-05,
+      "loss": 0.9012,
+      "step": 415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348047097412394e-05,
+      "loss": 0.8637,
+      "step": 416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934410981746604e-05,
+      "loss": 0.8909,
+      "step": 417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340161087325483e-05,
+      "loss": 0.9118,
+      "step": 418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336200911829463e-05,
+      "loss": 0.8449,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332229295830743e-05,
+      "loss": 0.8617,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328246244196117e-05,
+      "loss": 0.809,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324251761806376e-05,
+      "loss": 0.8874,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932024585355633e-05,
+      "loss": 0.8785,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931622852435478e-05,
+      "loss": 0.8292,
+      "step": 424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931219977912453e-05,
+      "loss": 0.9356,
+      "step": 425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930815962280237e-05,
+      "loss": 0.8249,
+      "step": 426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930410806033908e-05,
+      "loss": 0.8335,
+      "step": 427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300045096699404e-05,
+      "loss": 0.8365,
+      "step": 428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295970736862063e-05,
+      "loss": 0.8765,
+      "step": 429
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929188498581975e-05,
+      "loss": 0.8458,
+      "step": 430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287787848579104e-05,
+      "loss": 0.8185,
+      "step": 431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283679330160726e-05,
+      "loss": 0.8982,
+      "step": 432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279559435599164e-05,
+      "loss": 0.8464,
+      "step": 433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275428169942892e-05,
+      "loss": 0.9008,
+      "step": 434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271285538254336e-05,
+      "loss": 0.9377,
+      "step": 435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926713154560984e-05,
+      "loss": 0.8578,
+      "step": 436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262966197099676e-05,
+      "loss": 0.8027,
+      "step": 437
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258789497828022e-05,
+      "loss": 0.8875,
+      "step": 438
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254601452912972e-05,
+      "loss": 0.8961,
+      "step": 439
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250402067486523e-05,
+      "loss": 0.8733,
+      "step": 440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246191346694564e-05,
+      "loss": 0.9251,
+      "step": 441
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924196929569688e-05,
+      "loss": 0.8291,
+      "step": 442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237735919667133e-05,
+      "loss": 0.8623,
+      "step": 443
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233491223792874e-05,
+      "loss": 0.8931,
+      "step": 444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922923521327551e-05,
+      "loss": 0.8313,
+      "step": 445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224967893330324e-05,
+      "loss": 0.8715,
+      "step": 446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220689269186454e-05,
+      "loss": 0.8731,
+      "step": 447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216399346086893e-05,
+      "loss": 0.8048,
+      "step": 448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212098129288473e-05,
+      "loss": 0.9493,
+      "step": 449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207785624061873e-05,
+      "loss": 0.843,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203461835691596e-05,
+      "loss": 0.8473,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919912676947598e-05,
+      "loss": 0.8076,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194780430727177e-05,
+      "loss": 0.8931,
+      "step": 453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190422824771158e-05,
+      "loss": 0.8863,
+      "step": 454
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918605395694769e-05,
+      "loss": 0.8565,
+      "step": 455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181673832610354e-05,
+      "loss": 0.8385,
+      "step": 456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177282457126515e-05,
+      "loss": 0.8214,
+      "step": 457
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172879835877327e-05,
+      "loss": 0.8711,
+      "step": 458
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916846597425773e-05,
+      "loss": 0.8684,
+      "step": 459
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164040877676425e-05,
+      "loss": 0.8501,
+      "step": 460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159604551555895e-05,
+      "loss": 0.8553,
+      "step": 461
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155157001332374e-05,
+      "loss": 0.9008,
+      "step": 462
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150698232455853e-05,
+      "loss": 0.9085,
+      "step": 463
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146228250390074e-05,
+      "loss": 0.8872,
+      "step": 464
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914174706061251e-05,
+      "loss": 0.8615,
+      "step": 465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913725466861438e-05,
+      "loss": 0.8205,
+      "step": 466
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132751079900615e-05,
+      "loss": 0.8166,
+      "step": 467
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912823629998988e-05,
+      "loss": 0.9214,
+      "step": 468
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123710334414552e-05,
+      "loss": 0.9084,
+      "step": 469
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911917318872071e-05,
+      "loss": 0.8483,
+      "step": 470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911462486846813e-05,
+      "loss": 0.7946,
+      "step": 471
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911006537923029e-05,
+      "loss": 0.9031,
+      "step": 472
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105494726594344e-05,
+      "loss": 0.9079,
+      "step": 473
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910091291616114e-05,
+      "loss": 0.8558,
+      "step": 474
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 0.81,
+      "step": 475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909171584437466e-05,
+      "loss": 0.8742,
+      "step": 476
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087100594291396e-05,
+      "loss": 0.9393,
+      "step": 477
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908247420895089e-05,
+      "loss": 0.8485,
+      "step": 478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907783669402227e-05,
+      "loss": 0.8863,
+      "step": 479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073188055188313e-05,
+      "loss": 0.8156,
+      "step": 480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068528298145418e-05,
+      "loss": 0.8706,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063857428603616e-05,
+      "loss": 0.8619,
+      "step": 482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059175452286544e-05,
+      "loss": 0.8336,
+      "step": 483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905448237493147e-05,
+      "loss": 0.759,
+      "step": 484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049778202289237e-05,
+      "loss": 0.8742,
+      "step": 485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904506294012431e-05,
+      "loss": 0.8497,
+      "step": 486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040336594214727e-05,
+      "loss": 0.8756,
+      "step": 487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035599170352114e-05,
+      "loss": 0.8675,
+      "step": 488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030850674341667e-05,
+      "loss": 0.8256,
+      "step": 489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026091112002163e-05,
+      "loss": 0.8669,
+      "step": 490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021320489165916e-05,
+      "loss": 0.9242,
+      "step": 491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016538811678823e-05,
+      "loss": 0.8442,
+      "step": 492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90117460854003e-05,
+      "loss": 0.8496,
+      "step": 493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900694231620332e-05,
+      "loss": 0.8589,
+      "step": 494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002127509974376e-05,
+      "loss": 0.8253,
+      "step": 495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997301672613496e-05,
+      "loss": 0.8244,
+      "step": 496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899246481003422e-05,
+      "loss": 0.9097,
+      "step": 497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89876169281636e-05,
+      "loss": 0.8892,
+      "step": 498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982758032942184e-05,
+      "loss": 0.8638,
+      "step": 499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977888130324032e-05,
+      "loss": 0.8469,
+      "step": 500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897300722627667e-05,
+      "loss": 0.9107,
+      "step": 501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896811532678113e-05,
+      "loss": 0.8772,
+      "step": 502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963212437831894e-05,
+      "loss": 0.7867,
+      "step": 503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958298565436928e-05,
+      "loss": 0.8542,
+      "step": 504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953373715617646e-05,
+      "loss": 0.8441,
+      "step": 505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894843789440892e-05,
+      "loss": 0.8765,
+      "step": 506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943491107859066e-05,
+      "loss": 0.8828,
+      "step": 507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893853336202983e-05,
+      "loss": 0.869,
+      "step": 508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933564662996394e-05,
+      "loss": 0.8626,
+      "step": 509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892858501684736e-05,
+      "loss": 0.8708,
+      "step": 510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892359442968475e-05,
+      "loss": 0.859,
+      "step": 511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918592907623986e-05,
+      "loss": 0.8653,
+      "step": 512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913580456793884e-05,
+      "loss": 0.9201,
+      "step": 513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908557083336668e-05,
+      "loss": 0.8261,
+      "step": 514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903522793407935e-05,
+      "loss": 0.8521,
+      "step": 515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898477593176657e-05,
+      "loss": 0.8385,
+      "step": 516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889342148882519e-05,
+      "loss": 0.8087,
+      "step": 517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888354486549238e-05,
+      "loss": 0.843,
+      "step": 518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888327659255786e-05,
+      "loss": 0.8374,
+      "step": 519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878187813073465e-05,
+      "loss": 0.8606,
+      "step": 520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873088154331806e-05,
+      "loss": 0.7874,
+      "step": 521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886797762258196e-05,
+      "loss": 0.8164,
+      "step": 522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886285622408633e-05,
+      "loss": 0.8554,
+      "step": 523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885772396512063e-05,
+      "loss": 0.8801,
+      "step": 524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852580851973895e-05,
+      "loss": 0.9069,
+      "step": 525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847426890948447e-05,
+      "loss": 0.8037,
+      "step": 526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842262088359912e-05,
+      "loss": 0.8459,
+      "step": 527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837086450537195e-05,
+      "loss": 0.8246,
+      "step": 528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831899983822475e-05,
+      "loss": 0.8946,
+      "step": 529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826702694571207e-05,
+      "loss": 0.8973,
+      "step": 530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882149458915211e-05,
+      "loss": 0.8371,
+      "step": 531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816275673947148e-05,
+      "loss": 0.8452,
+      "step": 532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881104595535154e-05,
+      "loss": 0.8544,
+      "step": 533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805805439773732e-05,
+      "loss": 0.7706,
+      "step": 534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800554133635417e-05,
+      "loss": 0.8721,
+      "step": 535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87952920433715e-05,
+      "loss": 0.8497,
+      "step": 536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790019175430098e-05,
+      "loss": 0.8643,
+      "step": 537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784735536272543e-05,
+      "loss": 0.8762,
+      "step": 538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877944113237336e-05,
+      "loss": 0.9206,
+      "step": 539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774135970220272e-05,
+      "loss": 0.8372,
+      "step": 540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768820056314173e-05,
+      "loss": 0.8178,
+      "step": 541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876349339716915e-05,
+      "loss": 0.8208,
+      "step": 542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758155999312432e-05,
+      "loss": 0.8504,
+      "step": 543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875280786928444e-05,
+      "loss": 0.8893,
+      "step": 544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874744901363871e-05,
+      "loss": 0.8072,
+      "step": 545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874207943894195e-05,
+      "loss": 0.8517,
+      "step": 546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873669915177399e-05,
+      "loss": 0.8612,
+      "step": 547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731308158727784e-05,
+      "loss": 0.914,
+      "step": 548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725906466409422e-05,
+      "loss": 0.8518,
+      "step": 549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872049408143808e-05,
+      "loss": 0.9823,
+      "step": 550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871507101044605e-05,
+      "loss": 0.845,
+      "step": 551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709637260078728e-05,
+      "loss": 0.8652,
+      "step": 552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704192836994578e-05,
+      "loss": 0.8076,
+      "step": 553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698737747865154e-05,
+      "loss": 0.8796,
+      "step": 554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869327199937508e-05,
+      "loss": 0.9211,
+      "step": 555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687795598222024e-05,
+      "loss": 0.8345,
+      "step": 556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868230855111674e-05,
+      "loss": 0.9368,
+      "step": 557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676810864782998e-05,
+      "loss": 0.7902,
+      "step": 558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671302545957628e-05,
+      "loss": 0.8723,
+      "step": 559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866578360139047e-05,
+      "loss": 0.8118,
+      "step": 560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.9103,
+      "step": 561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654713862095272e-05,
+      "loss": 0.8369,
+      "step": 562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649163080932006e-05,
+      "loss": 0.8537,
+      "step": 563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643601701156467e-05,
+      "loss": 0.8382,
+      "step": 564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638029729583524e-05,
+      "loss": 0.889,
+      "step": 565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632447173041028e-05,
+      "loss": 0.8944,
+      "step": 566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626854038369793e-05,
+      "loss": 0.8435,
+      "step": 567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8621250332423603e-05,
+      "loss": 0.8756,
+      "step": 568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615636062069192e-05,
+      "loss": 0.8313,
+      "step": 569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861001123418624e-05,
+      "loss": 0.8585,
+      "step": 570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860437585566736e-05,
+      "loss": 0.8902,
+      "step": 571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598729933418102e-05,
+      "loss": 0.8557,
+      "step": 572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593073474356932e-05,
+      "loss": 0.7714,
+      "step": 573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587406485415226e-05,
+      "loss": 0.786,
+      "step": 574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858172897353727e-05,
+      "loss": 0.9173,
+      "step": 575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857604094568023e-05,
+      "loss": 0.7467,
+      "step": 576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570342408814173e-05,
+      "loss": 0.7477,
+      "step": 577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856463336992204e-05,
+      "loss": 0.8756,
+      "step": 578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855891383599964e-05,
+      "loss": 0.8458,
+      "step": 579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855318381405564e-05,
+      "loss": 0.8332,
+      "step": 580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547443311111565e-05,
+      "loss": 0.8625,
+      "step": 581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8541692334201773e-05,
+      "loss": 0.8111,
+      "step": 582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535930890373467e-05,
+      "loss": 0.8042,
+      "step": 583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853015898668667e-05,
+      "loss": 0.87,
+      "step": 584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524376630214227e-05,
+      "loss": 0.8756,
+      "step": 585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518583828041787e-05,
+      "loss": 0.9012,
+      "step": 586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512780587267806e-05,
+      "loss": 0.8454,
+      "step": 587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506966915003524e-05,
+      "loss": 0.8454,
+      "step": 588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501142818372964e-05,
+      "loss": 0.8532,
+      "step": 589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495308304512926e-05,
+      "loss": 0.8195,
+      "step": 590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848946338057298e-05,
+      "loss": 0.8762,
+      "step": 591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848360805371544e-05,
+      "loss": 0.8163,
+      "step": 592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847774233111538e-05,
+      "loss": 0.9047,
+      "step": 593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471866219960604e-05,
+      "loss": 0.8338,
+      "step": 594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465979727451653e-05,
+      "loss": 0.8574,
+      "step": 595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460082860801783e-05,
+      "loss": 0.8652,
+      "step": 596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454175627236966e-05,
+      "loss": 0.9215,
+      "step": 597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448258033995877e-05,
+      "loss": 0.7645,
+      "step": 598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8442330088329883e-05,
+      "loss": 0.8104,
+      "step": 599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436391797503045e-05,
+      "loss": 0.8642,
+      "step": 600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430443168792087e-05,
+      "loss": 0.8228,
+      "step": 601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842448420948642e-05,
+      "loss": 0.8585,
+      "step": 602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418514926888088e-05,
+      "loss": 0.7831,
+      "step": 603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.857,
+      "step": 604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840654542108494e-05,
+      "loss": 0.822,
+      "step": 605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400545212547452e-05,
+      "loss": 0.8406,
+      "step": 606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394534710051956e-05,
+      "loss": 0.8173,
+      "step": 607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388513920963668e-05,
+      "loss": 0.8893,
+      "step": 608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382482852660414e-05,
+      "loss": 0.8487,
+      "step": 609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376441512532617e-05,
+      "loss": 0.8478,
+      "step": 610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370389907983283e-05,
+      "loss": 0.8734,
+      "step": 611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364328046428e-05,
+      "loss": 0.8454,
+      "step": 612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835825593529492e-05,
+      "loss": 0.8027,
+      "step": 613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352173582024754e-05,
+      "loss": 0.8464,
+      "step": 614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834608099407077e-05,
+      "loss": 0.8026,
+      "step": 615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833997817889878e-05,
+      "loss": 0.8458,
+      "step": 616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833386514398712e-05,
+      "loss": 0.814,
+      "step": 617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327741896826646e-05,
+      "loss": 0.7735,
+      "step": 618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321608444920738e-05,
+      "loss": 0.8814,
+      "step": 619
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315464795785275e-05,
+      "loss": 0.7843,
+      "step": 620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309310956948637e-05,
+      "loss": 0.8217,
+      "step": 621
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830314693595169e-05,
+      "loss": 0.8849,
+      "step": 622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296972740347763e-05,
+      "loss": 0.7776,
+      "step": 623
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290788377702672e-05,
+      "loss": 0.7997,
+      "step": 624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828459385559468e-05,
+      "loss": 0.8593,
+      "step": 625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8278389181614504e-05,
+      "loss": 0.7846,
+      "step": 626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82721743633653e-05,
+      "loss": 0.8551,
+      "step": 627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265949408462657e-05,
+      "loss": 0.8966,
+      "step": 628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.825971432453457e-05,
+      "loss": 0.8931,
+      "step": 629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825346911922148e-05,
+      "loss": 0.8654,
+      "step": 630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247213800176192e-05,
+      "loss": 0.849,
+      "step": 631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240948375063927e-05,
+      "loss": 0.7933,
+      "step": 632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234672851562287e-05,
+      "loss": 0.8522,
+      "step": 633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228387237361245e-05,
+      "loss": 0.9224,
+      "step": 634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8222091540163137e-05,
+      "loss": 0.8714,
+      "step": 635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215785767682662e-05,
+      "loss": 0.8433,
+      "step": 636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209469927646863e-05,
+      "loss": 0.8454,
+      "step": 637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820314402779511e-05,
+      "loss": 0.9171,
+      "step": 638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196808075879112e-05,
+      "loss": 0.8514,
+      "step": 639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190462079662897e-05,
+      "loss": 0.7958,
+      "step": 640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184106046922787e-05,
+      "loss": 0.8364,
+      "step": 641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8177739985447415e-05,
+      "loss": 0.8304,
+      "step": 642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81713639030377e-05,
+      "loss": 0.8371,
+      "step": 643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816497780750684e-05,
+      "loss": 0.8084,
+      "step": 644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81585817066803e-05,
+      "loss": 0.8095,
+      "step": 645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152175608395814e-05,
+      "loss": 0.7951,
+      "step": 646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814575952050336e-05,
+      "loss": 0.7856,
+      "step": 647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139333450865156e-05,
+      "loss": 0.8894,
+      "step": 648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132897407355657e-05,
+      "loss": 0.7495,
+      "step": 649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8126451397861532e-05,
+      "loss": 0.84,
+      "step": 650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119995430281677e-05,
+      "loss": 0.8548,
+      "step": 651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811352951252717e-05,
+      "loss": 0.9157,
+      "step": 652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107053652521297e-05,
+      "loss": 0.9009,
+      "step": 653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100567858199525e-05,
+      "loss": 0.867,
+      "step": 654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809407213750949e-05,
+      "loss": 0.8672,
+      "step": 655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087566498410994e-05,
+      "loss": 0.8445,
+      "step": 656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8081050948875996e-05,
+      "loss": 0.9063,
+      "step": 657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807452549688859e-05,
+      "loss": 0.8129,
+      "step": 658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067990150445015e-05,
+      "loss": 0.8995,
+      "step": 659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806144491755363e-05,
+      "loss": 0.8375,
+      "step": 660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054889806234906e-05,
+      "loss": 0.8758,
+      "step": 661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804832482452142e-05,
+      "loss": 0.8236,
+      "step": 662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041749980457856e-05,
+      "loss": 0.884,
+      "step": 663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8035165282100963e-05,
+      "loss": 0.9068,
+      "step": 664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028570737519576e-05,
+      "loss": 0.8605,
+      "step": 665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80219663547946e-05,
+      "loss": 0.799,
+      "step": 666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015352142018984e-05,
+      "loss": 0.8143,
+      "step": 667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800872810729774e-05,
+      "loss": 0.7878,
+      "step": 668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800209425874789e-05,
+      "loss": 0.8811,
+      "step": 669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799545060449851e-05,
+      "loss": 0.7988,
+      "step": 670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798879715269067e-05,
+      "loss": 0.8077,
+      "step": 671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7982133911477462e-05,
+      "loss": 0.8602,
+      "step": 672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797546088902396e-05,
+      "loss": 0.8491,
+      "step": 673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796877809350723e-05,
+      "loss": 0.8884,
+      "step": 674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962085533116314e-05,
+      "loss": 0.8776,
+      "step": 675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955383216052224e-05,
+      "loss": 0.8325,
+      "step": 676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794867115052792e-05,
+      "loss": 0.7766,
+      "step": 677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941949344768308e-05,
+      "loss": 0.8774,
+      "step": 678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935217807010238e-05,
+      "loss": 0.7493,
+      "step": 679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792847654550248e-05,
+      "loss": 0.8448,
+      "step": 680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792172556850572e-05,
+      "loss": 0.8125,
+      "step": 681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914964884292543e-05,
+      "loss": 0.8669,
+      "step": 682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790819450114744e-05,
+      "loss": 0.7913,
+      "step": 683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901414427366783e-05,
+      "loss": 0.8357,
+      "step": 684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894624671258813e-05,
+      "loss": 0.8184,
+      "step": 685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7887825241143643e-05,
+      "loss": 0.912,
+      "step": 686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881016145353243e-05,
+      "loss": 0.8529,
+      "step": 687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874197392231414e-05,
+      "loss": 0.8427,
+      "step": 688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78673689901338e-05,
+      "loss": 0.8512,
+      "step": 689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860530947427878e-05,
+      "loss": 0.8048,
+      "step": 690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853683272492913e-05,
+      "loss": 0.8601,
+      "step": 691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784682597372e-05,
+      "loss": 0.7956,
+      "step": 692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839959059512016e-05,
+      "loss": 0.8326,
+      "step": 693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833082538283615e-05,
+      "loss": 0.7913,
+      "step": 694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7826196418461226e-05,
+      "loss": 0.7731,
+      "step": 695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781930070848305e-05,
+      "loss": 0.8677,
+      "step": 696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812395416799034e-05,
+      "loss": 0.7758,
+      "step": 697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805480551870855e-05,
+      "loss": 0.8716,
+      "step": 698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779855612217194e-05,
+      "loss": 0.8266,
+      "step": 699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7791622136187422e-05,
+      "loss": 0.8914,
+      "step": 700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778467860241415e-05,
+      "loss": 0.8121,
+      "step": 701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777725529360677e-05,
+      "loss": 0.8618,
+      "step": 702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770762925547235e-05,
+      "loss": 0.8581,
+      "step": 703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763790799505746e-05,
+      "loss": 0.8392,
+      "step": 704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775680915977979e-05,
+      "loss": 0.823,
+      "step": 705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7749818014924612e-05,
+      "loss": 0.7776,
+      "step": 706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77428173735071e-05,
+      "loss": 0.8328,
+      "step": 707
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735807244105788e-05,
+      "loss": 0.8544,
+      "step": 708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728787635310828e-05,
+      "loss": 0.8517,
+      "step": 709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7721758555723987e-05,
+      "loss": 0.7611,
+      "step": 710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771472001395864e-05,
+      "loss": 0.8365,
+      "step": 711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770767201863976e-05,
+      "loss": 0.8371,
+      "step": 712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77006145784039e-05,
+      "loss": 0.8845,
+      "step": 713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693547701899193e-05,
+      "loss": 0.8108,
+      "step": 714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7686471397785322e-05,
+      "loss": 0.7559,
+      "step": 715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679385674733538e-05,
+      "loss": 0.8845,
+      "step": 716
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7672290541426624e-05,
+      "loss": 0.928,
+      "step": 717
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76651860065589e-05,
+      "loss": 0.7775,
+      "step": 718
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76580720788362e-05,
+      "loss": 0.8024,
+      "step": 719
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650948766975874e-05,
+      "loss": 0.814,
+      "step": 720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764381607970677e-05,
+      "loss": 0.8028,
+      "step": 721
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636674025769215e-05,
+      "loss": 0.7106,
+      "step": 722
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629522613915035e-05,
+      "loss": 0.8084,
+      "step": 723
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622361852907506e-05,
+      "loss": 0.8056,
+      "step": 724
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761519175152136e-05,
+      "loss": 0.8444,
+      "step": 725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760801231854278e-05,
+      "loss": 0.8951,
+      "step": 726
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760082356276939e-05,
+      "loss": 0.7887,
+      "step": 727
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593625493010223e-05,
+      "loss": 0.818,
+      "step": 728
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586418118085745e-05,
+      "loss": 0.8076,
+      "step": 729
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75792014468278e-05,
+      "loss": 0.9197,
+      "step": 730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571975488079644e-05,
+      "loss": 0.808,
+      "step": 731
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564740250695905e-05,
+      "loss": 0.8532,
+      "step": 732
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 0.8727,
+      "step": 733
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550241975497036e-05,
+      "loss": 0.8636,
+      "step": 734
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542978955447975e-05,
+      "loss": 0.8082,
+      "step": 735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535706692295436e-05,
+      "loss": 0.8514,
+      "step": 736
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528425194950794e-05,
+      "loss": 0.8143,
+      "step": 737
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752113447233674e-05,
+      "loss": 0.8495,
+      "step": 738
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513834533387256e-05,
+      "loss": 0.8643,
+      "step": 739
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506525387047627e-05,
+      "loss": 0.7833,
+      "step": 740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499207042274428e-05,
+      "loss": 0.8294,
+      "step": 741
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491879508035488e-05,
+      "loss": 0.8071,
+      "step": 742
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484542793309916e-05,
+      "loss": 0.835,
+      "step": 743
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747719690708805e-05,
+      "loss": 0.8173,
+      "step": 744
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746984185837149e-05,
+      "loss": 0.8069,
+      "step": 745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7462477656173037e-05,
+      "loss": 0.7882,
+      "step": 746
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455104309516738e-05,
+      "loss": 0.8201,
+      "step": 747
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744772182743782e-05,
+      "loss": 0.9037,
+      "step": 748
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440330218982718e-05,
+      "loss": 0.8171,
+      "step": 749
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743292949320905e-05,
+      "loss": 0.8261,
+      "step": 750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7425519659185596e-05,
+      "loss": 0.8683,
+      "step": 751
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7418100725992317e-05,
+      "loss": 0.7829,
+      "step": 752
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410672702720298e-05,
+      "loss": 0.9031,
+      "step": 753
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740323559847179e-05,
+      "loss": 0.7632,
+      "step": 754
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739578942236015e-05,
+      "loss": 0.7753,
+      "step": 755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388334183509858e-05,
+      "loss": 0.7484,
+      "step": 756
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738086989105651e-05,
+      "loss": 0.8561,
+      "step": 757
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7373396554146784e-05,
+      "loss": 0.8062,
+      "step": 758
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736591418193844e-05,
+      "loss": 0.8221,
+      "step": 759
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735842278360032e-05,
+      "loss": 0.8469,
+      "step": 760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350922368312315e-05,
+      "loss": 0.7644,
+      "step": 761
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343412945265384e-05,
+      "loss": 0.7911,
+      "step": 762
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73358945236615e-05,
+      "loss": 0.8383,
+      "step": 763
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328367112713676e-05,
+      "loss": 0.9007,
+      "step": 764
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732083072164594e-05,
+      "loss": 0.8981,
+      "step": 765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313285359693322e-05,
+      "loss": 0.8534,
+      "step": 766
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305731036101846e-05,
+      "loss": 0.8271,
+      "step": 767
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298167760128515e-05,
+      "loss": 0.7397,
+      "step": 768
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7290595541041312e-05,
+      "loss": 0.8154,
+      "step": 769
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728301438811916e-05,
+      "loss": 0.7939,
+      "step": 770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727542431065195e-05,
+      "loss": 0.7673,
+      "step": 771
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267825317940494e-05,
+      "loss": 0.8567,
+      "step": 772
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260217419296536e-05,
+      "loss": 0.7524,
+      "step": 773
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725260062404273e-05,
+      "loss": 0.8726,
+      "step": 774
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724497494151264e-05,
+      "loss": 0.8705,
+      "step": 775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72373403810507e-05,
+      "loss": 0.8659,
+      "step": 776
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7229696952012254e-05,
+      "loss": 0.837,
+      "step": 777
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222044663763484e-05,
+      "loss": 0.8487,
+      "step": 778
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7214383525681443e-05,
+      "loss": 0.8661,
+      "step": 779
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206713547154022e-05,
+      "loss": 0.7689,
+      "step": 780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199034737579962e-05,
+      "loss": 0.8347,
+      "step": 781
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7191347106368796e-05,
+      "loss": 0.8214,
+      "step": 782
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183650662940893e-05,
+      "loss": 0.8108,
+      "step": 783
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175945416727405e-05,
+      "loss": 0.8394,
+      "step": 784
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168231377170284e-05,
+      "loss": 0.8503,
+      "step": 785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716050855372224e-05,
+      "loss": 0.8782,
+      "step": 786
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152776955846768e-05,
+      "loss": 0.8615,
+      "step": 787
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714503659301809e-05,
+      "loss": 0.8109,
+      "step": 788
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713728747472119e-05,
+      "loss": 0.8245,
+      "step": 789
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129529610451775e-05,
+      "loss": 0.8003,
+      "step": 790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121763009716262e-05,
+      "loss": 0.8535,
+      "step": 791
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711398768203178e-05,
+      "loss": 0.8667,
+      "step": 792
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106203636926154e-05,
+      "loss": 0.8748,
+      "step": 793
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709841088393788e-05,
+      "loss": 0.8413,
+      "step": 794
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090609432616142e-05,
+      "loss": 0.8021,
+      "step": 795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082799292520767e-05,
+      "loss": 0.863,
+      "step": 796
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707498047322224e-05,
+      "loss": 0.7679,
+      "step": 797
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7067152984301674e-05,
+      "loss": 0.8867,
+      "step": 798
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7059316835350806e-05,
+      "loss": 0.8306,
+      "step": 799
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051472035971993e-05,
+      "loss": 0.7416,
+      "step": 800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7043618595778187e-05,
+      "loss": 0.8578,
+      "step": 801
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7035756524392924e-05,
+      "loss": 0.8271,
+      "step": 802
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027885831450318e-05,
+      "loss": 0.7874,
+      "step": 803
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702000652659506e-05,
+      "loss": 0.9156,
+      "step": 804
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012118619482376e-05,
+      "loss": 0.8549,
+      "step": 805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004222119778044e-05,
+      "loss": 0.7897,
+      "step": 806
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699631703715837e-05,
+      "loss": 0.8054,
+      "step": 807
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988403381310177e-05,
+      "loss": 0.9034,
+      "step": 808
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6980481161930794e-05,
+      "loss": 0.7721,
+      "step": 809
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6972550388728035e-05,
+      "loss": 0.8287,
+      "step": 810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696461107142021e-05,
+      "loss": 0.8271,
+      "step": 811
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956663219736093e-05,
+      "loss": 0.8346,
+      "step": 812
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694870684341491e-05,
+      "loss": 0.846,
+      "step": 813
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940741952206342e-05,
+      "loss": 0.8489,
+      "step": 814
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69327685558705e-05,
+      "loss": 0.8416,
+      "step": 815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924786664177913e-05,
+      "loss": 0.795,
+      "step": 816
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691679628690953e-05,
+      "loss": 0.8044,
+      "step": 817
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690879743385669e-05,
+      "loss": 0.8419,
+      "step": 818
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900790114821122e-05,
+      "loss": 0.8193,
+      "step": 819
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892774339614927e-05,
+      "loss": 0.9069,
+      "step": 820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6884750118060568e-05,
+      "loss": 0.7914,
+      "step": 821
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6876717459990865e-05,
+      "loss": 0.8049,
+      "step": 822
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686867637524896e-05,
+      "loss": 0.7071,
+      "step": 823
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6860626873688343e-05,
+      "loss": 0.7802,
+      "step": 824
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6852568965172794e-05,
+      "loss": 0.7829,
+      "step": 825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844502659576414e-05,
+      "loss": 0.9022,
+      "step": 826
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836427966783586e-05,
+      "loss": 0.7505,
+      "step": 827
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828344896688973e-05,
+      "loss": 0.7764,
+      "step": 828
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820253459197493e-05,
+      "loss": 0.7875,
+      "step": 829
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812153664224338e-05,
+      "loss": 0.8166,
+      "step": 830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680404552169492e-05,
+      "loss": 0.8388,
+      "step": 831
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679592904154489e-05,
+      "loss": 0.8255,
+      "step": 832
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6787804233720124e-05,
+      "loss": 0.8836,
+      "step": 833
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6779671108176684e-05,
+      "loss": 0.8188,
+      "step": 834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677152967488084e-05,
+      "loss": 0.7749,
+      "step": 835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676337994380903e-05,
+      "loss": 0.8405,
+      "step": 836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755221924947872e-05,
+      "loss": 0.7975,
+      "step": 837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6747055628294134e-05,
+      "loss": 0.8322,
+      "step": 838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6738881063854725e-05,
+      "loss": 0.7207,
+      "step": 839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730698241646686e-05,
+      "loss": 0.7564,
+      "step": 840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722507171697184e-05,
+      "loss": 0.8344,
+      "step": 841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671430786404349e-05,
+      "loss": 0.8158,
+      "step": 842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670610032873296e-05,
+      "loss": 0.7697,
+      "step": 843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669788457582304e-05,
+      "loss": 0.7713,
+      "step": 844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6689660615381255e-05,
+      "loss": 0.8433,
+      "step": 845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6681428457485167e-05,
+      "loss": 0.8116,
+      "step": 846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6673188112222394e-05,
+      "loss": 0.7547,
+      "step": 847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6664939589690593e-05,
+      "loss": 0.8175,
+      "step": 848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665668289999743e-05,
+      "loss": 0.781,
+      "step": 849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6648418053260585e-05,
+      "loss": 0.8339,
+      "step": 850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640145059607728e-05,
+      "loss": 0.9208,
+      "step": 851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631863929176523e-05,
+      "loss": 0.7969,
+      "step": 852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623574672114596e-05,
+      "loss": 0.8362,
+      "step": 853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6615277298579534e-05,
+      "loss": 0.7613,
+      "step": 854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660697181873886e-05,
+      "loss": 0.8526,
+      "step": 855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6598658242770054e-05,
+      "loss": 0.8696,
+      "step": 856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65903365808605e-05,
+      "loss": 0.7745,
+      "step": 857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658200684320748e-05,
+      "loss": 0.8327,
+      "step": 858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573669040018202e-05,
+      "loss": 0.8988,
+      "step": 859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656532318150973e-05,
+      "loss": 0.8392,
+      "step": 860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6556969277909012e-05,
+      "loss": 0.8638,
+      "step": 861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.8425,
+      "step": 862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65402373763879e-05,
+      "loss": 0.7921,
+      "step": 863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653185939897064e-05,
+      "loss": 0.8209,
+      "step": 864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652347341746737e-05,
+      "loss": 0.8025,
+      "step": 865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651507944215421e-05,
+      "loss": 0.8219,
+      "step": 866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650667748331706e-05,
+      "loss": 0.7423,
+      "step": 867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498267551251618e-05,
+      "loss": 0.7737,
+      "step": 868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489849656263336e-05,
+      "loss": 0.8658,
+      "step": 869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481423808667438e-05,
+      "loss": 0.9068,
+      "step": 870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6472990018788884e-05,
+      "loss": 0.7723,
+      "step": 871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646454829696237e-05,
+      "loss": 0.908,
+      "step": 872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456098653532317e-05,
+      "loss": 0.8138,
+      "step": 873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644764109885284e-05,
+      "loss": 0.8086,
+      "step": 874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439175643287762e-05,
+      "loss": 0.8231,
+      "step": 875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643070229721058e-05,
+      "loss": 0.8066,
+      "step": 876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642222107100446e-05,
+      "loss": 0.8343,
+      "step": 877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6413731975062224e-05,
+      "loss": 0.7138,
+      "step": 878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6405235019786343e-05,
+      "loss": 0.8901,
+      "step": 879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396730215588913e-05,
+      "loss": 0.8172,
+      "step": 880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388217572891655e-05,
+      "loss": 0.8547,
+      "step": 881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637969710212588e-05,
+      "loss": 0.8592,
+      "step": 882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6371168813732514e-05,
+      "loss": 0.8519,
+      "step": 883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636263271816204e-05,
+      "loss": 0.8279,
+      "step": 884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6354088825874525e-05,
+      "loss": 0.8175,
+      "step": 885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345537147339578e-05,
+      "loss": 0.8977,
+      "step": 886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633697769303636e-05,
+      "loss": 0.9021,
+      "step": 887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328410473453554e-05,
+      "loss": 0.8514,
+      "step": 888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6319835499089358e-05,
+      "loss": 0.8666,
+      "step": 889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631125278045147e-05,
+      "loss": 0.8717,
+      "step": 890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630266232805709e-05,
+      "loss": 0.8873,
+      "step": 891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6294064152432878e-05,
+      "loss": 0.8577,
+      "step": 892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285458264114977e-05,
+      "loss": 0.8043,
+      "step": 893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276844673648955e-05,
+      "loss": 0.8502,
+      "step": 894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626822339158985e-05,
+      "loss": 0.8172,
+      "step": 895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259594428502106e-05,
+      "loss": 0.7852,
+      "step": 896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250957794959577e-05,
+      "loss": 0.7997,
+      "step": 897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6242313501545522e-05,
+      "loss": 0.7981,
+      "step": 898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623366155885259e-05,
+      "loss": 0.787,
+      "step": 899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225001977482795e-05,
+      "loss": 0.8165,
+      "step": 900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621633476804752e-05,
+      "loss": 0.7732,
+      "step": 901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6207659941167485e-05,
+      "loss": 0.8026,
+      "step": 902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619897750747276e-05,
+      "loss": 0.8451,
+      "step": 903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6190287477602716e-05,
+      "loss": 0.8712,
+      "step": 904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6181589862206053e-05,
+      "loss": 0.8714,
+      "step": 905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6172884671940753e-05,
+      "loss": 0.7595,
+      "step": 906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6164171917474078e-05,
+      "loss": 0.771,
+      "step": 907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615545160948257e-05,
+      "loss": 0.8757,
+      "step": 908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146723758652024e-05,
+      "loss": 0.8618,
+      "step": 909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137988375677466e-05,
+      "loss": 0.8238,
+      "step": 910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129245471263173e-05,
+      "loss": 0.8063,
+      "step": 911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120495056122623e-05,
+      "loss": 0.8424,
+      "step": 912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111737140978495e-05,
+      "loss": 0.8364,
+      "step": 913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6102971736562675e-05,
+      "loss": 0.8609,
+      "step": 914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6094198853616206e-05,
+      "loss": 0.8448,
+      "step": 915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6085418502889315e-05,
+      "loss": 0.7928,
+      "step": 916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607663069514137e-05,
+      "loss": 0.8102,
+      "step": 917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067835441140872e-05,
+      "loss": 0.8215,
+      "step": 918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059032751665454e-05,
+      "loss": 0.8307,
+      "step": 919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605022263750186e-05,
+      "loss": 0.8018,
+      "step": 920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6041405109445926e-05,
+      "loss": 0.8073,
+      "step": 921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6032580178302585e-05,
+      "loss": 0.8138,
+      "step": 922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6023747854885826e-05,
+      "loss": 0.9028,
+      "step": 923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6014908150018703e-05,
+      "loss": 0.8891,
+      "step": 924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600606107453333e-05,
+      "loss": 0.8358,
+      "step": 925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5997206639270825e-05,
+      "loss": 0.8722,
+      "step": 926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5988344855081347e-05,
+      "loss": 0.8408,
+      "step": 927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597947573282405e-05,
+      "loss": 0.8636,
+      "step": 928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597059928336708e-05,
+      "loss": 0.7778,
+      "step": 929
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5961715517587568e-05,
+      "loss": 0.7627,
+      "step": 930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5952824446371608e-05,
+      "loss": 0.8002,
+      "step": 931
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943926080614236e-05,
+      "loss": 0.8613,
+      "step": 932
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935020431219446e-05,
+      "loss": 0.8572,
+      "step": 933
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592610750910014e-05,
+      "loss": 0.8178,
+      "step": 934
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591718732517814e-05,
+      "loss": 0.7909,
+      "step": 935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590825989038416e-05,
+      "loss": 0.807,
+      "step": 936
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589932521565781e-05,
+      "loss": 0.77,
+      "step": 937
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589038331194757e-05,
+      "loss": 0.8519,
+      "step": 938
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881434190210765e-05,
+      "loss": 0.8114,
+      "step": 939
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587247786141358e-05,
+      "loss": 0.8458,
+      "step": 940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586351433653102e-05,
+      "loss": 0.8236,
+      "step": 941
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854543626546917e-05,
+      "loss": 0.8412,
+      "step": 942
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845565742453906e-05,
+      "loss": 0.7521,
+      "step": 943
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583658069525341e-05,
+      "loss": 0.7794,
+      "step": 944
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582758849595563e-05,
+      "loss": 0.8149,
+      "step": 945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581858915557953e-05,
+      "loss": 0.8104,
+      "step": 946
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5809582685152828e-05,
+      "loss": 0.8021,
+      "step": 947
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800569095711983e-05,
+      "loss": 0.8279,
+      "step": 948
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791548398302167e-05,
+      "loss": 0.7379,
+      "step": 949
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5782520603977265e-05,
+      "loss": 0.8008,
+      "step": 950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5773485723799868e-05,
+      "loss": 0.7551,
+      "step": 951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764443768841234e-05,
+      "loss": 0.8446,
+      "step": 952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575539475018131e-05,
+      "loss": 0.8412,
+      "step": 953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574633867890868e-05,
+      "loss": 0.7898,
+      "step": 954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737275566120577e-05,
+      "loss": 0.8144,
+      "step": 955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572820542292287e-05,
+      "loss": 0.8127,
+      "step": 956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571912826043003e-05,
+      "loss": 0.8002,
+      "step": 957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5710044089765144e-05,
+      "loss": 0.827,
+      "step": 958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5700952922059873e-05,
+      "loss": 0.748,
+      "step": 959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691854768454462e-05,
+      "loss": 0.8216,
+      "step": 960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682749640097708e-05,
+      "loss": 0.7876,
+      "step": 961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567363754814696e-05,
+      "loss": 0.8122,
+      "step": 962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5664518503768107e-05,
+      "loss": 0.7842,
+      "step": 963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565539251813554e-05,
+      "loss": 0.8507,
+      "step": 964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564625960243217e-05,
+      "loss": 0.8501,
+      "step": 965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637119767849392e-05,
+      "loss": 0.8449,
+      "step": 966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627973025587093e-05,
+      "loss": 0.8563,
+      "step": 967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5618819386853607e-05,
+      "loss": 0.8466,
+      "step": 968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5609658862865727e-05,
+      "loss": 0.8122,
+      "step": 969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560049146484868e-05,
+      "loss": 0.7614,
+      "step": 970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5591317204036123e-05,
+      "loss": 0.8012,
+      "step": 971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558213609167012e-05,
+      "loss": 0.8064,
+      "step": 972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5572948139001128e-05,
+      "loss": 0.7973,
+      "step": 973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5563753357287984e-05,
+      "loss": 0.8384,
+      "step": 974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5554551757797902e-05,
+      "loss": 0.7809,
+      "step": 975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545343351806443e-05,
+      "loss": 0.8521,
+      "step": 976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553612815059751e-05,
+      "loss": 0.7615,
+      "step": 977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5526906165463334e-05,
+      "loss": 0.8365,
+      "step": 978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551767740770446e-05,
+      "loss": 0.8691,
+      "step": 979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.550844188862973e-05,
+      "loss": 0.8336,
+      "step": 980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5499199619556272e-05,
+      "loss": 0.8197,
+      "step": 981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5489950611809484e-05,
+      "loss": 0.8121,
+      "step": 982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5480694876723027e-05,
+      "loss": 0.7496,
+      "step": 983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54714324256388e-05,
+      "loss": 0.7996,
+      "step": 984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5462163269906928e-05,
+      "loss": 0.8199,
+      "step": 985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545288742088576e-05,
+      "loss": 0.8592,
+      "step": 986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544360488994185e-05,
+      "loss": 0.7763,
+      "step": 987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5434315688449924e-05,
+      "loss": 0.8141,
+      "step": 988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425019827792903e-05,
+      "loss": 0.8339,
+      "step": 989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541571731936185e-05,
+      "loss": 0.7978,
+      "step": 990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5406408174555978e-05,
+      "loss": 0.787,
+      "step": 991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5397092404782643e-05,
+      "loss": 0.8761,
+      "step": 992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5387770021457306e-05,
+      "loss": 0.7862,
+      "step": 993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5378441036003543e-05,
+      "loss": 0.8253,
+      "step": 994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536910545985301e-05,
+      "loss": 0.8965,
+      "step": 995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5359763304445448e-05,
+      "loss": 0.8738,
+      "step": 996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535041458122865e-05,
+      "loss": 0.8601,
+      "step": 997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341059301658474e-05,
+      "loss": 0.8377,
+      "step": 998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331697477198792e-05,
+      "loss": 0.8086,
+      "step": 999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322329119321508e-05,
+      "loss": 0.8227,
+      "step": 1000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5312954239506536e-05,
+      "loss": 0.7645,
+      "step": 1001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303572849241762e-05,
+      "loss": 0.9442,
+      "step": 1002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529418496002308e-05,
+      "loss": 0.7583,
+      "step": 1003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284790583354317e-05,
+      "loss": 0.7795,
+      "step": 1004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527538973074727e-05,
+      "loss": 0.7305,
+      "step": 1005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5265982413721662e-05,
+      "loss": 0.789,
+      "step": 1006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525656864380515e-05,
+      "loss": 0.8154,
+      "step": 1007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5247148432533276e-05,
+      "loss": 0.892,
+      "step": 1008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237721791449497e-05,
+      "loss": 0.8235,
+      "step": 1009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5228288732105144e-05,
+      "loss": 0.8292,
+      "step": 1010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5218849266059396e-05,
+      "loss": 0.8582,
+      "step": 1011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209403404879305e-05,
+      "loss": 0.8147,
+      "step": 1012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5199951160139747e-05,
+      "loss": 0.8073,
+      "step": 1013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519049254342343e-05,
+      "loss": 0.8461,
+      "step": 1014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181027566320858e-05,
+      "loss": 0.8656,
+      "step": 1015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171556240430334e-05,
+      "loss": 0.7643,
+      "step": 1016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5162078577357947e-05,
+      "loss": 0.7767,
+      "step": 1017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152594588717544e-05,
+      "loss": 0.8699,
+      "step": 1018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143104286130723e-05,
+      "loss": 0.8026,
+      "step": 1019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133607681226825e-05,
+      "loss": 0.8686,
+      "step": 1020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124104785642909e-05,
+      "loss": 0.8137,
+      "step": 1021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114595611023743e-05,
+      "loss": 0.762,
+      "step": 1022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5105080169021792e-05,
+      "loss": 0.8076,
+      "step": 1023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095558471297196e-05,
+      "loss": 0.7659,
+      "step": 1024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508603052951777e-05,
+      "loss": 0.7494,
+      "step": 1025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076496355358971e-05,
+      "loss": 0.7526,
+      "step": 1026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066955960503893e-05,
+      "loss": 0.746,
+      "step": 1027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505740935664326e-05,
+      "loss": 0.8645,
+      "step": 1028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047856555475404e-05,
+      "loss": 0.7938,
+      "step": 1029
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038297568706244e-05,
+      "loss": 0.8308,
+      "step": 1030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028732408049277e-05,
+      "loss": 0.7988,
+      "step": 1031
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019161085225582e-05,
+      "loss": 0.8658,
+      "step": 1032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009583611963772e-05,
+      "loss": 0.7559,
+      "step": 1033
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.8277,
+      "step": 1034
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4990410261077952e-05,
+      "loss": 0.8305,
+      "step": 1035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980814406948806e-05,
+      "loss": 0.8338,
+      "step": 1036
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497121244937125e-05,
+      "loss": 0.7716,
+      "step": 1037
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961604400111437e-05,
+      "loss": 0.8001,
+      "step": 1038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951990270942991e-05,
+      "loss": 0.7909,
+      "step": 1039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942370073646986e-05,
+      "loss": 0.8413,
+      "step": 1040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4932743820011932e-05,
+      "loss": 0.7543,
+      "step": 1041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492311152183376e-05,
+      "loss": 0.8665,
+      "step": 1042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913473190915805e-05,
+      "loss": 0.788,
+      "step": 1043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49038288390688e-05,
+      "loss": 0.8714,
+      "step": 1044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4894178478110856e-05,
+      "loss": 0.8645,
+      "step": 1045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884522119867442e-05,
+      "loss": 0.8122,
+      "step": 1046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874859776171375e-05,
+      "loss": 0.8309,
+      "step": 1047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4865191458862816e-05,
+      "loss": 0.7836,
+      "step": 1048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4855517179789238e-05,
+      "loss": 0.8411,
+      "step": 1049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845836950805423e-05,
+      "loss": 0.7953,
+      "step": 1050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836150783773442e-05,
+      "loss": 0.8218,
+      "step": 1051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826458690562641e-05,
+      "loss": 0.7665,
+      "step": 1052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816760683049636e-05,
+      "loss": 0.7527,
+      "step": 1053
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807056773118276e-05,
+      "loss": 0.8256,
+      "step": 1054
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479734697265966e-05,
+      "loss": 0.8472,
+      "step": 1055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787631293572094e-05,
+      "loss": 0.8419,
+      "step": 1056
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777909747761085e-05,
+      "loss": 0.8087,
+      "step": 1057
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476818234713934e-05,
+      "loss": 0.7814,
+      "step": 1058
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475844910362673e-05,
+      "loss": 0.8288,
+      "step": 1059
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4748710029150296e-05,
+      "loss": 0.8365,
+      "step": 1060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738965135644214e-05,
+      "loss": 0.8232,
+      "step": 1061
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729214435049794e-05,
+      "loss": 0.7887,
+      "step": 1062
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719457939315468e-05,
+      "loss": 0.8274,
+      "step": 1063
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709695660396763e-05,
+      "loss": 0.7851,
+      "step": 1064
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4699927610256293e-05,
+      "loss": 0.7875,
+      "step": 1065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690153800863743e-05,
+      "loss": 0.82,
+      "step": 1066
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680374244195861e-05,
+      "loss": 0.7758,
+      "step": 1067
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4670588952236435e-05,
+      "loss": 0.8666,
+      "step": 1068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660797936976278e-05,
+      "loss": 0.811,
+      "step": 1069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651001210413216e-05,
+      "loss": 0.7913,
+      "step": 1070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4641198784552081e-05,
+      "loss": 0.8746,
+      "step": 1071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4631390671404682e-05,
+      "loss": 0.7486,
+      "step": 1072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4621576882989794e-05,
+      "loss": 0.7982,
+      "step": 1073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461175743133316e-05,
+      "loss": 0.8319,
+      "step": 1074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460193232846745e-05,
+      "loss": 0.8054,
+      "step": 1075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4592101586432261e-05,
+      "loss": 0.7607,
+      "step": 1076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4582265217274105e-05,
+      "loss": 0.8091,
+      "step": 1077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572423233046386e-05,
+      "loss": 0.7825,
+      "step": 1078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562575645809391e-05,
+      "loss": 0.8596,
+      "step": 1079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552722467630269e-05,
+      "loss": 0.8636,
+      "step": 1080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542863710583022e-05,
+      "loss": 0.7888,
+      "step": 1081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453299938674849e-05,
+      "loss": 0.7535,
+      "step": 1082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523129508214331e-05,
+      "loss": 0.8421,
+      "step": 1083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513254087075015e-05,
+      "loss": 0.8409,
+      "step": 1084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503373135431798e-05,
+      "loss": 0.7843,
+      "step": 1085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493486665392718e-05,
+      "loss": 0.7341,
+      "step": 1086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4483594689072571e-05,
+      "loss": 0.7953,
+      "step": 1087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4473697218592905e-05,
+      "loss": 0.8027,
+      "step": 1088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4463794266081994e-05,
+      "loss": 0.7408,
+      "step": 1089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4453885843674837e-05,
+      "loss": 0.8141,
+      "step": 1090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4443971963513128e-05,
+      "loss": 0.8408,
+      "step": 1091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434052637745256e-05,
+      "loss": 0.8315,
+      "step": 1092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424127878526278e-05,
+      "loss": 0.8097,
+      "step": 1093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414197698017915e-05,
+      "loss": 0.8437,
+      "step": 1094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404262108388523e-05,
+      "loss": 0.8483,
+      "step": 1095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4394321121813093e-05,
+      "loss": 0.7497,
+      "step": 1096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4384374750473227e-05,
+      "loss": 0.8063,
+      "step": 1097
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4374423006557123e-05,
+      "loss": 0.8612,
+      "step": 1098
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436446590225957e-05,
+      "loss": 0.8192,
+      "step": 1099
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354503449781914e-05,
+      "loss": 0.8198,
+      "step": 1100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344535661332067e-05,
+      "loss": 0.7851,
+      "step": 1101
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433456254912447e-05,
+      "loss": 0.7835,
+      "step": 1102
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324584125380092e-05,
+      "loss": 0.8633,
+      "step": 1103
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314600402326411e-05,
+      "loss": 0.8643,
+      "step": 1104
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304611392197399e-05,
+      "loss": 0.7335,
+      "step": 1105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294617107233504e-05,
+      "loss": 0.85,
+      "step": 1106
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428461755968164e-05,
+      "loss": 0.8364,
+      "step": 1107
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427461276179517e-05,
+      "loss": 0.8661,
+      "step": 1108
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4264602725833889e-05,
+      "loss": 0.8611,
+      "step": 1109
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254587464064009e-05,
+      "loss": 0.8052,
+      "step": 1110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244566988758152e-05,
+      "loss": 0.8095,
+      "step": 1111
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4234541312195322e-05,
+      "loss": 0.8474,
+      "step": 1112
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224510446660903e-05,
+      "loss": 0.7118,
+      "step": 1113
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214474404446633e-05,
+      "loss": 0.7667,
+      "step": 1114
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420443319785059e-05,
+      "loss": 0.7464,
+      "step": 1115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4194386839177192e-05,
+      "loss": 0.774,
+      "step": 1116
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184335340737158e-05,
+      "loss": 0.8035,
+      "step": 1117
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4174278714847512e-05,
+      "loss": 0.7868,
+      "step": 1118
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4164216973831555e-05,
+      "loss": 0.7537,
+      "step": 1119
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.7095,
+      "step": 1120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414407819574527e-05,
+      "loss": 0.8053,
+      "step": 1121
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4134001183352833e-05,
+      "loss": 0.8375,
+      "step": 1122
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123919105189836e-05,
+      "loss": 0.793,
+      "step": 1123
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411383197361078e-05,
+      "loss": 0.8454,
+      "step": 1124
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103739800976347e-05,
+      "loss": 0.783,
+      "step": 1125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4093642599653406e-05,
+      "loss": 0.8708,
+      "step": 1126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083540382014982e-05,
+      "loss": 0.8074,
+      "step": 1127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073433160440244e-05,
+      "loss": 0.937,
+      "step": 1128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40633209473145e-05,
+      "loss": 0.8172,
+      "step": 1129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4053203755029172e-05,
+      "loss": 0.8155,
+      "step": 1130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404308159598178e-05,
+      "loss": 0.8049,
+      "step": 1131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4032954482575938e-05,
+      "loss": 0.8119,
+      "step": 1132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022822427221325e-05,
+      "loss": 0.7724,
+      "step": 1133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4012685442333673e-05,
+      "loss": 0.8114,
+      "step": 1134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4002543540334766e-05,
+      "loss": 0.7805,
+      "step": 1135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992396733652397e-05,
+      "loss": 0.8658,
+      "step": 1136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982245034720388e-05,
+      "loss": 0.7968,
+      "step": 1137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3972088455978537e-05,
+      "loss": 0.8089,
+      "step": 1138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961927009872636e-05,
+      "loss": 0.7592,
+      "step": 1139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951760708854436e-05,
+      "loss": 0.7401,
+      "step": 1140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3941589565381635e-05,
+      "loss": 0.768,
+      "step": 1141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931413591917871e-05,
+      "loss": 0.8006,
+      "step": 1142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3921232800932694e-05,
+      "loss": 0.8352,
+      "step": 1143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391104720490156e-05,
+      "loss": 0.7917,
+      "step": 1144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3900856816305818e-05,
+      "loss": 0.7601,
+      "step": 1145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890661647632681e-05,
+      "loss": 0.7688,
+      "step": 1146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880461711375224e-05,
+      "loss": 0.7535,
+      "step": 1147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870257020032365e-05,
+      "loss": 0.7912,
+      "step": 1148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860047586108847e-05,
+      "loss": 0.829,
+      "step": 1149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849833422115221e-05,
+      "loss": 0.8052,
+      "step": 1150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3839614540567847e-05,
+      "loss": 0.7961,
+      "step": 1151
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3829390953988852e-05,
+      "loss": 0.8534,
+      "step": 1152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3819162674906134e-05,
+      "loss": 0.8027,
+      "step": 1153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380892971585334e-05,
+      "loss": 0.7971,
+      "step": 1154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3798692089369855e-05,
+      "loss": 0.8124,
+      "step": 1155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378844980800078e-05,
+      "loss": 0.8352,
+      "step": 1156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778202884296924e-05,
+      "loss": 0.7518,
+      "step": 1157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376795133081477e-05,
+      "loss": 0.7878,
+      "step": 1158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3757695160116502e-05,
+      "loss": 0.8329,
+      "step": 1159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747434384769937e-05,
+      "loss": 0.8047,
+      "step": 1160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3737169017348544e-05,
+      "loss": 0.7932,
+      "step": 1161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726899070431423e-05,
+      "loss": 0.8232,
+      "step": 1162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3716624556603275e-05,
+      "loss": 0.8506,
+      "step": 1163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3706345488454411e-05,
+      "loss": 0.8072,
+      "step": 1164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696061878580707e-05,
+      "loss": 0.7852,
+      "step": 1165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368577373958362e-05,
+      "loss": 0.857,
+      "step": 1166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675481084070144e-05,
+      "loss": 0.7945,
+      "step": 1167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665183924652817e-05,
+      "loss": 0.739,
+      "step": 1168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3654882273949687e-05,
+      "loss": 0.8133,
+      "step": 1169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364457614458432e-05,
+      "loss": 0.7989,
+      "step": 1170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634265549185755e-05,
+      "loss": 0.7838,
+      "step": 1171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3623950500388507e-05,
+      "loss": 0.7603,
+      "step": 1172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3613631010832555e-05,
+      "loss": 0.8304,
+      "step": 1173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603307093163319e-05,
+      "loss": 0.8183,
+      "step": 1174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3592978760031632e-05,
+      "loss": 0.7963,
+      "step": 1175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3582646024093756e-05,
+      "loss": 0.7733,
+      "step": 1176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572308898011328e-05,
+      "loss": 0.7915,
+      "step": 1177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356196739445139e-05,
+      "loss": 0.7953,
+      "step": 1178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551621526086327e-05,
+      "loss": 0.8269,
+      "step": 1179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541271305593878e-05,
+      "loss": 0.7532,
+      "step": 1180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530916745657122e-05,
+      "loss": 0.7939,
+      "step": 1181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520557858964448e-05,
+      "loss": 0.7691,
+      "step": 1182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3510194658209547e-05,
+      "loss": 0.7832,
+      "step": 1183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499827156091402e-05,
+      "loss": 0.7384,
+      "step": 1184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3489455365314262e-05,
+      "loss": 0.8326,
+      "step": 1185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3479079298587634e-05,
+      "loss": 0.8309,
+      "step": 1186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3468698968626263e-05,
+      "loss": 0.8593,
+      "step": 1187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458314388150115e-05,
+      "loss": 0.7938,
+      "step": 1188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3447925569884374e-05,
+      "loss": 0.8306,
+      "step": 1189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437532526559411e-05,
+      "loss": 0.7702,
+      "step": 1190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427135270910772e-05,
+      "loss": 0.7421,
+      "step": 1191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416733815679166e-05,
+      "loss": 0.8341,
+      "step": 1192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406328173610452e-05,
+      "loss": 0.7967,
+      "step": 1193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339591835745562e-05,
+      "loss": 0.8083,
+      "step": 1194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3385504379970764e-05,
+      "loss": 0.819,
+      "step": 1195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337508625391709e-05,
+      "loss": 0.714,
+      "step": 1196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3364663992060883e-05,
+      "loss": 0.8545,
+      "step": 1197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3354237607173494e-05,
+      "loss": 0.7919,
+      "step": 1198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343807112031329e-05,
+      "loss": 0.761,
+      "step": 1199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3333372519415828e-05,
+      "loss": 0.8247,
+      "step": 1200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3322933842113457e-05,
+      "loss": 0.7891,
+      "step": 1201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3312491092915683e-05,
+      "loss": 0.8217,
+      "step": 1202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3302044284618962e-05,
+      "loss": 0.7823,
+      "step": 1203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291593430024727e-05,
+      "loss": 0.8437,
+      "step": 1204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281138541939369e-05,
+      "loss": 0.8266,
+      "step": 1205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270679633174219e-05,
+      "loss": 0.8256,
+      "step": 1206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3260216716545534e-05,
+      "loss": 0.8393,
+      "step": 1207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324974980487449e-05,
+      "loss": 0.768,
+      "step": 1208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239278910987152e-05,
+      "loss": 0.7927,
+      "step": 1209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228804047714462e-05,
+      "loss": 0.8151,
+      "step": 1210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218325227892234e-05,
+      "loss": 0.7511,
+      "step": 1211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207842464361126e-05,
+      "loss": 0.7619,
+      "step": 1212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319735576996663e-05,
+      "loss": 0.7297,
+      "step": 1213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3186865157559052e-05,
+      "loss": 0.7675,
+      "step": 1214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3176370639993498e-05,
+      "loss": 0.8183,
+      "step": 1215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165872230129869e-05,
+      "loss": 0.8351,
+      "step": 1216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155369940832823e-05,
+      "loss": 0.8159,
+      "step": 1217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144863784971779e-05,
+      "loss": 0.7928,
+      "step": 1218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3134353775420895e-05,
+      "loss": 0.7969,
+      "step": 1219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123839925059045e-05,
+      "loss": 0.787,
+      "step": 1220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113322246769817e-05,
+      "loss": 0.8462,
+      "step": 1221
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102800753441488e-05,
+      "loss": 0.7878,
+      "step": 1222
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3092275457967003e-05,
+      "loss": 0.7856,
+      "step": 1223
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3081746373243975e-05,
+      "loss": 0.7541,
+      "step": 1224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071213512174655e-05,
+      "loss": 0.7826,
+      "step": 1225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3060676887665922e-05,
+      "loss": 0.7503,
+      "step": 1226
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305013651262927e-05,
+      "loss": 0.7732,
+      "step": 1227
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3039592399980785e-05,
+      "loss": 0.7196,
+      "step": 1228
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3029044562641136e-05,
+      "loss": 0.8065,
+      "step": 1229
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3018493013535551e-05,
+      "loss": 0.7451,
+      "step": 1230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3007937765593818e-05,
+      "loss": 0.7993,
+      "step": 1231
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2997378831750242e-05,
+      "loss": 0.7974,
+      "step": 1232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2986816224943661e-05,
+      "loss": 0.7224,
+      "step": 1233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2976249958117395e-05,
+      "loss": 0.7822,
+      "step": 1234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2965680044219264e-05,
+      "loss": 0.7923,
+      "step": 1235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2955106496201555e-05,
+      "loss": 0.8891,
+      "step": 1236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944529327021002e-05,
+      "loss": 0.7732,
+      "step": 1237
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933948549638778e-05,
+      "loss": 0.8138,
+      "step": 1238
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923364177020483e-05,
+      "loss": 0.8514,
+      "step": 1239
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291277622213612e-05,
+      "loss": 0.7822,
+      "step": 1240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2902184697960078e-05,
+      "loss": 0.7795,
+      "step": 1241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2891589617471123e-05,
+      "loss": 0.762,
+      "step": 1242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2880990993652379e-05,
+      "loss": 0.7798,
+      "step": 1243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870388839491308e-05,
+      "loss": 0.8623,
+      "step": 1244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859783167979705e-05,
+      "loss": 0.8243,
+      "step": 1245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849173992113669e-05,
+      "loss": 0.7387,
+      "step": 1246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2838561324893594e-05,
+      "loss": 0.8003,
+      "step": 1247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827945179324162e-05,
+      "loss": 0.784,
+      "step": 1248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817325568414299e-05,
+      "loss": 0.8053,
+      "step": 1249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2806702505177193e-05,
+      "loss": 0.7963,
+      "step": 1250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796076002630256e-05,
+      "loss": 0.8729,
+      "step": 1251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785446073795118e-05,
+      "loss": 0.7412,
+      "step": 1252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774812731697602e-05,
+      "loss": 0.7563,
+      "step": 1253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2764175989367717e-05,
+      "loss": 0.796,
+      "step": 1254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2753535859839638e-05,
+      "loss": 0.7924,
+      "step": 1255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2742892356151692e-05,
+      "loss": 0.7679,
+      "step": 1256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273224549134634e-05,
+      "loss": 0.7996,
+      "step": 1257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272159527847016e-05,
+      "loss": 0.7986,
+      "step": 1258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2710941730573836e-05,
+      "loss": 0.8219,
+      "step": 1259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2700284860712135e-05,
+      "loss": 0.7923,
+      "step": 1260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689624681943897e-05,
+      "loss": 0.8021,
+      "step": 1261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2678961207332014e-05,
+      "loss": 0.772,
+      "step": 1262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2668294449943423e-05,
+      "loss": 0.7792,
+      "step": 1263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657624422849077e-05,
+      "loss": 0.7937,
+      "step": 1264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2646951139123935e-05,
+      "loss": 0.8057,
+      "step": 1265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2636274611846957e-05,
+      "loss": 0.847,
+      "step": 1266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2625594854101066e-05,
+      "loss": 0.8286,
+      "step": 1267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614911878973145e-05,
+      "loss": 0.768,
+      "step": 1268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2604225699554032e-05,
+      "loss": 0.7879,
+      "step": 1269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2593536328938471e-05,
+      "loss": 0.8072,
+      "step": 1270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2582843780225138e-05,
+      "loss": 0.7758,
+      "step": 1271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2572148066516585e-05,
+      "loss": 0.8292,
+      "step": 1272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561449200919253e-05,
+      "loss": 0.8062,
+      "step": 1273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550747196543445e-05,
+      "loss": 0.8943,
+      "step": 1274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2540042066503305e-05,
+      "loss": 0.7292,
+      "step": 1275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529333823916807e-05,
+      "loss": 0.7768,
+      "step": 1276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518622481905749e-05,
+      "loss": 0.7437,
+      "step": 1277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507908053595718e-05,
+      "loss": 0.7892,
+      "step": 1278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497190552116082e-05,
+      "loss": 0.8191,
+      "step": 1279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486469990599983e-05,
+      "loss": 0.729,
+      "step": 1280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2475746382184302e-05,
+      "loss": 0.7887,
+      "step": 1281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465019740009662e-05,
+      "loss": 0.7672,
+      "step": 1282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24542900772204e-05,
+      "loss": 0.7703,
+      "step": 1283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443557406964557e-05,
+      "loss": 0.7822,
+      "step": 1284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432821742393854e-05,
+      "loss": 0.8294,
+      "step": 1285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242208309666369e-05,
+      "loss": 0.7386,
+      "step": 1286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241134148293311e-05,
+      "loss": 0.8674,
+      "step": 1287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2400596914364792e-05,
+      "loss": 0.7416,
+      "step": 1288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2389849404125052e-05,
+      "loss": 0.82,
+      "step": 1289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2379098965383788e-05,
+      "loss": 0.7694,
+      "step": 1290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2368345611314508e-05,
+      "loss": 0.7568,
+      "step": 1291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2357589355094275e-05,
+      "loss": 0.7108,
+      "step": 1292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346830209903718e-05,
+      "loss": 0.7632,
+      "step": 1293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2336068188927002e-05,
+      "loss": 0.8368,
+      "step": 1294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2325303305351819e-05,
+      "loss": 0.8015,
+      "step": 1295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2314535572369368e-05,
+      "loss": 0.7834,
+      "step": 1296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2303765003174342e-05,
+      "loss": 0.7649,
+      "step": 1297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292991610964902e-05,
+      "loss": 0.7927,
+      "step": 1298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2282215408942676e-05,
+      "loss": 0.8497,
+      "step": 1299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2271436410312727e-05,
+      "loss": 0.8281,
+      "step": 1300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2260654628283556e-05,
+      "loss": 0.8478,
+      "step": 1301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2249870076067068e-05,
+      "loss": 0.8083,
+      "step": 1302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239082766878557e-05,
+      "loss": 0.7655,
+      "step": 1303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222829271393671e-05,
+      "loss": 0.8082,
+      "step": 1304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2217499930463563e-05,
+      "loss": 0.8073,
+      "step": 1305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2206704429684504e-05,
+      "loss": 0.6905,
+      "step": 1306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2195906224828246e-05,
+      "loss": 0.8293,
+      "step": 1307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185105329126827e-05,
+      "loss": 0.7729,
+      "step": 1308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2174301755815572e-05,
+      "loss": 0.7728,
+      "step": 1309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163495518133084e-05,
+      "loss": 0.8131,
+      "step": 1310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152686629321243e-05,
+      "loss": 0.8277,
+      "step": 1311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141875102625166e-05,
+      "loss": 0.8327,
+      "step": 1312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213106095129321e-05,
+      "loss": 0.7617,
+      "step": 1313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120244188576946e-05,
+      "loss": 0.8105,
+      "step": 1314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109424827731144e-05,
+      "loss": 0.7047,
+      "step": 1315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2098602882013761e-05,
+      "loss": 0.8298,
+      "step": 1316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2087778364685917e-05,
+      "loss": 0.7625,
+      "step": 1317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2076951289011884e-05,
+      "loss": 0.7538,
+      "step": 1318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066121668259072e-05,
+      "loss": 0.7481,
+      "step": 1319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055289515698008e-05,
+      "loss": 0.8134,
+      "step": 1320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204445484460232e-05,
+      "loss": 0.7938,
+      "step": 1321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2033617668248723e-05,
+      "loss": 0.8081,
+      "step": 1322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2022777999917005e-05,
+      "loss": 0.7861,
+      "step": 1323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2011935852890004e-05,
+      "loss": 0.8134,
+      "step": 1324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001091240453597e-05,
+      "loss": 0.7831,
+      "step": 1325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1990244175896679e-05,
+      "loss": 0.8342,
+      "step": 1326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979394672511156e-05,
+      "loss": 0.8234,
+      "step": 1327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1968542743591917e-05,
+      "loss": 0.7925,
+      "step": 1328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957688402436827e-05,
+      "loss": 0.8256,
+      "step": 1329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19468316623467e-05,
+      "loss": 0.8341,
+      "step": 1330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935972536625302e-05,
+      "loss": 0.7642,
+      "step": 1331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1925111038579309e-05,
+      "loss": 0.7402,
+      "step": 1332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914247181518312e-05,
+      "loss": 0.8072,
+      "step": 1333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1903380978754792e-05,
+      "loss": 0.8471,
+      "step": 1334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892512443604103e-05,
+      "loss": 0.7128,
+      "step": 1335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1881641589384456e-05,
+      "loss": 0.7602,
+      "step": 1336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1870768429416906e-05,
+      "loss": 0.766,
+      "step": 1337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1859892977025333e-05,
+      "loss": 0.6963,
+      "step": 1338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849015245536424e-05,
+      "loss": 0.7539,
+      "step": 1339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838135248279663e-05,
+      "loss": 0.7345,
+      "step": 1340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827252998587302e-05,
+      "loss": 0.8156,
+      "step": 1341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816368509794365e-05,
+      "loss": 0.8135,
+      "step": 1342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805481795238614e-05,
+      "loss": 0.8611,
+      "step": 1343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794592868260535e-05,
+      "loss": 0.8171,
+      "step": 1344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783701742203326e-05,
+      "loss": 0.7766,
+      "step": 1345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1772808430412888e-05,
+      "loss": 0.842,
+      "step": 1346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761912946237793e-05,
+      "loss": 0.7699,
+      "step": 1347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1751015303029272e-05,
+      "loss": 0.8001,
+      "step": 1348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174011551414121e-05,
+      "loss": 0.7699,
+      "step": 1349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729213592930117e-05,
+      "loss": 0.8527,
+      "step": 1350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1718309552755118e-05,
+      "loss": 0.766,
+      "step": 1351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707403406977929e-05,
+      "loss": 0.7909,
+      "step": 1352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1696495168962848e-05,
+      "loss": 0.826,
+      "step": 1353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685584852076746e-05,
+      "loss": 0.8267,
+      "step": 1354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167467246968903e-05,
+      "loss": 0.7847,
+      "step": 1355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663758035171642e-05,
+      "loss": 0.7535,
+      "step": 1356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1652841561899042e-05,
+      "loss": 0.7571,
+      "step": 1357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641923063248185e-05,
+      "loss": 0.7737,
+      "step": 1358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1631002552598503e-05,
+      "loss": 0.7614,
+      "step": 1359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620080043331901e-05,
+      "loss": 0.794,
+      "step": 1360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609155548832734e-05,
+      "loss": 0.7739,
+      "step": 1361
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1598229082487785e-05,
+      "loss": 0.8046,
+      "step": 1362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1587300657686254e-05,
+      "loss": 0.8097,
+      "step": 1363
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1576370287819737e-05,
+      "loss": 0.8389,
+      "step": 1364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1565437986282227e-05,
+      "loss": 0.848,
+      "step": 1365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1554503766470069e-05,
+      "loss": 0.8289,
+      "step": 1366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543567641781967e-05,
+      "loss": 0.854,
+      "step": 1367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532629625618956e-05,
+      "loss": 0.8443,
+      "step": 1368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1521689731384391e-05,
+      "loss": 0.7437,
+      "step": 1369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510747972483924e-05,
+      "loss": 0.7764,
+      "step": 1370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14998043623255e-05,
+      "loss": 0.8074,
+      "step": 1371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488858914319321e-05,
+      "loss": 0.8148,
+      "step": 1372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477911641877853e-05,
+      "loss": 0.8158,
+      "step": 1373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1466962558415791e-05,
+      "loss": 0.7365,
+      "step": 1374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1456011677350052e-05,
+      "loss": 0.7632,
+      "step": 1375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445059012099748e-05,
+      "loss": 0.7506,
+      "step": 1376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434104576086193e-05,
+      "loss": 0.7475,
+      "step": 1377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.8012,
+      "step": 1378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141219044546536e-05,
+      "loss": 0.7565,
+      "step": 1379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140123077771148e-05,
+      "loss": 0.776,
+      "step": 1380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1390269392901096e-05,
+      "loss": 0.7372,
+      "step": 1381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379306304466197e-05,
+      "loss": 0.8248,
+      "step": 1382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368341525840862e-05,
+      "loss": 0.8236,
+      "step": 1383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1357375070461241e-05,
+      "loss": 0.7328,
+      "step": 1384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346406951765534e-05,
+      "loss": 0.8219,
+      "step": 1385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133543718319398e-05,
+      "loss": 0.7934,
+      "step": 1386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324465778188846e-05,
+      "loss": 0.8313,
+      "step": 1387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313492750194399e-05,
+      "loss": 0.7495,
+      "step": 1388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1302518112656898e-05,
+      "loss": 0.7918,
+      "step": 1389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1291541879024568e-05,
+      "loss": 0.8058,
+      "step": 1390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280564062747592e-05,
+      "loss": 0.8149,
+      "step": 1391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269584677278103e-05,
+      "loss": 0.7443,
+      "step": 1392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1258603736070145e-05,
+      "loss": 0.7242,
+      "step": 1393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247621252579668e-05,
+      "loss": 0.7761,
+      "step": 1394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236637240264517e-05,
+      "loss": 0.7955,
+      "step": 1395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1225651712584413e-05,
+      "loss": 0.7282,
+      "step": 1396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1214664683000927e-05,
+      "loss": 0.806,
+      "step": 1397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1203676164977471e-05,
+      "loss": 0.7797,
+      "step": 1398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192686171979288e-05,
+      "loss": 0.7589,
+      "step": 1399
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1181694717473422e-05,
+      "loss": 0.8228,
+      "step": 1400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170701814928707e-05,
+      "loss": 0.7855,
+      "step": 1401
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1159707477815756e-05,
+      "loss": 0.7645,
+      "step": 1402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114871171960694e-05,
+      "loss": 0.7742,
+      "step": 1403
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1137714553776368e-05,
+      "loss": 0.7195,
+      "step": 1404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1126715993799875e-05,
+      "loss": 0.7656,
+      "step": 1405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1115716053155003e-05,
+      "loss": 0.743,
+      "step": 1406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1104714745320991e-05,
+      "loss": 0.7032,
+      "step": 1407
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1093712083778748e-05,
+      "loss": 0.7734,
+      "step": 1408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108270808201084e-05,
+      "loss": 0.7763,
+      "step": 1409
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071702753501484e-05,
+      "loss": 0.7284,
+      "step": 1410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1060696111736515e-05,
+      "loss": 0.7098,
+      "step": 1411
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1049688170203384e-05,
+      "loss": 0.7733,
+      "step": 1412
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1038678942391125e-05,
+      "loss": 0.744,
+      "step": 1413
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1027668441790358e-05,
+      "loss": 0.7443,
+      "step": 1414
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1016656681893256e-05,
+      "loss": 0.7692,
+      "step": 1415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1005643676193545e-05,
+      "loss": 0.7403,
+      "step": 1416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099462943818646e-05,
+      "loss": 0.776,
+      "step": 1417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0983613981368757e-05,
+      "loss": 0.7917,
+      "step": 1418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0972597319238692e-05,
+      "loss": 0.697,
+      "step": 1419
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0961579465295987e-05,
+      "loss": 0.7658,
+      "step": 1420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950560433041825e-05,
+      "loss": 0.8223,
+      "step": 1421
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939540235978844e-05,
+      "loss": 0.7763,
+      "step": 1422
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928518887611099e-05,
+      "loss": 0.7556,
+      "step": 1423
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917496401444055e-05,
+      "loss": 0.8115,
+      "step": 1424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0906472790984575e-05,
+      "loss": 0.7328,
+      "step": 1425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0895448069740902e-05,
+      "loss": 0.7738,
+      "step": 1426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0884422251222636e-05,
+      "loss": 0.7163,
+      "step": 1427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873395348940721e-05,
+      "loss": 0.7998,
+      "step": 1428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862367376407433e-05,
+      "loss": 0.7892,
+      "step": 1429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0851338347136358e-05,
+      "loss": 0.7935,
+      "step": 1430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840308274642377e-05,
+      "loss": 0.7603,
+      "step": 1431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0829277172441648e-05,
+      "loss": 0.7803,
+      "step": 1432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818245054051589e-05,
+      "loss": 0.7683,
+      "step": 1433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080721193299087e-05,
+      "loss": 0.785,
+      "step": 1434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0796177822779384e-05,
+      "loss": 0.822,
+      "step": 1435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785142736938237e-05,
+      "loss": 0.751,
+      "step": 1436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774106688989729e-05,
+      "loss": 0.7504,
+      "step": 1437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0763069692457346e-05,
+      "loss": 0.8584,
+      "step": 1438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0752031760865728e-05,
+      "loss": 0.7699,
+      "step": 1439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740992907740667e-05,
+      "loss": 0.8188,
+      "step": 1440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0729953146609076e-05,
+      "loss": 0.722,
+      "step": 1441
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718912490998992e-05,
+      "loss": 0.7548,
+      "step": 1442
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707870954439541e-05,
+      "loss": 0.7261,
+      "step": 1443
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696828550460928e-05,
+      "loss": 0.7715,
+      "step": 1444
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685785292594424e-05,
+      "loss": 0.7308,
+      "step": 1445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674741194372345e-05,
+      "loss": 0.819,
+      "step": 1446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663696269328034e-05,
+      "loss": 0.8171,
+      "step": 1447
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0652650530995855e-05,
+      "loss": 0.7532,
+      "step": 1448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0641603992911158e-05,
+      "loss": 0.7698,
+      "step": 1449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0630556668610286e-05,
+      "loss": 0.7578,
+      "step": 1450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0619508571630531e-05,
+      "loss": 0.862,
+      "step": 1451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060845971551014e-05,
+      "loss": 0.7109,
+      "step": 1452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059741011378829e-05,
+      "loss": 0.7925,
+      "step": 1453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586359780005076e-05,
+      "loss": 0.7881,
+      "step": 1454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057530872770148e-05,
+      "loss": 0.7607,
+      "step": 1455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0564256970419367e-05,
+      "loss": 0.7129,
+      "step": 1456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055320452170147e-05,
+      "loss": 0.7513,
+      "step": 1457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542151395091369e-05,
+      "loss": 0.7652,
+      "step": 1458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0531097604133473e-05,
+      "loss": 0.7801,
+      "step": 1459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0520043162373001e-05,
+      "loss": 0.7601,
+      "step": 1460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050898808335598e-05,
+      "loss": 0.8199,
+      "step": 1461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0497932380629207e-05,
+      "loss": 0.7929,
+      "step": 1462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486876067740253e-05,
+      "loss": 0.7482,
+      "step": 1463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0475819158237426e-05,
+      "loss": 0.7066,
+      "step": 1464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0464761665669771e-05,
+      "loss": 0.8501,
+      "step": 1465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453703603587051e-05,
+      "loss": 0.7951,
+      "step": 1466
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442644985539718e-05,
+      "loss": 0.7334,
+      "step": 1467
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0431585825078916e-05,
+      "loss": 0.786,
+      "step": 1468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420526135756443e-05,
+      "loss": 0.8158,
+      "step": 1469
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0409465931124755e-05,
+      "loss": 0.7702,
+      "step": 1470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0398405224736927e-05,
+      "loss": 0.8186,
+      "step": 1471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0387344030146665e-05,
+      "loss": 0.7471,
+      "step": 1472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037628236090826e-05,
+      "loss": 0.6832,
+      "step": 1473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0365220230576592e-05,
+      "loss": 0.8193,
+      "step": 1474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0354157652707095e-05,
+      "loss": 0.7071,
+      "step": 1475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034309464085577e-05,
+      "loss": 0.7918,
+      "step": 1476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0332031208579133e-05,
+      "loss": 0.6822,
+      "step": 1477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0320967369434223e-05,
+      "loss": 0.7758,
+      "step": 1478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0309903136978576e-05,
+      "loss": 0.7912,
+      "step": 1479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0298838524770212e-05,
+      "loss": 0.7791,
+      "step": 1480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0287773546367614e-05,
+      "loss": 0.7352,
+      "step": 1481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0276708215329712e-05,
+      "loss": 0.804,
+      "step": 1482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265642545215872e-05,
+      "loss": 0.8244,
+      "step": 1483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0254576549585875e-05,
+      "loss": 0.7758,
+      "step": 1484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0243510241999898e-05,
+      "loss": 0.767,
+      "step": 1485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232443636018502e-05,
+      "loss": 0.7736,
+      "step": 1486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0221376745202612e-05,
+      "loss": 0.6948,
+      "step": 1487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0210309583113508e-05,
+      "loss": 0.792,
+      "step": 1488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199242163312794e-05,
+      "loss": 0.7601,
+      "step": 1489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188174499362395e-05,
+      "loss": 0.8114,
+      "step": 1490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177106604824536e-05,
+      "loss": 0.749,
+      "step": 1491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166038493261723e-05,
+      "loss": 0.7917,
+      "step": 1492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154970178236723e-05,
+      "loss": 0.7744,
+      "step": 1493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143901673312562e-05,
+      "loss": 0.787,
+      "step": 1494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013283299205249e-05,
+      "loss": 0.7628,
+      "step": 1495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0121764148019977e-05,
+      "loss": 0.7504,
+      "step": 1496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110695154778694e-05,
+      "loss": 0.8343,
+      "step": 1497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099626025892491e-05,
+      "loss": 0.8261,
+      "step": 1498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0088556774925386e-05,
+      "loss": 0.7402,
+      "step": 1499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077487415441552e-05,
+      "loss": 0.7642,
+      "step": 1500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0066417961005283e-05,
+      "loss": 0.7618,
+      "step": 1501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055348425181e-05,
+      "loss": 0.7884,
+      "step": 1502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0044278821533218e-05,
+      "loss": 0.832,
+      "step": 1503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033209163626539e-05,
+      "loss": 0.7721,
+      "step": 1504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0022139465025627e-05,
+      "loss": 0.7644,
+      "step": 1505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0011069739295196e-05,
+      "loss": 0.8078,
+      "step": 1506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.7854,
+      "step": 1507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988930260704805e-06,
+      "loss": 0.8412,
+      "step": 1508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97786053497438e-06,
+      "loss": 0.7661,
+      "step": 1509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966790836373465e-06,
+      "loss": 0.7488,
+      "step": 1510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.955721178466785e-06,
+      "loss": 0.7991,
+      "step": 1511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.944651574819004e-06,
+      "loss": 0.7254,
+      "step": 1512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.933582038994719e-06,
+      "loss": 0.7725,
+      "step": 1513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922512584558453e-06,
+      "loss": 0.7173,
+      "step": 1514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.911443225074615e-06,
+      "loss": 0.7749,
+      "step": 1515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90037397410751e-06,
+      "loss": 0.8129,
+      "step": 1516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.889304845221311e-06,
+      "loss": 0.7538,
+      "step": 1517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.878235851980027e-06,
+      "loss": 0.7951,
+      "step": 1518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.867167007947511e-06,
+      "loss": 0.7331,
+      "step": 1519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856098326687441e-06,
+      "loss": 0.7164,
+      "step": 1520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84502982176328e-06,
+      "loss": 0.783,
+      "step": 1521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833961506738282e-06,
+      "loss": 0.8363,
+      "step": 1522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822893395175467e-06,
+      "loss": 0.7682,
+      "step": 1523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811825500637607e-06,
+      "loss": 0.7769,
+      "step": 1524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80075783668721e-06,
+      "loss": 0.7805,
+      "step": 1525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789690416886496e-06,
+      "loss": 0.7269,
+      "step": 1526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77862325479739e-06,
+      "loss": 0.7786,
+      "step": 1527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767556363981503e-06,
+      "loss": 0.8009,
+      "step": 1528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756489758000105e-06,
+      "loss": 0.7317,
+      "step": 1529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.745423450414128e-06,
+      "loss": 0.7576,
+      "step": 1530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734357454784131e-06,
+      "loss": 0.7244,
+      "step": 1531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72329178467029e-06,
+      "loss": 0.7466,
+      "step": 1532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.712226453632392e-06,
+      "loss": 0.7664,
+      "step": 1533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701161475229791e-06,
+      "loss": 0.7475,
+      "step": 1534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690096863021426e-06,
+      "loss": 0.8184,
+      "step": 1535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.679032630565782e-06,
+      "loss": 0.8088,
+      "step": 1536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66796879142087e-06,
+      "loss": 0.765,
+      "step": 1537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656905359144232e-06,
+      "loss": 0.7221,
+      "step": 1538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.645842347292907e-06,
+      "loss": 0.7248,
+      "step": 1539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634779769423412e-06,
+      "loss": 0.7817,
+      "step": 1540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.623717639091743e-06,
+      "loss": 0.8211,
+      "step": 1541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.612655969853337e-06,
+      "loss": 0.7298,
+      "step": 1542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601594775263073e-06,
+      "loss": 0.7471,
+      "step": 1543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59053406887525e-06,
+      "loss": 0.7985,
+      "step": 1544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579473864243559e-06,
+      "loss": 0.8229,
+      "step": 1545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.568414174921085e-06,
+      "loss": 0.8016,
+      "step": 1546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557355014460285e-06,
+      "loss": 0.6921,
+      "step": 1547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.546296396412952e-06,
+      "loss": 0.808,
+      "step": 1548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.535238334330234e-06,
+      "loss": 0.7922,
+      "step": 1549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.524180841762577e-06,
+      "loss": 0.7792,
+      "step": 1550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51312393225975e-06,
+      "loss": 0.737,
+      "step": 1551
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.502067619370794e-06,
+      "loss": 0.7482,
+      "step": 1552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491011916644024e-06,
+      "loss": 0.7005,
+      "step": 1553
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.479956837627e-06,
+      "loss": 0.76,
+      "step": 1554
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468902395866532e-06,
+      "loss": 0.7738,
+      "step": 1555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457848604908634e-06,
+      "loss": 0.7339,
+      "step": 1556
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44679547829853e-06,
+      "loss": 0.7199,
+      "step": 1557
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.435743029580638e-06,
+      "loss": 0.7689,
+      "step": 1558
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424691272298523e-06,
+      "loss": 0.8299,
+      "step": 1559
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413640219994928e-06,
+      "loss": 0.839,
+      "step": 1560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402589886211711e-06,
+      "loss": 0.7333,
+      "step": 1561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391540284489862e-06,
+      "loss": 0.8143,
+      "step": 1562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.380491428369474e-06,
+      "loss": 0.7701,
+      "step": 1563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.369443331389718e-06,
+      "loss": 0.7473,
+      "step": 1564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358396007088843e-06,
+      "loss": 0.7748,
+      "step": 1565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34734946900415e-06,
+      "loss": 0.802,
+      "step": 1566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336303730671968e-06,
+      "loss": 0.7968,
+      "step": 1567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32525880562766e-06,
+      "loss": 0.7842,
+      "step": 1568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31421470740558e-06,
+      "loss": 0.7844,
+      "step": 1569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303171449539074e-06,
+      "loss": 0.7308,
+      "step": 1570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292129045560464e-06,
+      "loss": 0.7189,
+      "step": 1571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281087509001012e-06,
+      "loss": 0.7264,
+      "step": 1572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270046853390924e-06,
+      "loss": 0.7699,
+      "step": 1573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259007092259338e-06,
+      "loss": 0.8072,
+      "step": 1574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.247968239134273e-06,
+      "loss": 0.7383,
+      "step": 1575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236930307542654e-06,
+      "loss": 0.7652,
+      "step": 1576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225893311010273e-06,
+      "loss": 0.84,
+      "step": 1577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214857263061765e-06,
+      "loss": 0.7405,
+      "step": 1578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203822177220621e-06,
+      "loss": 0.8457,
+      "step": 1579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192788067009134e-06,
+      "loss": 0.8226,
+      "step": 1580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181754945948413e-06,
+      "loss": 0.7527,
+      "step": 1581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170722827558357e-06,
+      "loss": 0.7848,
+      "step": 1582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159691725357627e-06,
+      "loss": 0.7947,
+      "step": 1583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148661652863644e-06,
+      "loss": 0.7447,
+      "step": 1584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13763262359257e-06,
+      "loss": 0.8021,
+      "step": 1585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126604651059282e-06,
+      "loss": 0.7461,
+      "step": 1586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11557774877737e-06,
+      "loss": 0.8068,
+      "step": 1587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104551930259101e-06,
+      "loss": 0.6662,
+      "step": 1588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.093527209015428e-06,
+      "loss": 0.7723,
+      "step": 1589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.082503598555949e-06,
+      "loss": 0.7826,
+      "step": 1590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071481112388905e-06,
+      "loss": 0.7824,
+      "step": 1591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060459764021156e-06,
+      "loss": 0.7508,
+      "step": 1592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049439566958176e-06,
+      "loss": 0.7651,
+      "step": 1593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.038420534704015e-06,
+      "loss": 0.8089,
+      "step": 1594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02740268076131e-06,
+      "loss": 0.851,
+      "step": 1595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016386018631246e-06,
+      "loss": 0.7054,
+      "step": 1596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.005370561813545e-06,
+      "loss": 0.8317,
+      "step": 1597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994356323806462e-06,
+      "loss": 0.7489,
+      "step": 1598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.983343318106745e-06,
+      "loss": 0.7818,
+      "step": 1599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.972331558209644e-06,
+      "loss": 0.7738,
+      "step": 1600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961321057608878e-06,
+      "loss": 0.7962,
+      "step": 1601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95031182979662e-06,
+      "loss": 0.762,
+      "step": 1602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939303888263485e-06,
+      "loss": 0.8044,
+      "step": 1603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92829724649852e-06,
+      "loss": 0.8282,
+      "step": 1604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917291917989162e-06,
+      "loss": 0.711,
+      "step": 1605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.906287916221259e-06,
+      "loss": 0.7672,
+      "step": 1606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.895285254679012e-06,
+      "loss": 0.7659,
+      "step": 1607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.884283946845e-06,
+      "loss": 0.8199,
+      "step": 1608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.873284006200129e-06,
+      "loss": 0.8194,
+      "step": 1609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.862285446223635e-06,
+      "loss": 0.7941,
+      "step": 1610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85128828039306e-06,
+      "loss": 0.7144,
+      "step": 1611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.840292522184247e-06,
+      "loss": 0.7516,
+      "step": 1612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829298185071295e-06,
+      "loss": 0.8141,
+      "step": 1613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.818305282526585e-06,
+      "loss": 0.754,
+      "step": 1614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807313828020715e-06,
+      "loss": 0.7625,
+      "step": 1615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79632383502253e-06,
+      "loss": 0.7273,
+      "step": 1616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.785335316999078e-06,
+      "loss": 0.7344,
+      "step": 1617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.774348287415589e-06,
+      "loss": 0.7945,
+      "step": 1618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763362759735485e-06,
+      "loss": 0.7958,
+      "step": 1619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752378747420339e-06,
+      "loss": 0.7765,
+      "step": 1620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74139626392986e-06,
+      "loss": 0.7375,
+      "step": 1621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730415322721897e-06,
+      "loss": 0.7198,
+      "step": 1622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71943593725241e-06,
+      "loss": 0.7573,
+      "step": 1623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.708458120975436e-06,
+      "loss": 0.7675,
+      "step": 1624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.697481887343107e-06,
+      "loss": 0.7511,
+      "step": 1625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.686507249805603e-06,
+      "loss": 0.7847,
+      "step": 1626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675534221811156e-06,
+      "loss": 0.7783,
+      "step": 1627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.664562816806022e-06,
+      "loss": 0.7573,
+      "step": 1628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65359304823447e-06,
+      "loss": 0.7411,
+      "step": 1629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64262492953876e-06,
+      "loss": 0.7132,
+      "step": 1630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63165847415914e-06,
+      "loss": 0.7773,
+      "step": 1631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620693695533804e-06,
+      "loss": 0.7756,
+      "step": 1632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60973060709891e-06,
+      "loss": 0.7043,
+      "step": 1633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.598769222288523e-06,
+      "loss": 0.7381,
+      "step": 1634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587809554534641e-06,
+      "loss": 0.827,
+      "step": 1635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.7788,
+      "step": 1636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56589542391381e-06,
+      "loss": 0.7249,
+      "step": 1637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.554940987900253e-06,
+      "loss": 0.7665,
+      "step": 1638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.543988322649954e-06,
+      "loss": 0.7144,
+      "step": 1639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533037441584212e-06,
+      "loss": 0.6814,
+      "step": 1640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522088358122149e-06,
+      "loss": 0.8243,
+      "step": 1641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511141085680684e-06,
+      "loss": 0.7941,
+      "step": 1642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500195637674504e-06,
+      "loss": 0.7662,
+      "step": 1643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489252027516081e-06,
+      "loss": 0.7559,
+      "step": 1644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478310268615612e-06,
+      "loss": 0.7824,
+      "step": 1645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467370374381047e-06,
+      "loss": 0.7616,
+      "step": 1646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456432358218036e-06,
+      "loss": 0.7602,
+      "step": 1647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445496233529934e-06,
+      "loss": 0.8289,
+      "step": 1648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434562013717775e-06,
+      "loss": 0.8091,
+      "step": 1649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.423629712180265e-06,
+      "loss": 0.8172,
+      "step": 1650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41269934231375e-06,
+      "loss": 0.7381,
+      "step": 1651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40177091751222e-06,
+      "loss": 0.7306,
+      "step": 1652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390844451167268e-06,
+      "loss": 0.6838,
+      "step": 1653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3799199566681e-06,
+      "loss": 0.6902,
+      "step": 1654
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.368997447401503e-06,
+      "loss": 0.7162,
+      "step": 1655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358076936751819e-06,
+      "loss": 0.7169,
+      "step": 1656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34715843810096e-06,
+      "loss": 0.7274,
+      "step": 1657
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336241964828363e-06,
+      "loss": 0.8323,
+      "step": 1658
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325327530310972e-06,
+      "loss": 0.8447,
+      "step": 1659
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314415147923254e-06,
+      "loss": 0.7883,
+      "step": 1660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.303504831037154e-06,
+      "loss": 0.7603,
+      "step": 1661
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.292596593022076e-06,
+      "loss": 0.7005,
+      "step": 1662
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.281690447244887e-06,
+      "loss": 0.8289,
+      "step": 1663
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270786407069884e-06,
+      "loss": 0.7315,
+      "step": 1664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.259884485858791e-06,
+      "loss": 0.7544,
+      "step": 1665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.248984696970732e-06,
+      "loss": 0.7452,
+      "step": 1666
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238087053762212e-06,
+      "loss": 0.7812,
+      "step": 1667
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227191569587112e-06,
+      "loss": 0.8054,
+      "step": 1668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216298257796677e-06,
+      "loss": 0.7489,
+      "step": 1669
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205407131739468e-06,
+      "loss": 0.6939,
+      "step": 1670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19451820476139e-06,
+      "loss": 0.7256,
+      "step": 1671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183631490205636e-06,
+      "loss": 0.7719,
+      "step": 1672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1727470014127e-06,
+      "loss": 0.7795,
+      "step": 1673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161864751720343e-06,
+      "loss": 0.8299,
+      "step": 1674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150984754463578e-06,
+      "loss": 0.8302,
+      "step": 1675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140107022974669e-06,
+      "loss": 0.7696,
+      "step": 1676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.129231570583097e-06,
+      "loss": 0.7536,
+      "step": 1677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118358410615545e-06,
+      "loss": 0.8011,
+      "step": 1678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.107487556395902e-06,
+      "loss": 0.7073,
+      "step": 1679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096619021245211e-06,
+      "loss": 0.7573,
+      "step": 1680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08575281848169e-06,
+      "loss": 0.7238,
+      "step": 1681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074888961420695e-06,
+      "loss": 0.7747,
+      "step": 1682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064027463374702e-06,
+      "loss": 0.7926,
+      "step": 1683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0531683376533e-06,
+      "loss": 0.7642,
+      "step": 1684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042311597563177e-06,
+      "loss": 0.7369,
+      "step": 1685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.031457256408086e-06,
+      "loss": 0.7808,
+      "step": 1686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.020605327488846e-06,
+      "loss": 0.685,
+      "step": 1687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.009755824103324e-06,
+      "loss": 0.691,
+      "step": 1688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.998908759546405e-06,
+      "loss": 0.7003,
+      "step": 1689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.988064147110001e-06,
+      "loss": 0.8612,
+      "step": 1690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.977222000082998e-06,
+      "loss": 0.7839,
+      "step": 1691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966382331751278e-06,
+      "loss": 0.8015,
+      "step": 1692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.955545155397684e-06,
+      "loss": 0.8039,
+      "step": 1693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944710484301995e-06,
+      "loss": 0.8049,
+      "step": 1694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93387833174093e-06,
+      "loss": 0.7374,
+      "step": 1695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923048710988119e-06,
+      "loss": 0.7674,
+      "step": 1696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.912221635314085e-06,
+      "loss": 0.8209,
+      "step": 1697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.901397117986244e-06,
+      "loss": 0.7953,
+      "step": 1698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890575172268858e-06,
+      "loss": 0.7266,
+      "step": 1699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.879755811423056e-06,
+      "loss": 0.7808,
+      "step": 1700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868939048706793e-06,
+      "loss": 0.7672,
+      "step": 1701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858124897374837e-06,
+      "loss": 0.8494,
+      "step": 1702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847313370678759e-06,
+      "loss": 0.7479,
+      "step": 1703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83650448186692e-06,
+      "loss": 0.7601,
+      "step": 1704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.825698244184432e-06,
+      "loss": 0.7267,
+      "step": 1705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814894670873171e-06,
+      "loss": 0.743,
+      "step": 1706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804093775171756e-06,
+      "loss": 0.7353,
+      "step": 1707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.7932955703155e-06,
+      "loss": 0.7561,
+      "step": 1708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.782500069536444e-06,
+      "loss": 0.8029,
+      "step": 1709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771707286063294e-06,
+      "loss": 0.7942,
+      "step": 1710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760917233121443e-06,
+      "loss": 0.7557,
+      "step": 1711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750129923932939e-06,
+      "loss": 0.793,
+      "step": 1712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.739345371716447e-06,
+      "loss": 0.7915,
+      "step": 1713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728563589687275e-06,
+      "loss": 0.7655,
+      "step": 1714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71778459105733e-06,
+      "loss": 0.802,
+      "step": 1715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707008389035102e-06,
+      "loss": 0.8471,
+      "step": 1716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.696234996825663e-06,
+      "loss": 0.8058,
+      "step": 1717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685464427630634e-06,
+      "loss": 0.7521,
+      "step": 1718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.674696694648183e-06,
+      "loss": 0.7505,
+      "step": 1719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.663931811073003e-06,
+      "loss": 0.8163,
+      "step": 1720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653169790096287e-06,
+      "loss": 0.7133,
+      "step": 1721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642410644905726e-06,
+      "loss": 0.7818,
+      "step": 1722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631654388685496e-06,
+      "loss": 0.7651,
+      "step": 1723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6209010346162126e-06,
+      "loss": 0.7926,
+      "step": 1724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.610150595874951e-06,
+      "loss": 0.6738,
+      "step": 1725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.599403085635208e-06,
+      "loss": 0.7544,
+      "step": 1726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588658517066893e-06,
+      "loss": 0.7853,
+      "step": 1727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5779169033363135e-06,
+      "loss": 0.783,
+      "step": 1728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567178257606147e-06,
+      "loss": 0.6888,
+      "step": 1729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556442593035445e-06,
+      "loss": 0.8304,
+      "step": 1730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545709922779604e-06,
+      "loss": 0.7328,
+      "step": 1731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534980259990341e-06,
+      "loss": 0.7972,
+      "step": 1732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524253617815701e-06,
+      "loss": 0.8316,
+      "step": 1733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513530009400021e-06,
+      "loss": 0.757,
+      "step": 1734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50280944788392e-06,
+      "loss": 0.7931,
+      "step": 1735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492091946404287e-06,
+      "loss": 0.6622,
+      "step": 1736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481377518094253e-06,
+      "loss": 0.7031,
+      "step": 1737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470666176083193e-06,
+      "loss": 0.6596,
+      "step": 1738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4599579334967e-06,
+      "loss": 0.798,
+      "step": 1739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449252803456557e-06,
+      "loss": 0.7182,
+      "step": 1740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438550799080746e-06,
+      "loss": 0.8107,
+      "step": 1741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4278519334834175e-06,
+      "loss": 0.8051,
+      "step": 1742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.417156219774866e-06,
+      "loss": 0.6802,
+      "step": 1743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40646367106153e-06,
+      "loss": 0.7486,
+      "step": 1744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.395774300445973e-06,
+      "loss": 0.7747,
+      "step": 1745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3850881210268546e-06,
+      "loss": 0.7633,
+      "step": 1746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3744051458989395e-06,
+      "loss": 0.6904,
+      "step": 1747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363725388153046e-06,
+      "loss": 0.7674,
+      "step": 1748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3530488608760645e-06,
+      "loss": 0.8098,
+      "step": 1749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342375577150928e-06,
+      "loss": 0.6673,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.331705550056579e-06,
+      "loss": 0.7782,
+      "step": 1751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.321038792667987e-06,
+      "loss": 0.8076,
+      "step": 1752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310375318056107e-06,
+      "loss": 0.7472,
+      "step": 1753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.299715139287868e-06,
+      "loss": 0.7902,
+      "step": 1754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289058269426169e-06,
+      "loss": 0.8081,
+      "step": 1755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278404721529843e-06,
+      "loss": 0.7922,
+      "step": 1756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.267754508653662e-06,
+      "loss": 0.853,
+      "step": 1757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.257107643848311e-06,
+      "loss": 0.7449,
+      "step": 1758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246464140160365e-06,
+      "loss": 0.763,
+      "step": 1759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.235824010632284e-06,
+      "loss": 0.8198,
+      "step": 1760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2251872683024025e-06,
+      "loss": 0.7661,
+      "step": 1761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214553926204884e-06,
+      "loss": 0.7196,
+      "step": 1762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.203923997369746e-06,
+      "loss": 0.7354,
+      "step": 1763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.193297494822811e-06,
+      "loss": 0.7318,
+      "step": 1764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182674431585703e-06,
+      "loss": 0.7208,
+      "step": 1765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1720548206758445e-06,
+      "loss": 0.6823,
+      "step": 1766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161438675106408e-06,
+      "loss": 0.7605,
+      "step": 1767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150826007886334e-06,
+      "loss": 0.7341,
+      "step": 1768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1402168320203e-06,
+      "loss": 0.7656,
+      "step": 1769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.129611160508695e-06,
+      "loss": 0.7318,
+      "step": 1770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119009006347625e-06,
+      "loss": 0.8075,
+      "step": 1771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10841038252888e-06,
+      "loss": 0.7556,
+      "step": 1772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.097815302039924e-06,
+      "loss": 0.7709,
+      "step": 1773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087223777863883e-06,
+      "loss": 0.8149,
+      "step": 1774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076635822979518e-06,
+      "loss": 0.7856,
+      "step": 1775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066051450361222e-06,
+      "loss": 0.7621,
+      "step": 1776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055470672979003e-06,
+      "loss": 0.7099,
+      "step": 1777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.044893503798448e-06,
+      "loss": 0.7667,
+      "step": 1778
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.034319955780735e-06,
+      "loss": 0.7002,
+      "step": 1779
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.023750041882609e-06,
+      "loss": 0.7576,
+      "step": 1780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013183775056341e-06,
+      "loss": 0.7922,
+      "step": 1781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.002621168249759e-06,
+      "loss": 0.7896,
+      "step": 1782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992062234406185e-06,
+      "loss": 0.7176,
+      "step": 1783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.981506986464448e-06,
+      "loss": 0.7805,
+      "step": 1784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.970955437358868e-06,
+      "loss": 0.8315,
+      "step": 1785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960407600019217e-06,
+      "loss": 0.772,
+      "step": 1786
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9498634873707314e-06,
+      "loss": 0.7774,
+      "step": 1787
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939323112334082e-06,
+      "loss": 0.7284,
+      "step": 1788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9287864878253475e-06,
+      "loss": 0.6832,
+      "step": 1789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91825362675603e-06,
+      "loss": 0.7796,
+      "step": 1790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.907724542032999e-06,
+      "loss": 0.7574,
+      "step": 1791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897199246558515e-06,
+      "loss": 0.7567,
+      "step": 1792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886677753230184e-06,
+      "loss": 0.7633,
+      "step": 1793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876160074940959e-06,
+      "loss": 0.7483,
+      "step": 1794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865646224579108e-06,
+      "loss": 0.817,
+      "step": 1795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855136215028226e-06,
+      "loss": 0.782,
+      "step": 1796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.844630059167181e-06,
+      "loss": 0.7541,
+      "step": 1797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834127769870134e-06,
+      "loss": 0.7891,
+      "step": 1798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.823629360006504e-06,
+      "loss": 0.7735,
+      "step": 1799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.813134842440949e-06,
+      "loss": 0.7648,
+      "step": 1800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802644230033373e-06,
+      "loss": 0.6781,
+      "step": 1801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792157535638874e-06,
+      "loss": 0.695,
+      "step": 1802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781674772107766e-06,
+      "loss": 0.7305,
+      "step": 1803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771195952285541e-06,
+      "loss": 0.7475,
+      "step": 1804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760721089012851e-06,
+      "loss": 0.7545,
+      "step": 1805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750250195125509e-06,
+      "loss": 0.7589,
+      "step": 1806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.739783283454469e-06,
+      "loss": 0.7747,
+      "step": 1807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729320366825785e-06,
+      "loss": 0.7673,
+      "step": 1808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718861458060635e-06,
+      "loss": 0.7122,
+      "step": 1809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708406569975274e-06,
+      "loss": 0.7411,
+      "step": 1810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697955715381039e-06,
+      "loss": 0.7778,
+      "step": 1811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687508907084319e-06,
+      "loss": 0.7479,
+      "step": 1812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6770661578865444e-06,
+      "loss": 0.8416,
+      "step": 1813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666627480584173e-06,
+      "loss": 0.7654,
+      "step": 1814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656192887968675e-06,
+      "loss": 0.7306,
+      "step": 1815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.645762392826509e-06,
+      "loss": 0.7384,
+      "step": 1816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635336007939119e-06,
+      "loss": 0.768,
+      "step": 1817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624913746082913e-06,
+      "loss": 0.7259,
+      "step": 1818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614495620029238e-06,
+      "loss": 0.7592,
+      "step": 1819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604081642544386e-06,
+      "loss": 0.7832,
+      "step": 1820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593671826389551e-06,
+      "loss": 0.7652,
+      "step": 1821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.583266184320836e-06,
+      "loss": 0.7011,
+      "step": 1822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.572864729089234e-06,
+      "loss": 0.6973,
+      "step": 1823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562467473440591e-06,
+      "loss": 0.7546,
+      "step": 1824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.552074430115624e-06,
+      "loss": 0.7676,
+      "step": 1825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5416856118498874e-06,
+      "loss": 0.7464,
+      "step": 1826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.53130103137374e-06,
+      "loss": 0.7816,
+      "step": 1827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.520920701412371e-06,
+      "loss": 0.7482,
+      "step": 1828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.510544634685741e-06,
+      "loss": 0.7167,
+      "step": 1829
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500172843908601e-06,
+      "loss": 0.7092,
+      "step": 1830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.489805341790456e-06,
+      "loss": 0.7236,
+      "step": 1831
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479442141035557e-06,
+      "loss": 0.8502,
+      "step": 1832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46908325434288e-06,
+      "loss": 0.7129,
+      "step": 1833
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458728694406124e-06,
+      "loss": 0.8093,
+      "step": 1834
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448378473913677e-06,
+      "loss": 0.7717,
+      "step": 1835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438032605548611e-06,
+      "loss": 0.6707,
+      "step": 1836
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.427691101988673e-06,
+      "loss": 0.8128,
+      "step": 1837
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.417353975906248e-06,
+      "loss": 0.7489,
+      "step": 1838
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4070212399683716e-06,
+      "loss": 0.7349,
+      "step": 1839
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.396692906836686e-06,
+      "loss": 0.7388,
+      "step": 1840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386368989167445e-06,
+      "loss": 0.7564,
+      "step": 1841
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3760494996114965e-06,
+      "loss": 0.7372,
+      "step": 1842
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3657344508142495e-06,
+      "loss": 0.7764,
+      "step": 1843
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.355423855415681e-06,
+      "loss": 0.7621,
+      "step": 1844
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3451177260503135e-06,
+      "loss": 0.7119,
+      "step": 1845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334816075347185e-06,
+      "loss": 0.7611,
+      "step": 1846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.32451891592986e-06,
+      "loss": 0.7653,
+      "step": 1847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.314226260416383e-06,
+      "loss": 0.7929,
+      "step": 1848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303938121419295e-06,
+      "loss": 0.7287,
+      "step": 1849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293654511545593e-06,
+      "loss": 0.7598,
+      "step": 1850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.283375443396726e-06,
+      "loss": 0.7773,
+      "step": 1851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273100929568579e-06,
+      "loss": 0.7984,
+      "step": 1852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.262830982651458e-06,
+      "loss": 0.7341,
+      "step": 1853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252565615230065e-06,
+      "loss": 0.6987,
+      "step": 1854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242304839883502e-06,
+      "loss": 0.8173,
+      "step": 1855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.232048669185231e-06,
+      "loss": 0.7343,
+      "step": 1856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221797115703081e-06,
+      "loss": 0.7373,
+      "step": 1857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.211550191999223e-06,
+      "loss": 0.8179,
+      "step": 1858
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2013079106301454e-06,
+      "loss": 0.7894,
+      "step": 1859
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1910702841466605e-06,
+      "loss": 0.7727,
+      "step": 1860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18083732509387e-06,
+      "loss": 0.7943,
+      "step": 1861
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170609046011151e-06,
+      "loss": 0.7461,
+      "step": 1862
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.160385459432153e-06,
+      "loss": 0.7364,
+      "step": 1863
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.150166577884781e-06,
+      "loss": 0.7481,
+      "step": 1864
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.139952413891157e-06,
+      "loss": 0.6619,
+      "step": 1865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12974297996764e-06,
+      "loss": 0.7911,
+      "step": 1866
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119538288624778e-06,
+      "loss": 0.7984,
+      "step": 1867
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109338352367322e-06,
+      "loss": 0.6932,
+      "step": 1868
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099143183694185e-06,
+      "loss": 0.7584,
+      "step": 1869
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088952795098442e-06,
+      "loss": 0.7639,
+      "step": 1870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.078767199067308e-06,
+      "loss": 0.7721,
+      "step": 1871
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068586408082133e-06,
+      "loss": 0.7326,
+      "step": 1872
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.058410434618367e-06,
+      "loss": 0.7333,
+      "step": 1873
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.04823929114557e-06,
+      "loss": 0.8021,
+      "step": 1874
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.038072990127369e-06,
+      "loss": 0.6811,
+      "step": 1875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027911544021465e-06,
+      "loss": 0.7763,
+      "step": 1876
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0177549652796185e-06,
+      "loss": 0.6702,
+      "step": 1877
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.007603266347603e-06,
+      "loss": 0.7344,
+      "step": 1878
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.997456459665237e-06,
+      "loss": 0.7081,
+      "step": 1879
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.987314557666331e-06,
+      "loss": 0.7221,
+      "step": 1880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977177572778679e-06,
+      "loss": 0.8041,
+      "step": 1881
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.967045517424062e-06,
+      "loss": 0.7679,
+      "step": 1882
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956918404018221e-06,
+      "loss": 0.7632,
+      "step": 1883
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946796244970831e-06,
+      "loss": 0.7741,
+      "step": 1884
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936679052685505e-06,
+      "loss": 0.7119,
+      "step": 1885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.92656683955976e-06,
+      "loss": 0.829,
+      "step": 1886
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916459617985022e-06,
+      "loss": 0.8005,
+      "step": 1887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.906357400346596e-06,
+      "loss": 0.7388,
+      "step": 1888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896260199023654e-06,
+      "loss": 0.7735,
+      "step": 1889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886168026389222e-06,
+      "loss": 0.7254,
+      "step": 1890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876080894810167e-06,
+      "loss": 0.7477,
+      "step": 1891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8659988166471715e-06,
+      "loss": 0.7786,
+      "step": 1892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.855921804254734e-06,
+      "loss": 0.7038,
+      "step": 1893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.754,
+      "step": 1894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.835783026168447e-06,
+      "loss": 0.785,
+      "step": 1895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.825721285152497e-06,
+      "loss": 0.7111,
+      "step": 1896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815664659262845e-06,
+      "loss": 0.7855,
+      "step": 1897
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80561316082281e-06,
+      "loss": 0.7123,
+      "step": 1898
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79556680214941e-06,
+      "loss": 0.7762,
+      "step": 1899
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.78552559555337e-06,
+      "loss": 0.777,
+      "step": 1900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7754895533390975e-06,
+      "loss": 0.7526,
+      "step": 1901
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7654586878046795e-06,
+      "loss": 0.819,
+      "step": 1902
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755433011241851e-06,
+      "loss": 0.7897,
+      "step": 1903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745412535935995e-06,
+      "loss": 0.7872,
+      "step": 1904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.735397274166117e-06,
+      "loss": 0.7466,
+      "step": 1905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.725387238204831e-06,
+      "loss": 0.7102,
+      "step": 1906
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.715382440318364e-06,
+      "loss": 0.7083,
+      "step": 1907
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705382892766498e-06,
+      "loss": 0.7266,
+      "step": 1908
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695388607802603e-06,
+      "loss": 0.7349,
+      "step": 1909
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685399597673591e-06,
+      "loss": 0.7345,
+      "step": 1910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675415874619911e-06,
+      "loss": 0.7268,
+      "step": 1911
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665437450875534e-06,
+      "loss": 0.6828,
+      "step": 1912
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6554643386679375e-06,
+      "loss": 0.7637,
+      "step": 1913
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645496550218089e-06,
+      "loss": 0.6887,
+      "step": 1914
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635534097740435e-06,
+      "loss": 0.7347,
+      "step": 1915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62557699344288e-06,
+      "loss": 0.6705,
+      "step": 1916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.615625249526774e-06,
+      "loss": 0.756,
+      "step": 1917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605678878186911e-06,
+      "loss": 0.6577,
+      "step": 1918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5957378916114764e-06,
+      "loss": 0.7758,
+      "step": 1919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.58580230198209e-06,
+      "loss": 0.738,
+      "step": 1920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575872121473722e-06,
+      "loss": 0.7586,
+      "step": 1921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565947362254747e-06,
+      "loss": 0.7882,
+      "step": 1922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.556028036486873e-06,
+      "loss": 0.7415,
+      "step": 1923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.546114156325166e-06,
+      "loss": 0.6844,
+      "step": 1924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5362057339180075e-06,
+      "loss": 0.708,
+      "step": 1925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526302781407098e-06,
+      "loss": 0.777,
+      "step": 1926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.516405310927431e-06,
+      "loss": 0.7669,
+      "step": 1927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506513334607281e-06,
+      "loss": 0.7369,
+      "step": 1928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496626864568204e-06,
+      "loss": 0.7132,
+      "step": 1929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4867459129249846e-06,
+      "loss": 0.7148,
+      "step": 1930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.476870491785673e-06,
+      "loss": 0.7466,
+      "step": 1931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.467000613251516e-06,
+      "loss": 0.8082,
+      "step": 1932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4571362894169795e-06,
+      "loss": 0.766,
+      "step": 1933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447277532369738e-06,
+      "loss": 0.7378,
+      "step": 1934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437424354190611e-06,
+      "loss": 0.7709,
+      "step": 1935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427576766953615e-06,
+      "loss": 0.7748,
+      "step": 1936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.417734782725896e-06,
+      "loss": 0.7621,
+      "step": 1937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.407898413567741e-06,
+      "loss": 0.7058,
+      "step": 1938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398067671532554e-06,
+      "loss": 0.7164,
+      "step": 1939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3882425686668425e-06,
+      "loss": 0.7672,
+      "step": 1940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3784231170102075e-06,
+      "loss": 0.7559,
+      "step": 1941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368609328595323e-06,
+      "loss": 0.7112,
+      "step": 1942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.358801215447923e-06,
+      "loss": 0.78,
+      "step": 1943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3489987895867835e-06,
+      "loss": 0.7688,
+      "step": 1944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.339202063023727e-06,
+      "loss": 0.7743,
+      "step": 1945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3294110477635664e-06,
+      "loss": 0.7461,
+      "step": 1946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3196257558041386e-06,
+      "loss": 0.7867,
+      "step": 1947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309846199136258e-06,
+      "loss": 0.7288,
+      "step": 1948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30007238974371e-06,
+      "loss": 0.7523,
+      "step": 1949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2903043396032405e-06,
+      "loss": 0.7569,
+      "step": 1950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280542060684535e-06,
+      "loss": 0.7755,
+      "step": 1951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2707855649502085e-06,
+      "loss": 0.6702,
+      "step": 1952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261034864355791e-06,
+      "loss": 0.7655,
+      "step": 1953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2512899708497086e-06,
+      "loss": 0.7354,
+      "step": 1954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.24155089637327e-06,
+      "loss": 0.653,
+      "step": 1955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231817652860665e-06,
+      "loss": 0.8004,
+      "step": 1956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222090252238916e-06,
+      "loss": 0.7385,
+      "step": 1957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212368706427913e-06,
+      "loss": 0.7171,
+      "step": 1958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2026530273403405e-06,
+      "loss": 0.6798,
+      "step": 1959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192943226881724e-06,
+      "loss": 0.757,
+      "step": 1960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183239316950367e-06,
+      "loss": 0.6789,
+      "step": 1961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.17354130943736e-06,
+      "loss": 0.7688,
+      "step": 1962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163849216226562e-06,
+      "loss": 0.7396,
+      "step": 1963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15416304919458e-06,
+      "loss": 0.7838,
+      "step": 1964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.144482820210764e-06,
+      "loss": 0.706,
+      "step": 1965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.134808541137183e-06,
+      "loss": 0.7628,
+      "step": 1966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.125140223828629e-06,
+      "loss": 0.7376,
+      "step": 1967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.11547788013256e-06,
+      "loss": 0.7938,
+      "step": 1968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105821521889147e-06,
+      "loss": 0.7601,
+      "step": 1969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096171160931198e-06,
+      "loss": 0.7701,
+      "step": 1970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0865268090841956e-06,
+      "loss": 0.7406,
+      "step": 1971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.076888478166247e-06,
+      "loss": 0.7652,
+      "step": 1972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.06725617998807e-06,
+      "loss": 0.8217,
+      "step": 1973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057629926353016e-06,
+      "loss": 0.7206,
+      "step": 1974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048009729057012e-06,
+      "loss": 0.7303,
+      "step": 1975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038395599888566e-06,
+      "loss": 0.7615,
+      "step": 1976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.028787550628752e-06,
+      "loss": 0.737,
+      "step": 1977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0191855930511946e-06,
+      "loss": 0.7345,
+      "step": 1978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.009589738922051e-06,
+      "loss": 0.7772,
+      "step": 1979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.7974,
+      "step": 1980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990416388036233e-06,
+      "loss": 0.7537,
+      "step": 1981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9808389147744195e-06,
+      "loss": 0.6977,
+      "step": 1982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971267591950727e-06,
+      "loss": 0.6988,
+      "step": 1983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961702431293759e-06,
+      "loss": 0.687,
+      "step": 1984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9521434445246006e-06,
+      "loss": 0.7203,
+      "step": 1985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94259064335674e-06,
+      "loss": 0.738,
+      "step": 1986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933044039496107e-06,
+      "loss": 0.7711,
+      "step": 1987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9235036446410325e-06,
+      "loss": 0.7575,
+      "step": 1988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913969470482233e-06,
+      "loss": 0.8017,
+      "step": 1989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904441528702806e-06,
+      "loss": 0.7932,
+      "step": 1990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.894919830978212e-06,
+      "loss": 0.7555,
+      "step": 1991
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8854043889762615e-06,
+      "loss": 0.7255,
+      "step": 1992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875895214357093e-06,
+      "loss": 0.7532,
+      "step": 1993
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86639231877318e-06,
+      "loss": 0.7143,
+      "step": 1994
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856895713869279e-06,
+      "loss": 0.7628,
+      "step": 1995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847405411282462e-06,
+      "loss": 0.6857,
+      "step": 1996
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837921422642054e-06,
+      "loss": 0.7672,
+      "step": 1997
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828443759569666e-06,
+      "loss": 0.7113,
+      "step": 1998
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818972433679145e-06,
+      "loss": 0.7821,
+      "step": 1999
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809507456576573e-06,
+      "loss": 0.6969,
+      "step": 2000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.800048839860254e-06,
+      "loss": 0.7209,
+      "step": 2001
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790596595120699e-06,
+      "loss": 0.6855,
+      "step": 2002
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781150733940608e-06,
+      "loss": 0.7236,
+      "step": 2003
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771711267894863e-06,
+      "loss": 0.6922,
+      "step": 2004
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762278208550505e-06,
+      "loss": 0.7754,
+      "step": 2005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.752851567466723e-06,
+      "loss": 0.6949,
+      "step": 2006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7434313561948565e-06,
+      "loss": 0.7952,
+      "step": 2007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.734017586278337e-06,
+      "loss": 0.7204,
+      "step": 2008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724610269252732e-06,
+      "loss": 0.7366,
+      "step": 2009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715209416645685e-06,
+      "loss": 0.7807,
+      "step": 2010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7058150399769245e-06,
+      "loss": 0.6887,
+      "step": 2011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6964271507582385e-06,
+      "loss": 0.7473,
+      "step": 2012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.687045760493468e-06,
+      "loss": 0.7112,
+      "step": 2013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677670880678493e-06,
+      "loss": 0.738,
+      "step": 2014
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6683025228012105e-06,
+      "loss": 0.7399,
+      "step": 2015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65894069834153e-06,
+      "loss": 0.7508,
+      "step": 2016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649585418771348e-06,
+      "loss": 0.6844,
+      "step": 2017
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6402366955545566e-06,
+      "loss": 0.6923,
+      "step": 2018
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630894540146994e-06,
+      "loss": 0.8087,
+      "step": 2019
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621558963996458e-06,
+      "loss": 0.7697,
+      "step": 2020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.612229978542698e-06,
+      "loss": 0.7906,
+      "step": 2021
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60290759521736e-06,
+      "loss": 0.7692,
+      "step": 2022
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593591825444028e-06,
+      "loss": 0.7652,
+      "step": 2023
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584282680638155e-06,
+      "loss": 0.7472,
+      "step": 2024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574980172207101e-06,
+      "loss": 0.7189,
+      "step": 2025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565684311550077e-06,
+      "loss": 0.7372,
+      "step": 2026
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5563951100581534e-06,
+      "loss": 0.71,
+      "step": 2027
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547112579114242e-06,
+      "loss": 0.6685,
+      "step": 2028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537836730093077e-06,
+      "loss": 0.7462,
+      "step": 2029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.528567574361207e-06,
+      "loss": 0.7547,
+      "step": 2030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.519305123276974e-06,
+      "loss": 0.7998,
+      "step": 2031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.510049388190518e-06,
+      "loss": 0.6935,
+      "step": 2032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500800380443729e-06,
+      "loss": 0.7302,
+      "step": 2033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491558111370274e-06,
+      "loss": 0.74,
+      "step": 2034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482322592295541e-06,
+      "loss": 0.7484,
+      "step": 2035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.473093834536667e-06,
+      "loss": 0.7194,
+      "step": 2036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463871849402493e-06,
+      "loss": 0.7435,
+      "step": 2037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454656648193559e-06,
+      "loss": 0.7617,
+      "step": 2038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4454482422021005e-06,
+      "loss": 0.6526,
+      "step": 2039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436246642712019e-06,
+      "loss": 0.7258,
+      "step": 2040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427051860998877e-06,
+      "loss": 0.7892,
+      "step": 2041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.417863908329883e-06,
+      "loss": 0.7223,
+      "step": 2042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40868279596388e-06,
+      "loss": 0.7274,
+      "step": 2043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399508535151321e-06,
+      "loss": 0.799,
+      "step": 2044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39034113713428e-06,
+      "loss": 0.7587,
+      "step": 2045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381180613146396e-06,
+      "loss": 0.7069,
+      "step": 2046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372026974412907e-06,
+      "loss": 0.7594,
+      "step": 2047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3628802321506055e-06,
+      "loss": 0.6681,
+      "step": 2048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353740397567832e-06,
+      "loss": 0.7552,
+      "step": 2049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344607481864466e-06,
+      "loss": 0.74,
+      "step": 2050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335481496231896e-06,
+      "loss": 0.7225,
+      "step": 2051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32636245185304e-06,
+      "loss": 0.823,
+      "step": 2052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317250359902295e-06,
+      "loss": 0.721,
+      "step": 2053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3081452315455415e-06,
+      "loss": 0.8099,
+      "step": 2054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299047077940126e-06,
+      "loss": 0.7988,
+      "step": 2055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2899559102348585e-06,
+      "loss": 0.7906,
+      "step": 2056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280871739569972e-06,
+      "loss": 0.726,
+      "step": 2057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.271794577077132e-06,
+      "loss": 0.7338,
+      "step": 2058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262724433879427e-06,
+      "loss": 0.6879,
+      "step": 2059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253661321091322e-06,
+      "loss": 0.822,
+      "step": 2060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244605249818696e-06,
+      "loss": 0.7377,
+      "step": 2061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235556231158765e-06,
+      "loss": 0.7181,
+      "step": 2062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.226514276200134e-06,
+      "loss": 0.7413,
+      "step": 2063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.217479396022737e-06,
+      "loss": 0.6444,
+      "step": 2064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208451601697836e-06,
+      "loss": 0.7465,
+      "step": 2065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19943090428802e-06,
+      "loss": 0.7299,
+      "step": 2066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190417314847172e-06,
+      "loss": 0.8116,
+      "step": 2067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181410844420473e-06,
+      "loss": 0.7209,
+      "step": 2068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1724115040443755e-06,
+      "loss": 0.7978,
+      "step": 2069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1634193047465956e-06,
+      "loss": 0.7618,
+      "step": 2070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154434257546095e-06,
+      "loss": 0.668,
+      "step": 2071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.145456373453088e-06,
+      "loss": 0.7213,
+      "step": 2072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.136485663468982e-06,
+      "loss": 0.7435,
+      "step": 2073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.127522138586424e-06,
+      "loss": 0.6776,
+      "step": 2074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118565809789238e-06,
+      "loss": 0.7475,
+      "step": 2075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109616688052432e-06,
+      "loss": 0.7419,
+      "step": 2076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.10067478434219e-06,
+      "loss": 0.7239,
+      "step": 2077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091740109615842e-06,
+      "loss": 0.7403,
+      "step": 2078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082812674821865e-06,
+      "loss": 0.6385,
+      "step": 2079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.073892490899865e-06,
+      "loss": 0.6826,
+      "step": 2080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064979568780558e-06,
+      "loss": 0.7361,
+      "step": 2081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0560739193857625e-06,
+      "loss": 0.7281,
+      "step": 2082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047175553628397e-06,
+      "loss": 0.7595,
+      "step": 2083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038284482412432e-06,
+      "loss": 0.7339,
+      "step": 2084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02940071663292e-06,
+      "loss": 0.7435,
+      "step": 2085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.020524267175954e-06,
+      "loss": 0.7516,
+      "step": 2086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0116551449186555e-06,
+      "loss": 0.7732,
+      "step": 2087
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.002793360729177e-06,
+      "loss": 0.7432,
+      "step": 2088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.993938925466674e-06,
+      "loss": 0.7389,
+      "step": 2089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9850918499812976e-06,
+      "loss": 0.7403,
+      "step": 2090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.976252145114179e-06,
+      "loss": 0.727,
+      "step": 2091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.96741982169742e-06,
+      "loss": 0.7263,
+      "step": 2092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958594890554073e-06,
+      "loss": 0.726,
+      "step": 2093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949777362498145e-06,
+      "loss": 0.7819,
+      "step": 2094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9409672483345465e-06,
+      "loss": 0.6682,
+      "step": 2095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932164558859133e-06,
+      "loss": 0.7376,
+      "step": 2096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.923369304858635e-06,
+      "loss": 0.7637,
+      "step": 2097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914581497110684e-06,
+      "loss": 0.761,
+      "step": 2098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905801146383798e-06,
+      "loss": 0.7392,
+      "step": 2099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897028263437329e-06,
+      "loss": 0.7567,
+      "step": 2100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888262859021508e-06,
+      "loss": 0.7702,
+      "step": 2101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879504943877383e-06,
+      "loss": 0.7582,
+      "step": 2102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870754528736829e-06,
+      "loss": 0.7536,
+      "step": 2103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862011624322534e-06,
+      "loss": 0.7354,
+      "step": 2104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85327624134798e-06,
+      "loss": 0.7447,
+      "step": 2105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844548390517433e-06,
+      "loss": 0.7511,
+      "step": 2106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.835828082525925e-06,
+      "loss": 0.7516,
+      "step": 2107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827115328059253e-06,
+      "loss": 0.759,
+      "step": 2108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818410137793947e-06,
+      "loss": 0.7408,
+      "step": 2109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8097125223972864e-06,
+      "loss": 0.7326,
+      "step": 2110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801022492527242e-06,
+      "loss": 0.6813,
+      "step": 2111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7923400588325156e-06,
+      "loss": 0.7315,
+      "step": 2112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7836652319524835e-06,
+      "loss": 0.751,
+      "step": 2113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7749980225172077e-06,
+      "loss": 0.7737,
+      "step": 2114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7663384411474145e-06,
+      "loss": 0.7801,
+      "step": 2115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7576864984544814e-06,
+      "loss": 0.6797,
+      "step": 2116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7490422050404273e-06,
+      "loss": 0.8028,
+      "step": 2117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.740405571497897e-06,
+      "loss": 0.7324,
+      "step": 2118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.73177660841015e-06,
+      "loss": 0.7478,
+      "step": 2119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7231553263510424e-06,
+      "loss": 0.7111,
+      "step": 2120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.71454173588503e-06,
+      "loss": 0.7156,
+      "step": 2121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7059358475671225e-06,
+      "loss": 0.6761,
+      "step": 2122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6973376719429134e-06,
+      "loss": 0.7638,
+      "step": 2123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.688747219548532e-06,
+      "loss": 0.6947,
+      "step": 2124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680164500910646e-06,
+      "loss": 0.7282,
+      "step": 2125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6715895265464485e-06,
+      "loss": 0.7244,
+      "step": 2126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6630223069636416e-06,
+      "loss": 0.7673,
+      "step": 2127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.654462852660423e-06,
+      "loss": 0.7623,
+      "step": 2128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6459111741254792e-06,
+      "loss": 0.7457,
+      "step": 2129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6373672818379637e-06,
+      "loss": 0.7488,
+      "step": 2130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6288311862674885e-06,
+      "loss": 0.696,
+      "step": 2131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.620302897874123e-06,
+      "loss": 0.7565,
+      "step": 2132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.611782427108348e-06,
+      "loss": 0.7035,
+      "step": 2133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6032697844110896e-06,
+      "loss": 0.7462,
+      "step": 2134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.594764980213661e-06,
+      "loss": 0.6845,
+      "step": 2135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5862680249377768e-06,
+      "loss": 0.7575,
+      "step": 2136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5777789289955454e-06,
+      "loss": 0.6908,
+      "step": 2137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5692977027894214e-06,
+      "loss": 0.728,
+      "step": 2138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560824356712238e-06,
+      "loss": 0.7786,
+      "step": 2139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5523589011471592e-06,
+      "loss": 0.7609,
+      "step": 2140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5439013464676842e-06,
+      "loss": 0.8087,
+      "step": 2141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5354517030376266e-06,
+      "loss": 0.7865,
+      "step": 2142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.527009981211119e-06,
+      "loss": 0.7743,
+      "step": 2143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5185761913325665e-06,
+      "loss": 0.7346,
+      "step": 2144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5101503437366678e-06,
+      "loss": 0.7141,
+      "step": 2145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5017324487483873e-06,
+      "loss": 0.7759,
+      "step": 2146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49332251668294e-06,
+      "loss": 0.7444,
+      "step": 2147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4849205578457944e-06,
+      "loss": 0.7337,
+      "step": 2148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.47652658253263e-06,
+      "loss": 0.7575,
+      "step": 2149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.468140601029363e-06,
+      "loss": 0.7834,
+      "step": 2150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.459762623612102e-06,
+      "loss": 0.8519,
+      "step": 2151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.7751,
+      "step": 2152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.443030722090991e-06,
+      "loss": 0.801,
+      "step": 2153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4346768184902736e-06,
+      "loss": 0.7322,
+      "step": 2154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4263309599818017e-06,
+      "loss": 0.739,
+      "step": 2155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4179931567925216e-06,
+      "loss": 0.8122,
+      "step": 2156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409663419139506e-06,
+      "loss": 0.7919,
+      "step": 2157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4013417572299446e-06,
+      "loss": 0.6767,
+      "step": 2158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393028181261142e-06,
+      "loss": 0.7648,
+      "step": 2159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3847227014204697e-06,
+      "loss": 0.7071,
+      "step": 2160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.37642532788541e-06,
+      "loss": 0.6924,
+      "step": 2161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.368136070823478e-06,
+      "loss": 0.7317,
+      "step": 2162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.359854940392273e-06,
+      "loss": 0.7774,
+      "step": 2163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3515819467394184e-06,
+      "loss": 0.751,
+      "step": 2164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3433171000025733e-06,
+      "loss": 0.7345,
+      "step": 2165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.33506041030941e-06,
+      "loss": 0.8038,
+      "step": 2166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.326811887777607e-06,
+      "loss": 0.7828,
+      "step": 2167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.318571542514837e-06,
+      "loss": 0.8157,
+      "step": 2168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.310339384618745e-06,
+      "loss": 0.7516,
+      "step": 2169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3021154241769606e-06,
+      "loss": 0.6922,
+      "step": 2170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.29389967126704e-06,
+      "loss": 0.7337,
+      "step": 2171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2856921359565154e-06,
+      "loss": 0.7277,
+      "step": 2172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2774928283028153e-06,
+      "loss": 0.7205,
+      "step": 2173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.269301758353315e-06,
+      "loss": 0.7482,
+      "step": 2174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2611189361452823e-06,
+      "loss": 0.722,
+      "step": 2175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2529443717058693e-06,
+      "loss": 0.7902,
+      "step": 2176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2447780750521307e-06,
+      "loss": 0.7001,
+      "step": 2177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.236620056190972e-06,
+      "loss": 0.7217,
+      "step": 2178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228470325119164e-06,
+      "loss": 0.7345,
+      "step": 2179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.220328891823318e-06,
+      "loss": 0.7072,
+      "step": 2180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2121957662798788e-06,
+      "loss": 0.7519,
+      "step": 2181
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.20407095845511e-06,
+      "loss": 0.6942,
+      "step": 2182
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1959544783050833e-06,
+      "loss": 0.6545,
+      "step": 2183
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.187846335775666e-06,
+      "loss": 0.6904,
+      "step": 2184
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.179746540802506e-06,
+      "loss": 0.7609,
+      "step": 2185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1716551033110334e-06,
+      "loss": 0.7015,
+      "step": 2186
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1635720332164154e-06,
+      "loss": 0.8222,
+      "step": 2187
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.155497340423588e-06,
+      "loss": 0.7033,
+      "step": 2188
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1474310348272084e-06,
+      "loss": 0.7144,
+      "step": 2189
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.139373126311662e-06,
+      "loss": 0.7314,
+      "step": 2190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1313236247510414e-06,
+      "loss": 0.7424,
+      "step": 2191
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1232825400091395e-06,
+      "loss": 0.7452,
+      "step": 2192
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.115249881939434e-06,
+      "loss": 0.7176,
+      "step": 2193
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.107225660385077e-06,
+      "loss": 0.7704,
+      "step": 2194
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.099209885178882e-06,
+      "loss": 0.6783,
+      "step": 2195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091202566143311e-06,
+      "loss": 0.6906,
+      "step": 2196
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0832037130904748e-06,
+      "loss": 0.6824,
+      "step": 2197
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.075213335822089e-06,
+      "loss": 0.7151,
+      "step": 2198
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.067231444129507e-06,
+      "loss": 0.7215,
+      "step": 2199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0592580477936606e-06,
+      "loss": 0.7387,
+      "step": 2200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0512931565850936e-06,
+      "loss": 0.7302,
+      "step": 2201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0433367802639115e-06,
+      "loss": 0.7461,
+      "step": 2202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.035388928579792e-06,
+      "loss": 0.7465,
+      "step": 2203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0274496112719675e-06,
+      "loss": 0.7223,
+      "step": 2204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.019518838069211e-06,
+      "loss": 0.6867,
+      "step": 2205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011596618689825e-06,
+      "loss": 0.7225,
+      "step": 2206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.003682962841629e-06,
+      "loss": 0.7445,
+      "step": 2207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.995777880221958e-06,
+      "loss": 0.7429,
+      "step": 2208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9878813805176252e-06,
+      "loss": 0.8414,
+      "step": 2209
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.979993473404945e-06,
+      "loss": 0.6792,
+      "step": 2210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9721141685496825e-06,
+      "loss": 0.7805,
+      "step": 2211
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9642434756070793e-06,
+      "loss": 0.7609,
+      "step": 2212
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9563814042218185e-06,
+      "loss": 0.7296,
+      "step": 2213
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.948527964028007e-06,
+      "loss": 0.7684,
+      "step": 2214
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.940683164649194e-06,
+      "loss": 0.6919,
+      "step": 2215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.932847015698329e-06,
+      "loss": 0.7239,
+      "step": 2216
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9250195267777624e-06,
+      "loss": 0.7353,
+      "step": 2217
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9172007074792342e-06,
+      "loss": 0.7663,
+      "step": 2218
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9093905673838607e-06,
+      "loss": 0.7002,
+      "step": 2219
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.901589116062119e-06,
+      "loss": 0.8039,
+      "step": 2220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8937963630738517e-06,
+      "loss": 0.7218,
+      "step": 2221
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8860123179682244e-06,
+      "loss": 0.6934,
+      "step": 2222
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.878236990283739e-06,
+      "loss": 0.659,
+      "step": 2223
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.87047038954823e-06,
+      "loss": 0.7523,
+      "step": 2224
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8627125252788104e-06,
+      "loss": 0.7109,
+      "step": 2225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8549634069819155e-06,
+      "loss": 0.6927,
+      "step": 2226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8472230441532365e-06,
+      "loss": 0.7725,
+      "step": 2227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.83949144627776e-06,
+      "loss": 0.7397,
+      "step": 2228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.831768622829719e-06,
+      "loss": 0.768,
+      "step": 2229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8240545832725963e-06,
+      "loss": 0.7487,
+      "step": 2230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.81634933705911e-06,
+      "loss": 0.7756,
+      "step": 2231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8086528936312073e-06,
+      "loss": 0.7918,
+      "step": 2232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8009652624200436e-06,
+      "loss": 0.6792,
+      "step": 2233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.793286452845976e-06,
+      "loss": 0.7122,
+      "step": 2234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.785616474318562e-06,
+      "loss": 0.6942,
+      "step": 2235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7779553362365184e-06,
+      "loss": 0.7068,
+      "step": 2236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7703030479877515e-06,
+      "loss": 0.7091,
+      "step": 2237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7626596189492983e-06,
+      "loss": 0.6332,
+      "step": 2238
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.755025058487364e-06,
+      "loss": 0.6397,
+      "step": 2239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.74739937595727e-06,
+      "loss": 0.7712,
+      "step": 2240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7397825807034663e-06,
+      "loss": 0.694,
+      "step": 2241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7321746820595084e-06,
+      "loss": 0.7328,
+      "step": 2242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7245756893480525e-06,
+      "loss": 0.7178,
+      "step": 2243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7169856118808414e-06,
+      "loss": 0.7258,
+      "step": 2244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.709404458958693e-06,
+      "loss": 0.7697,
+      "step": 2245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701832239871487e-06,
+      "loss": 0.7446,
+      "step": 2246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6942689638981556e-06,
+      "loss": 0.7554,
+      "step": 2247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6867146403066833e-06,
+      "loss": 0.6876,
+      "step": 2248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6791692783540634e-06,
+      "loss": 0.7304,
+      "step": 2249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.671632887286327e-06,
+      "loss": 0.7483,
+      "step": 2250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6641054763385044e-06,
+      "loss": 0.7261,
+      "step": 2251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6565870547346193e-06,
+      "loss": 0.7472,
+      "step": 2252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6490776316876853e-06,
+      "loss": 0.762,
+      "step": 2253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6415772163996845e-06,
+      "loss": 0.7429,
+      "step": 2254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.634085818061565e-06,
+      "loss": 0.7536,
+      "step": 2255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.626603445853222e-06,
+      "loss": 0.7519,
+      "step": 2256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.619130108943494e-06,
+      "loss": 0.7673,
+      "step": 2257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.611665816490142e-06,
+      "loss": 0.7077,
+      "step": 2258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.604210577639855e-06,
+      "loss": 0.7707,
+      "step": 2259
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5967644015282146e-06,
+      "loss": 0.7267,
+      "step": 2260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5893272972797e-06,
+      "loss": 0.7048,
+      "step": 2261
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5818992740076876e-06,
+      "loss": 0.7421,
+      "step": 2262
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5744803408144026e-06,
+      "loss": 0.6808,
+      "step": 2263
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.567070506790955e-06,
+      "loss": 0.7432,
+      "step": 2264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559669781017283e-06,
+      "loss": 0.7106,
+      "step": 2265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5522781725621814e-06,
+      "loss": 0.7499,
+      "step": 2266
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5448956904832643e-06,
+      "loss": 0.7378,
+      "step": 2267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5375223438269635e-06,
+      "loss": 0.7156,
+      "step": 2268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.530158141628515e-06,
+      "loss": 0.7211,
+      "step": 2269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5228030929119517e-06,
+      "loss": 0.7642,
+      "step": 2270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5154572066900895e-06,
+      "loss": 0.743,
+      "step": 2271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.508120491964512e-06,
+      "loss": 0.7702,
+      "step": 2272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5007929577255773e-06,
+      "loss": 0.7588,
+      "step": 2273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4934746129523735e-06,
+      "loss": 0.753,
+      "step": 2274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.486165466612751e-06,
+      "loss": 0.7788,
+      "step": 2275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4788655276632633e-06,
+      "loss": 0.7141,
+      "step": 2276
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.471574805049206e-06,
+      "loss": 0.7603,
+      "step": 2277
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.464293307704566e-06,
+      "loss": 0.754,
+      "step": 2278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4570210445520284e-06,
+      "loss": 0.7698,
+      "step": 2279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.449758024502964e-06,
+      "loss": 0.6991,
+      "step": 2280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4425042564574186e-06,
+      "loss": 0.7423,
+      "step": 2281
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4352597493040965e-06,
+      "loss": 0.7352,
+      "step": 2282
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4280245119203584e-06,
+      "loss": 0.714,
+      "step": 2283
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4207985531722034e-06,
+      "loss": 0.723,
+      "step": 2284
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4135818819142576e-06,
+      "loss": 0.6981,
+      "step": 2285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4063745069897783e-06,
+      "loss": 0.7811,
+      "step": 2286
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3991764372306113e-06,
+      "loss": 0.7278,
+      "step": 2287
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3919876814572197e-06,
+      "loss": 0.6961,
+      "step": 2288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.384808248478643e-06,
+      "loss": 0.7359,
+      "step": 2289
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.377638147092497e-06,
+      "loss": 0.718,
+      "step": 2290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3704773860849683e-06,
+      "loss": 0.7298,
+      "step": 2291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3633259742307846e-06,
+      "loss": 0.7112,
+      "step": 2292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3561839202932344e-06,
+      "loss": 0.7081,
+      "step": 2293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3490512330241287e-06,
+      "loss": 0.7443,
+      "step": 2294
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3419279211638025e-06,
+      "loss": 0.6905,
+      "step": 2295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3348139934411008e-06,
+      "loss": 0.7227,
+      "step": 2296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.327709458573378e-06,
+      "loss": 0.7614,
+      "step": 2297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3206143252664626e-06,
+      "loss": 0.7075,
+      "step": 2298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3135286022146785e-06,
+      "loss": 0.7477,
+      "step": 2299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3064522981008133e-06,
+      "loss": 0.5961,
+      "step": 2300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.299385421596101e-06,
+      "loss": 0.7875,
+      "step": 2301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.292327981360245e-06,
+      "loss": 0.7163,
+      "step": 2302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285279986041363e-06,
+      "loss": 0.7766,
+      "step": 2303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.278241444276017e-06,
+      "loss": 0.7631,
+      "step": 2304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.271212364689176e-06,
+      "loss": 0.7514,
+      "step": 2305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2641927558942135e-06,
+      "loss": 0.7592,
+      "step": 2306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.257182626492901e-06,
+      "loss": 0.6437,
+      "step": 2307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2501819850753925e-06,
+      "loss": 0.7244,
+      "step": 2308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.243190840220214e-06,
+      "loss": 0.7553,
+      "step": 2309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2362092004942583e-06,
+      "loss": 0.6959,
+      "step": 2310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229237074452768e-06,
+      "loss": 0.7888,
+      "step": 2311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.222274470639324e-06,
+      "loss": 0.6848,
+      "step": 2312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2153213975858522e-06,
+      "loss": 0.7004,
+      "step": 2313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2083778638125796e-06,
+      "loss": 0.7265,
+      "step": 2314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2014438778280624e-06,
+      "loss": 0.8043,
+      "step": 2315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1945194481291464e-06,
+      "loss": 0.7453,
+      "step": 2316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1876045832009694e-06,
+      "loss": 0.7073,
+      "step": 2317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18069929151695e-06,
+      "loss": 0.718,
+      "step": 2318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.173803581538775e-06,
+      "loss": 0.6452,
+      "step": 2319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16691746171639e-06,
+      "loss": 0.6899,
+      "step": 2320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1600409404879875e-06,
+      "loss": 0.7281,
+      "step": 2321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1531740262800003e-06,
+      "loss": 0.6975,
+      "step": 2322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1463167275070863e-06,
+      "loss": 0.6844,
+      "step": 2323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1394690525721275e-06,
+      "loss": 0.7188,
+      "step": 2324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1326310098661986e-06,
+      "loss": 0.678,
+      "step": 2325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.125802607768588e-06,
+      "loss": 0.7599,
+      "step": 2326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.11898385464676e-06,
+      "loss": 0.7077,
+      "step": 2327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1121747588563567e-06,
+      "loss": 0.7875,
+      "step": 2328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1053753287411895e-06,
+      "loss": 0.7731,
+      "step": 2329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0985855726332205e-06,
+      "loss": 0.7513,
+      "step": 2330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0918054988525627e-06,
+      "loss": 0.691,
+      "step": 2331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08503511570746e-06,
+      "loss": 0.7456,
+      "step": 2332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.078274431494285e-06,
+      "loss": 0.7921,
+      "step": 2333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0715234544975206e-06,
+      "loss": 0.7106,
+      "step": 2334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064782192989765e-06,
+      "loss": 0.6268,
+      "step": 2335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0580506552316924e-06,
+      "loss": 0.6945,
+      "step": 2336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.051328849472084e-06,
+      "loss": 0.7211,
+      "step": 2337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0446167839477815e-06,
+      "loss": 0.7213,
+      "step": 2338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037914466883688e-06,
+      "loss": 0.712,
+      "step": 2339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0312219064927764e-06,
+      "loss": 0.7355,
+      "step": 2340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0245391109760437e-06,
+      "loss": 0.7393,
+      "step": 2341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.017866088522541e-06,
+      "loss": 0.7209,
+      "step": 2342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0112028473093294e-06,
+      "loss": 0.7426,
+      "step": 2343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0045493955014915e-06,
+      "loss": 0.7598,
+      "step": 2344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9979057412521095e-06,
+      "loss": 0.7647,
+      "step": 2345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9912718927022644e-06,
+      "loss": 0.6744,
+      "step": 2346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.984647857981017e-06,
+      "loss": 0.7835,
+      "step": 2347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9780336452054037e-06,
+      "loss": 0.7388,
+      "step": 2348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9714292624804277e-06,
+      "loss": 0.7446,
+      "step": 2349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.96483471789904e-06,
+      "loss": 0.7126,
+      "step": 2350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9582500195421494e-06,
+      "loss": 0.6991,
+      "step": 2351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951675175478579e-06,
+      "loss": 0.7705,
+      "step": 2352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9451101937650963e-06,
+      "loss": 0.7246,
+      "step": 2353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9385550824463727e-06,
+      "loss": 0.7459,
+      "step": 2354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9320098495549865e-06,
+      "loss": 0.6749,
+      "step": 2355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925474503111412e-06,
+      "loss": 0.7147,
+      "step": 2356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.918949051124007e-06,
+      "loss": 0.7212,
+      "step": 2357
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912433501589007e-06,
+      "loss": 0.7489,
+      "step": 2358
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.905927862490512e-06,
+      "loss": 0.7963,
+      "step": 2359
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8994321418004779e-06,
+      "loss": 0.633,
+      "step": 2360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8929463474787036e-06,
+      "loss": 0.7623,
+      "step": 2361
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8864704874728346e-06,
+      "loss": 0.8141,
+      "step": 2362
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8800045697183267e-06,
+      "loss": 0.7869,
+      "step": 2363
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8735486021384686e-06,
+      "loss": 0.6841,
+      "step": 2364
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8671025926443464e-06,
+      "loss": 0.7142,
+      "step": 2365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8606665491348463e-06,
+      "loss": 0.6952,
+      "step": 2366
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.854240479496643e-06,
+      "loss": 0.6521,
+      "step": 2367
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8478243916041882e-06,
+      "loss": 0.7278,
+      "step": 2368
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8414182933197022e-06,
+      "loss": 0.7456,
+      "step": 2369
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8350221924931655e-06,
+      "loss": 0.6846,
+      "step": 2370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.828636096962304e-06,
+      "loss": 0.7649,
+      "step": 2371
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.822260014552587e-06,
+      "loss": 0.7057,
+      "step": 2372
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8158939530772168e-06,
+      "loss": 0.7162,
+      "step": 2373
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8095379203371044e-06,
+      "loss": 0.6666,
+      "step": 2374
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8031919241208884e-06,
+      "loss": 0.7371,
+      "step": 2375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7968559722048906e-06,
+      "loss": 0.756,
+      "step": 2376
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7905300723531393e-06,
+      "loss": 0.7819,
+      "step": 2377
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7842142323173396e-06,
+      "loss": 0.745,
+      "step": 2378
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7779084598368635e-06,
+      "loss": 0.6568,
+      "step": 2379
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.771612762638758e-06,
+      "loss": 0.7489,
+      "step": 2380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7653271484377154e-06,
+      "loss": 0.7559,
+      "step": 2381
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7590516249360757e-06,
+      "loss": 0.7456,
+      "step": 2382
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7527861998238094e-06,
+      "loss": 0.7182,
+      "step": 2383
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7465308807785253e-06,
+      "loss": 0.7444,
+      "step": 2384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7402856754654297e-06,
+      "loss": 0.7403,
+      "step": 2385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7340505915373495e-06,
+      "loss": 0.755,
+      "step": 2386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7278256366347034e-06,
+      "loss": 0.7227,
+      "step": 2387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7216108183854964e-06,
+      "loss": 0.7925,
+      "step": 2388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7154061444053239e-06,
+      "loss": 0.7733,
+      "step": 2389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7092116222973298e-06,
+      "loss": 0.7234,
+      "step": 2390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703027259652238e-06,
+      "loss": 0.698,
+      "step": 2391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6968530640483126e-06,
+      "loss": 0.6818,
+      "step": 2392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6906890430513623e-06,
+      "loss": 0.788,
+      "step": 2393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6845352042147255e-06,
+      "loss": 0.7643,
+      "step": 2394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6783915550792652e-06,
+      "loss": 0.7263,
+      "step": 2395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6722581031733587e-06,
+      "loss": 0.758,
+      "step": 2396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6661348560128853e-06,
+      "loss": 0.7389,
+      "step": 2397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.660021821101222e-06,
+      "loss": 0.6679,
+      "step": 2398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6539190059292288e-06,
+      "loss": 0.776,
+      "step": 2399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6478264179752502e-06,
+      "loss": 0.7543,
+      "step": 2400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6417440647050853e-06,
+      "loss": 0.7227,
+      "step": 2401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6356719535720057e-06,
+      "loss": 0.7604,
+      "step": 2402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.629610092016719e-06,
+      "loss": 0.6839,
+      "step": 2403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6235584874673848e-06,
+      "loss": 0.7451,
+      "step": 2404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6175171473395878e-06,
+      "loss": 0.7179,
+      "step": 2405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6114860790363351e-06,
+      "loss": 0.7642,
+      "step": 2406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6054652899480472e-06,
+      "loss": 0.6917,
+      "step": 2407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5994547874525501e-06,
+      "loss": 0.687,
+      "step": 2408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5934545789150625e-06,
+      "loss": 0.7679,
+      "step": 2409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.7747,
+      "step": 2410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5814850731119146e-06,
+      "loss": 0.7514,
+      "step": 2411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5755157905135843e-06,
+      "loss": 0.6613,
+      "step": 2412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5695568312079156e-06,
+      "loss": 0.686,
+      "step": 2413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5636082024969579e-06,
+      "loss": 0.7624,
+      "step": 2414
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5576699116701178e-06,
+      "loss": 0.6861,
+      "step": 2415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5517419660041277e-06,
+      "loss": 0.7431,
+      "step": 2416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5458243727630363e-06,
+      "loss": 0.6985,
+      "step": 2417
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5399171391982194e-06,
+      "loss": 0.7769,
+      "step": 2418
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.534020272548349e-06,
+      "loss": 0.7417,
+      "step": 2419
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.528133780039397e-06,
+      "loss": 0.7371,
+      "step": 2420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5222576688846223e-06,
+      "loss": 0.736,
+      "step": 2421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5163919462845622e-06,
+      "loss": 0.7504,
+      "step": 2422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5105366194270244e-06,
+      "loss": 0.7786,
+      "step": 2423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5046916954870761e-06,
+      "loss": 0.733,
+      "step": 2424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4988571816270402e-06,
+      "loss": 0.76,
+      "step": 2425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.493033084996479e-06,
+      "loss": 0.7744,
+      "step": 2426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4872194127321971e-06,
+      "loss": 0.816,
+      "step": 2427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4814161719582132e-06,
+      "loss": 0.7017,
+      "step": 2428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4756233697857747e-06,
+      "loss": 0.7609,
+      "step": 2429
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4698410133133322e-06,
+      "loss": 0.7217,
+      "step": 2430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4640691096265358e-06,
+      "loss": 0.6998,
+      "step": 2431
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4583076657982297e-06,
+      "loss": 0.7123,
+      "step": 2432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.452556688888438e-06,
+      "loss": 0.7313,
+      "step": 2433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4468161859443609e-06,
+      "loss": 0.7337,
+      "step": 2434
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4410861640003605e-06,
+      "loss": 0.667,
+      "step": 2435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4353666300779613e-06,
+      "loss": 0.7691,
+      "step": 2436
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4296575911858268e-06,
+      "loss": 0.7096,
+      "step": 2437
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4239590543197746e-06,
+      "loss": 0.7014,
+      "step": 2438
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4182710264627343e-06,
+      "loss": 0.6584,
+      "step": 2439
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.412593514584777e-06,
+      "loss": 0.6742,
+      "step": 2440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4069265256430697e-06,
+      "loss": 0.7402,
+      "step": 2441
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.401270066581899e-06,
+      "loss": 0.7021,
+      "step": 2442
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3956241443326423e-06,
+      "loss": 0.7272,
+      "step": 2443
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3899887658137634e-06,
+      "loss": 0.6651,
+      "step": 2444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3843639379308104e-06,
+      "loss": 0.674,
+      "step": 2445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.378749667576399e-06,
+      "loss": 0.7257,
+      "step": 2446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.373145961630209e-06,
+      "loss": 0.6906,
+      "step": 2447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3675528269589732e-06,
+      "loss": 0.7174,
+      "step": 2448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3619702704164783e-06,
+      "loss": 0.7105,
+      "step": 2449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3563982988435354e-06,
+      "loss": 0.7531,
+      "step": 2450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3508369190679992e-06,
+      "loss": 0.7346,
+      "step": 2451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3452861379047289e-06,
+      "loss": 0.7361,
+      "step": 2452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.7833,
+      "step": 2453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3342163986095347e-06,
+      "loss": 0.6634,
+      "step": 2454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3286974540423747e-06,
+      "loss": 0.7003,
+      "step": 2455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3231891352170011e-06,
+      "loss": 0.6866,
+      "step": 2456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3176914488832616e-06,
+      "loss": 0.7841,
+      "step": 2457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3122044017779768e-06,
+      "loss": 0.7034,
+      "step": 2458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3067280006249261e-06,
+      "loss": 0.6494,
+      "step": 2459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3012622521348483e-06,
+      "loss": 0.7753,
+      "step": 2460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2958071630054214e-06,
+      "loss": 0.6309,
+      "step": 2461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2903627399212748e-06,
+      "loss": 0.7631,
+      "step": 2462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.284928989553952e-06,
+      "loss": 0.7554,
+      "step": 2463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.279505918561923e-06,
+      "loss": 0.7604,
+      "step": 2464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2740935335905835e-06,
+      "loss": 0.7249,
+      "step": 2465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.268691841272215e-06,
+      "loss": 0.7479,
+      "step": 2466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2633008482260146e-06,
+      "loss": 0.767,
+      "step": 2467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2579205610580514e-06,
+      "loss": 0.7104,
+      "step": 2468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2525509863612917e-06,
+      "loss": 0.7468,
+      "step": 2469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2471921307155655e-06,
+      "loss": 0.7723,
+      "step": 2470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2418440006875676e-06,
+      "loss": 0.7011,
+      "step": 2471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2365066028308548e-06,
+      "loss": 0.7718,
+      "step": 2472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2311799436858275e-06,
+      "loss": 0.7773,
+      "step": 2473
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.225864029779731e-06,
+      "loss": 0.7486,
+      "step": 2474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.220558867626639e-06,
+      "loss": 0.7728,
+      "step": 2475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2152644637274603e-06,
+      "loss": 0.7796,
+      "step": 2476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.209980824569903e-06,
+      "loss": 0.8313,
+      "step": 2477
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2047079566285037e-06,
+      "loss": 0.6406,
+      "step": 2478
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1994458663645836e-06,
+      "loss": 0.7422,
+      "step": 2479
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1941945602262684e-06,
+      "loss": 0.6981,
+      "step": 2480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1889540446484638e-06,
+      "loss": 0.6816,
+      "step": 2481
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1837243260528542e-06,
+      "loss": 0.6769,
+      "step": 2482
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1785054108478933e-06,
+      "loss": 0.7037,
+      "step": 2483
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1732973054287945e-06,
+      "loss": 0.7428,
+      "step": 2484
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.168100016177528e-06,
+      "loss": 0.7237,
+      "step": 2485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1629135494628097e-06,
+      "loss": 0.6883,
+      "step": 2486
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.157737911640091e-06,
+      "loss": 0.6958,
+      "step": 2487
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1525731090515536e-06,
+      "loss": 0.7538,
+      "step": 2488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1474191480261098e-06,
+      "loss": 0.6889,
+      "step": 2489
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1422760348793715e-06,
+      "loss": 0.7371,
+      "step": 2490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.137143775913675e-06,
+      "loss": 0.6737,
+      "step": 2491
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1320223774180428e-06,
+      "loss": 0.7548,
+      "step": 2492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.126911845668196e-06,
+      "loss": 0.762,
+      "step": 2493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1218121869265365e-06,
+      "loss": 0.7117,
+      "step": 2494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.116723407442144e-06,
+      "loss": 0.7031,
+      "step": 2495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1116455134507665e-06,
+      "loss": 0.6746,
+      "step": 2496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1065785111748117e-06,
+      "loss": 0.6616,
+      "step": 2497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1015224068233433e-06,
+      "loss": 0.7384,
+      "step": 2498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.096477206592067e-06,
+      "loss": 0.7083,
+      "step": 2499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0914429166633355e-06,
+      "loss": 0.7536,
+      "step": 2500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0864195432061165e-06,
+      "loss": 0.779,
+      "step": 2501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0814070923760179e-06,
+      "loss": 0.7667,
+      "step": 2502
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.076405570315252e-06,
+      "loss": 0.7331,
+      "step": 2503
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0714149831526398e-06,
+      "loss": 0.7835,
+      "step": 2504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0664353370036107e-06,
+      "loss": 0.7323,
+      "step": 2505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0614666379701732e-06,
+      "loss": 0.7301,
+      "step": 2506
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0565088921409383e-06,
+      "loss": 0.7324,
+      "step": 2507
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.051562105591082e-06,
+      "loss": 0.7047,
+      "step": 2508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.046626284382356e-06,
+      "loss": 0.7236,
+      "step": 2509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.041701434563075e-06,
+      "loss": 0.7062,
+      "step": 2510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0367875621681078e-06,
+      "loss": 0.7131,
+      "step": 2511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0318846732188737e-06,
+      "loss": 0.7483,
+      "step": 2512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0269927737233287e-06,
+      "loss": 0.7318,
+      "step": 2513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0221118696759724e-06,
+      "loss": 0.7224,
+      "step": 2514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.017241967057816e-06,
+      "loss": 0.7361,
+      "step": 2515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0123830718364047e-06,
+      "loss": 0.733,
+      "step": 2516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0075351899657814e-06,
+      "loss": 0.7612,
+      "step": 2517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0026983273865055e-06,
+      "loss": 0.7744,
+      "step": 2518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.978724900256265e-07,
+      "loss": 0.6342,
+      "step": 2519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.930576837966842e-07,
+      "loss": 0.7042,
+      "step": 2520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.882539145997027e-07,
+      "loss": 0.6451,
+      "step": 2521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.834611883211797e-07,
+      "loss": 0.7615,
+      "step": 2522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.78679510834084e-07,
+      "loss": 0.6946,
+      "step": 2523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.739088879978409e-07,
+      "loss": 0.7225,
+      "step": 2524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.69149325658334e-07,
+      "loss": 0.6844,
+      "step": 2525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.644008296478878e-07,
+      "loss": 0.7673,
+      "step": 2526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59663405785277e-07,
+      "loss": 0.7151,
+      "step": 2527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.549370598756913e-07,
+      "loss": 0.6815,
+      "step": 2528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.502217977107642e-07,
+      "loss": 0.7406,
+      "step": 2529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.455176250685338e-07,
+      "loss": 0.733,
+      "step": 2530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.408245477134559e-07,
+      "loss": 0.7086,
+      "step": 2531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.361425713963879e-07,
+      "loss": 0.6835,
+      "step": 2532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.314717018545838e-07,
+      "loss": 0.6997,
+      "step": 2533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.268119448116886e-07,
+      "loss": 0.7771,
+      "step": 2534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.221633059777301e-07,
+      "loss": 0.8005,
+      "step": 2535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.17525791049112e-07,
+      "loss": 0.7756,
+      "step": 2536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.128994057086038e-07,
+      "loss": 0.7067,
+      "step": 2537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.082841556253441e-07,
+      "loss": 0.6746,
+      "step": 2538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.036800464548157e-07,
+      "loss": 0.7822,
+      "step": 2539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.99087083838861e-07,
+      "loss": 0.7362,
+      "step": 2540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.945052734056581e-07,
+      "loss": 0.7725,
+      "step": 2541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.899346207697135e-07,
+      "loss": 0.7117,
+      "step": 2542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.853751315318737e-07,
+      "loss": 0.7927,
+      "step": 2543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.808268112792917e-07,
+      "loss": 0.6636,
+      "step": 2544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.762896655854481e-07,
+      "loss": 0.7542,
+      "step": 2545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.717637000101186e-07,
+      "loss": 0.6962,
+      "step": 2546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.672489200993872e-07,
+      "loss": 0.699,
+      "step": 2547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.627453313856249e-07,
+      "loss": 0.7221,
+      "step": 2548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.582529393874928e-07,
+      "loss": 0.7065,
+      "step": 2549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.537717496099296e-07,
+      "loss": 0.6737,
+      "step": 2550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.493017675441495e-07,
+      "loss": 0.7264,
+      "step": 2551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.448429986676298e-07,
+      "loss": 0.7012,
+      "step": 2552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.403954484441079e-07,
+      "loss": 0.7031,
+      "step": 2553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.359591223235785e-07,
+      "loss": 0.7382,
+      "step": 2554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.315340257422733e-07,
+      "loss": 0.7662,
+      "step": 2555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.271201641226734e-07,
+      "loss": 0.6788,
+      "step": 2556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.227175428734868e-07,
+      "loss": 0.7195,
+      "step": 2557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.18326167389647e-07,
+      "loss": 0.6863,
+      "step": 2558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.139460430523105e-07,
+      "loss": 0.7763,
+      "step": 2559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.095771752288451e-07,
+      "loss": 0.6836,
+      "step": 2560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.052195692728248e-07,
+      "loss": 0.7831,
+      "step": 2561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.00873230524023e-07,
+      "loss": 0.706,
+      "step": 2562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.965381643084069e-07,
+      "loss": 0.7068,
+      "step": 2563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.922143759381296e-07,
+      "loss": 0.6972,
+      "step": 2564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.879018707115294e-07,
+      "loss": 0.7562,
+      "step": 2565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.83600653913108e-07,
+      "loss": 0.7233,
+      "step": 2566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.793107308135461e-07,
+      "loss": 0.7382,
+      "step": 2567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.750321066696776e-07,
+      "loss": 0.6676,
+      "step": 2568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.707647867244927e-07,
+      "loss": 0.6655,
+      "step": 2569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.665087762071299e-07,
+      "loss": 0.7481,
+      "step": 2570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.622640803328685e-07,
+      "loss": 0.6271,
+      "step": 2571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.580307043031232e-07,
+      "loss": 0.7229,
+      "step": 2572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.53808653305439e-07,
+      "loss": 0.7512,
+      "step": 2573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.495979325134806e-07,
+      "loss": 0.7212,
+      "step": 2574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.453985470870284e-07,
+      "loss": 0.68,
+      "step": 2575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.412105021719807e-07,
+      "loss": 0.7565,
+      "step": 2576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.370338029003254e-07,
+      "loss": 0.7545,
+      "step": 2577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.328684543901598e-07,
+      "loss": 0.7314,
+      "step": 2578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.287144617456654e-07,
+      "loss": 0.6825,
+      "step": 2579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.245718300571092e-07,
+      "loss": 0.7507,
+      "step": 2580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.204405644008416e-07,
+      "loss": 0.7319,
+      "step": 2581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.163206698392744e-07,
+      "loss": 0.6887,
+      "step": 2582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.122121514208969e-07,
+      "loss": 0.7219,
+      "step": 2583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.081150141802518e-07,
+      "loss": 0.7188,
+      "step": 2584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.040292631379386e-07,
+      "loss": 0.657,
+      "step": 2585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.999549033005981e-07,
+      "loss": 0.6802,
+      "step": 2586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.958919396609231e-07,
+      "loss": 0.7273,
+      "step": 2587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.918403771976312e-07,
+      "loss": 0.7282,
+      "step": 2588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.878002208754731e-07,
+      "loss": 0.715,
+      "step": 2589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.837714756452241e-07,
+      "loss": 0.7907,
+      "step": 2590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.79754146443673e-07,
+      "loss": 0.6744,
+      "step": 2591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.757482381936265e-07,
+      "loss": 0.6897,
+      "step": 2592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.717537558038845e-07,
+      "loss": 0.749,
+      "step": 2593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.677707041692571e-07,
+      "loss": 0.7096,
+      "step": 2594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.637990881705403e-07,
+      "loss": 0.7758,
+      "step": 2595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.598389126745209e-07,
+      "loss": 0.7749,
+      "step": 2596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.558901825339625e-07,
+      "loss": 0.7417,
+      "step": 2597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.519529025876081e-07,
+      "loss": 0.7075,
+      "step": 2598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.480270776601682e-07,
+      "loss": 0.6995,
+      "step": 2599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.441127125623137e-07,
+      "loss": 0.6938,
+      "step": 2600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.402098120906763e-07,
+      "loss": 0.7622,
+      "step": 2601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.36318381027835e-07,
+      "loss": 0.7109,
+      "step": 2602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.324384241423232e-07,
+      "loss": 0.7063,
+      "step": 2603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.285699461886008e-07,
+      "loss": 0.7498,
+      "step": 2604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.247129519070728e-07,
+      "loss": 0.6581,
+      "step": 2605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.20867446024066e-07,
+      "loss": 0.7431,
+      "step": 2606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.170334332518325e-07,
+      "loss": 0.7119,
+      "step": 2607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.132109182885382e-07,
+      "loss": 0.7169,
+      "step": 2608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.09399905818262e-07,
+      "loss": 0.7697,
+      "step": 2609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.056004005109861e-07,
+      "loss": 0.6225,
+      "step": 2610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.018124070225928e-07,
+      "loss": 0.7622,
+      "step": 2611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.980359299948568e-07,
+      "loss": 0.7039,
+      "step": 2612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.942709740554419e-07,
+      "loss": 0.7048,
+      "step": 2613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.905175438178979e-07,
+      "loss": 0.7224,
+      "step": 2614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.867756438816396e-07,
+      "loss": 0.6642,
+      "step": 2615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.830452788319662e-07,
+      "loss": 0.659,
+      "step": 2616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.793264532400311e-07,
+      "loss": 0.6532,
+      "step": 2617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.756191716628556e-07,
+      "loss": 0.7466,
+      "step": 2618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.719234386433126e-07,
+      "loss": 0.6755,
+      "step": 2619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.68239258710116e-07,
+      "loss": 0.6892,
+      "step": 2620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.645666363778357e-07,
+      "loss": 0.7093,
+      "step": 2621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.609055761468706e-07,
+      "loss": 0.7495,
+      "step": 2622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.572560825034523e-07,
+      "loss": 0.7268,
+      "step": 2623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.536181599196388e-07,
+      "loss": 0.7466,
+      "step": 2624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.499918128533155e-07,
+      "loss": 0.6938,
+      "step": 2625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.463770457481732e-07,
+      "loss": 0.7292,
+      "step": 2626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.427738630337176e-07,
+      "loss": 0.7725,
+      "step": 2627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.39182269125259e-07,
+      "loss": 0.6861,
+      "step": 2628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.35602268423906e-07,
+      "loss": 0.7289,
+      "step": 2629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.320338653165657e-07,
+      "loss": 0.6626,
+      "step": 2630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28477064175924e-07,
+      "loss": 0.7371,
+      "step": 2631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.249318693604577e-07,
+      "loss": 0.7422,
+      "step": 2632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.213982852144206e-07,
+      "loss": 0.7062,
+      "step": 2633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.178763160678346e-07,
+      "loss": 0.6897,
+      "step": 2634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.143659662364931e-07,
+      "loss": 0.7143,
+      "step": 2635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.108672400219483e-07,
+      "loss": 0.702,
+      "step": 2636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.073801417115109e-07,
+      "loss": 0.6899,
+      "step": 2637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.039046755782417e-07,
+      "loss": 0.7369,
+      "step": 2638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.004408458809473e-07,
+      "loss": 0.7311,
+      "step": 2639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.969886568641757e-07,
+      "loss": 0.7182,
+      "step": 2640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.935481127582131e-07,
+      "loss": 0.7116,
+      "step": 2641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.901192177790693e-07,
+      "loss": 0.6719,
+      "step": 2642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.867019761284875e-07,
+      "loss": 0.6721,
+      "step": 2643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.83296391993926e-07,
+      "loss": 0.6945,
+      "step": 2644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.799024695485577e-07,
+      "loss": 0.6734,
+      "step": 2645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7652021295126804e-07,
+      "loss": 0.7567,
+      "step": 2646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7314962634664616e-07,
+      "loss": 0.6996,
+      "step": 2647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.697907138649793e-07,
+      "loss": 0.7553,
+      "step": 2648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.664434796222528e-07,
+      "loss": 0.7007,
+      "step": 2649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.631079277201389e-07,
+      "loss": 0.6567,
+      "step": 2650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.597840622459937e-07,
+      "loss": 0.7505,
+      "step": 2651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5647188727285687e-07,
+      "loss": 0.7302,
+      "step": 2652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5317140685943726e-07,
+      "loss": 0.6874,
+      "step": 2653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.498826250501187e-07,
+      "loss": 0.7323,
+      "step": 2654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.466055458749441e-07,
+      "loss": 0.712,
+      "step": 2655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.433401733496201e-07,
+      "loss": 0.6931,
+      "step": 2656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4008651147550816e-07,
+      "loss": 0.707,
+      "step": 2657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3684456423961797e-07,
+      "loss": 0.6845,
+      "step": 2658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3361433561460274e-07,
+      "loss": 0.7127,
+      "step": 2659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.303958295587596e-07,
+      "loss": 0.7104,
+      "step": 2660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.271890500160192e-07,
+      "loss": 0.7212,
+      "step": 2661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2399400091594154e-07,
+      "loss": 0.7328,
+      "step": 2662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2081068617371266e-07,
+      "loss": 0.6754,
+      "step": 2663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1763910969013886e-07,
+      "loss": 0.687,
+      "step": 2664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.14479275351648e-07,
+      "loss": 0.7002,
+      "step": 2665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.113311870302716e-07,
+      "loss": 0.7723,
+      "step": 2666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0819484858364844e-07,
+      "loss": 0.6724,
+      "step": 2667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.7132,
+      "step": 2668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.019574366732426e-07,
+      "loss": 0.7742,
+      "step": 2669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9885637085273044e-07,
+      "loss": 0.69,
+      "step": 2670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9576707019350903e-07,
+      "loss": 0.7239,
+      "step": 2671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.926895384811835e-07,
+      "loss": 0.7406,
+      "step": 2672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8962377948693395e-07,
+      "loss": 0.7888,
+      "step": 2673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.865697969675164e-07,
+      "loss": 0.7542,
+      "step": 2674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.83527594665255e-07,
+      "loss": 0.7088,
+      "step": 2675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.804971763080401e-07,
+      "loss": 0.716,
+      "step": 2676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7747854560931996e-07,
+      "loss": 0.7207,
+      "step": 2677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7447170626809915e-07,
+      "loss": 0.689,
+      "step": 2678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.714766619689369e-07,
+      "loss": 0.6904,
+      "step": 2679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.684934163819309e-07,
+      "loss": 0.7173,
+      "step": 2680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.655219731627302e-07,
+      "loss": 0.6797,
+      "step": 2681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6256233595250987e-07,
+      "loss": 0.6768,
+      "step": 2682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.596145083779912e-07,
+      "loss": 0.7227,
+      "step": 2683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.566784940514145e-07,
+      "loss": 0.6876,
+      "step": 2684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.537542965705476e-07,
+      "loss": 0.6966,
+      "step": 2685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.508419195186774e-07,
+      "loss": 0.7543,
+      "step": 2686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4794136646460695e-07,
+      "loss": 0.7136,
+      "step": 2687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.450526409626498e-07,
+      "loss": 0.7446,
+      "step": 2688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.421757465526243e-07,
+      "loss": 0.7471,
+      "step": 2689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3931068675985726e-07,
+      "loss": 0.7075,
+      "step": 2690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.364574650951624e-07,
+      "loss": 0.7402,
+      "step": 2691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.33616085054862e-07,
+      "loss": 0.741,
+      "step": 2692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.307865501207508e-07,
+      "loss": 0.6618,
+      "step": 2693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2796886376012413e-07,
+      "loss": 0.6881,
+      "step": 2694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2516302942574794e-07,
+      "loss": 0.6615,
+      "step": 2695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2236905055586854e-07,
+      "loss": 0.6575,
+      "step": 2696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1958693057420496e-07,
+      "loss": 0.7083,
+      "step": 2697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1681667288994353e-07,
+      "loss": 0.639,
+      "step": 2698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.140582808977355e-07,
+      "loss": 0.7493,
+      "step": 2699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.113117579776914e-07,
+      "loss": 0.739,
+      "step": 2700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0857710749537585e-07,
+      "loss": 0.717,
+      "step": 2701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.058543328018071e-07,
+      "loss": 0.7202,
+      "step": 2702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0314343723345295e-07,
+      "loss": 0.7123,
+      "step": 2703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0044442411222066e-07,
+      "loss": 0.7202,
+      "step": 2704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9775729674545673e-07,
+      "loss": 0.7077,
+      "step": 2705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9508205842594727e-07,
+      "loss": 0.6441,
+      "step": 2706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9241871243190555e-07,
+      "loss": 0.7043,
+      "step": 2707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8976726202697313e-07,
+      "loss": 0.7474,
+      "step": 2708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.871277104602177e-07,
+      "loss": 0.6816,
+      "step": 2709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.845000609661208e-07,
+      "loss": 0.7332,
+      "step": 2710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.818843167645835e-07,
+      "loss": 0.7394,
+      "step": 2711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7928048106091733e-07,
+      "loss": 0.7399,
+      "step": 2712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7668855704583997e-07,
+      "loss": 0.7338,
+      "step": 2713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7410854789547305e-07,
+      "loss": 0.6756,
+      "step": 2714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.715404567713398e-07,
+      "loss": 0.771,
+      "step": 2715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.689842868203563e-07,
+      "loss": 0.7164,
+      "step": 2716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6644004117483357e-07,
+      "loss": 0.7257,
+      "step": 2717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.639077229524656e-07,
+      "loss": 0.7402,
+      "step": 2718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6138733525633896e-07,
+      "loss": 0.651,
+      "step": 2719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.58878881174911e-07,
+      "loss": 0.7645,
+      "step": 2720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.56382363782024e-07,
+      "loss": 0.7082,
+      "step": 2721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5389778613688744e-07,
+      "loss": 0.7657,
+      "step": 2722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5142515128408263e-07,
+      "loss": 0.7727,
+      "step": 2723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4896446225355477e-07,
+      "loss": 0.6529,
+      "step": 2724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.46515722060614e-07,
+      "loss": 0.749,
+      "step": 2725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.440789337059235e-07,
+      "loss": 0.698,
+      "step": 2726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4165410017550575e-07,
+      "loss": 0.773,
+      "step": 2727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.392412244407294e-07,
+      "loss": 0.712,
+      "step": 2728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.368403094583116e-07,
+      "loss": 0.7278,
+      "step": 2729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3445135817031783e-07,
+      "loss": 0.644,
+      "step": 2730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3207437350414418e-07,
+      "loss": 0.7642,
+      "step": 2731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2970935837253182e-07,
+      "loss": 0.7307,
+      "step": 2732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.273563156735481e-07,
+      "loss": 0.6878,
+      "step": 2733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2501524829059208e-07,
+      "loss": 0.64,
+      "step": 2734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.22686159092389e-07,
+      "loss": 0.7563,
+      "step": 2735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2036905093298477e-07,
+      "loss": 0.7188,
+      "step": 2736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.180639266517448e-07,
+      "loss": 0.6556,
+      "step": 2737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.157707890733496e-07,
+      "loss": 0.7588,
+      "step": 2738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1348964100778914e-07,
+      "loss": 0.7205,
+      "step": 2739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1122048525036409e-07,
+      "loss": 0.7009,
+      "step": 2740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0896332458168024e-07,
+      "loss": 0.6714,
+      "step": 2741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.067181617676406e-07,
+      "loss": 0.7393,
+      "step": 2742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0448499955945223e-07,
+      "loss": 0.7064,
+      "step": 2743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0226384069361283e-07,
+      "loss": 0.6755,
+      "step": 2744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0005468789190852e-07,
+      "loss": 0.7165,
+      "step": 2745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9785754386142164e-07,
+      "loss": 0.7498,
+      "step": 2746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9567241129450964e-07,
+      "loss": 0.7172,
+      "step": 2747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.934992928688195e-07,
+      "loss": 0.6713,
+      "step": 2748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9133819124727003e-07,
+      "loss": 0.699,
+      "step": 2749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8918910907805733e-07,
+      "loss": 0.7011,
+      "step": 2750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8705204899465034e-07,
+      "loss": 0.7181,
+      "step": 2751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8492701361578326e-07,
+      "loss": 0.6477,
+      "step": 2752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8281400554545637e-07,
+      "loss": 0.7038,
+      "step": 2753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8071302737293294e-07,
+      "loss": 0.712,
+      "step": 2754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7862408167273472e-07,
+      "loss": 0.7057,
+      "step": 2755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7654717100463625e-07,
+      "loss": 0.7025,
+      "step": 2756
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7448229791366956e-07,
+      "loss": 0.6973,
+      "step": 2757
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.724294649301095e-07,
+      "loss": 0.6945,
+      "step": 2758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7038867456948382e-07,
+      "loss": 0.7369,
+      "step": 2759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6835992933255775e-07,
+      "loss": 0.7402,
+      "step": 2760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6634323170533928e-07,
+      "loss": 0.7158,
+      "step": 2761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.643385841590728e-07,
+      "loss": 0.7335,
+      "step": 2762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6234598915023657e-07,
+      "loss": 0.7984,
+      "step": 2763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6036544912054087e-07,
+      "loss": 0.7188,
+      "step": 2764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5839696649692204e-07,
+      "loss": 0.7142,
+      "step": 2765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5644054369154282e-07,
+      "loss": 0.7488,
+      "step": 2766
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.544961831017855e-07,
+      "loss": 0.6855,
+      "step": 2767
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5256388711025638e-07,
+      "loss": 0.7306,
+      "step": 2768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5064365808477255e-07,
+      "loss": 0.7335,
+      "step": 2769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.487354983783673e-07,
+      "loss": 0.7541,
+      "step": 2770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4683941032928362e-07,
+      "loss": 0.7269,
+      "step": 2771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4495539626097289e-07,
+      "loss": 0.661,
+      "step": 2772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.430834584820895e-07,
+      "loss": 0.7202,
+      "step": 2773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4122359928648965e-07,
+      "loss": 0.7417,
+      "step": 2774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3937582095322923e-07,
+      "loss": 0.7264,
+      "step": 2775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.375401257465625e-07,
+      "loss": 0.7639,
+      "step": 2776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.357165159159324e-07,
+      "loss": 0.7034,
+      "step": 2777
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3390499369597686e-07,
+      "loss": 0.7199,
+      "step": 2778
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3210556130652031e-07,
+      "loss": 0.6905,
+      "step": 2779
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3031822095257108e-07,
+      "loss": 0.6657,
+      "step": 2780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2854297482432166e-07,
+      "loss": 0.6753,
+      "step": 2781
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2677982509714415e-07,
+      "loss": 0.6995,
+      "step": 2782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2502877393158587e-07,
+      "loss": 0.7508,
+      "step": 2783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.232898234733726e-07,
+      "loss": 0.6968,
+      "step": 2784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2156297585339872e-07,
+      "loss": 0.6487,
+      "step": 2785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.198482331877282e-07,
+      "loss": 0.7349,
+      "step": 2786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1814559757759358e-07,
+      "loss": 0.7576,
+      "step": 2787
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1645507110938925e-07,
+      "loss": 0.7516,
+      "step": 2788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1477665585467256e-07,
+      "loss": 0.7279,
+      "step": 2789
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1311035387016167e-07,
+      "loss": 0.7388,
+      "step": 2790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1145616719772545e-07,
+      "loss": 0.7759,
+      "step": 2791
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0981409786439356e-07,
+      "loss": 0.7531,
+      "step": 2792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.08184147882342e-07,
+      "loss": 0.697,
+      "step": 2793
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0656631924889749e-07,
+      "loss": 0.7171,
+      "step": 2794
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0496061394653756e-07,
+      "loss": 0.7105,
+      "step": 2795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0336703394287495e-07,
+      "loss": 0.666,
+      "step": 2796
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0178558119067316e-07,
+      "loss": 0.7079,
+      "step": 2797
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0021625762782872e-07,
+      "loss": 0.6933,
+      "step": 2798
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.865906517738e-08,
+      "loss": 0.7113,
+      "step": 2799
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.711400574749507e-08,
+      "loss": 0.6376,
+      "step": 2800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.558108123147947e-08,
+      "loss": 0.7325,
+      "step": 2801
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.40602935077639e-08,
+      "loss": 0.7433,
+      "step": 2802
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.255164443990994e-08,
+      "loss": 0.7283,
+      "step": 2803
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.10551358766032e-08,
+      "loss": 0.6573,
+      "step": 2804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.957076965165234e-08,
+      "loss": 0.7551,
+      "step": 2805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.809854758399017e-08,
+      "loss": 0.6597,
+      "step": 2806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.663847147766136e-08,
+      "loss": 0.7607,
+      "step": 2807
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.519054312183361e-08,
+      "loss": 0.7009,
+      "step": 2808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.375476429078543e-08,
+      "loss": 0.7109,
+      "step": 2809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.233113674390836e-08,
+      "loss": 0.6901,
+      "step": 2810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.091966222570358e-08,
+      "loss": 0.7702,
+      "step": 2811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.952034246577977e-08,
+      "loss": 0.6876,
+      "step": 2812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.81331791788531e-08,
+      "loss": 0.7316,
+      "step": 2813
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.67581740647394e-08,
+      "loss": 0.6676,
+      "step": 2814
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.539532880836087e-08,
+      "loss": 0.7361,
+      "step": 2815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.404464507973608e-08,
+      "loss": 0.6917,
+      "step": 2816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.270612453398107e-08,
+      "loss": 0.7091,
+      "step": 2817
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.137976881130826e-08,
+      "loss": 0.6832,
+      "step": 2818
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.006557953702198e-08,
+      "loss": 0.7619,
+      "step": 2819
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.87635583215196e-08,
+      "loss": 0.6969,
+      "step": 2820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.747370676028819e-08,
+      "loss": 0.6808,
+      "step": 2821
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.6196026433899e-08,
+      "loss": 0.767,
+      "step": 2822
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.493051890801072e-08,
+      "loss": 0.6622,
+      "step": 2823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.367718573336845e-08,
+      "loss": 0.7744,
+      "step": 2824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.243602844579255e-08,
+      "loss": 0.7358,
+      "step": 2825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.120704856618975e-08,
+      "loss": 0.7729,
+      "step": 2826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.999024760054095e-08,
+      "loss": 0.7261,
+      "step": 2827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.878562703990342e-08,
+      "loss": 0.7048,
+      "step": 2828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7593188360408613e-08,
+      "loss": 0.7074,
+      "step": 2829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.641293302326323e-08,
+      "loss": 0.6587,
+      "step": 2830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.524486247473926e-08,
+      "loss": 0.7555,
+      "step": 2831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.408897814618175e-08,
+      "loss": 0.6951,
+      "step": 2832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2945281454003236e-08,
+      "loss": 0.7145,
+      "step": 2833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.181377379967823e-08,
+      "loss": 0.7353,
+      "step": 2834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0694456569747583e-08,
+      "loss": 0.7366,
+      "step": 2835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.958733113581415e-08,
+      "loss": 0.7047,
+      "step": 2836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.849239885453827e-08,
+      "loss": 0.7297,
+      "step": 2837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.740966106764222e-08,
+      "loss": 0.7196,
+      "step": 2838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6339119101902475e-08,
+      "loss": 0.7485,
+      "step": 2839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.528077426915412e-08,
+      "loss": 0.7004,
+      "step": 2840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.423462786628308e-08,
+      "loss": 0.6532,
+      "step": 2841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.320068117522835e-08,
+      "loss": 0.7201,
+      "step": 2842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2178935462978686e-08,
+      "loss": 0.751,
+      "step": 2843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1169391981574766e-08,
+      "loss": 0.6991,
+      "step": 2844
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0172051968101474e-08,
+      "loss": 0.7154,
+      "step": 2845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.918691664469121e-08,
+      "loss": 0.7389,
+      "step": 2846
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8213987218520544e-08,
+      "loss": 0.7562,
+      "step": 2847
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7253264881809137e-08,
+      "loss": 0.7293,
+      "step": 2848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.630475081181861e-08,
+      "loss": 0.672,
+      "step": 2849
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5368446170849225e-08,
+      "loss": 0.7363,
+      "step": 2850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4444352106242086e-08,
+      "loss": 0.7231,
+      "step": 2851
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.35324697503725e-08,
+      "loss": 0.7224,
+      "step": 2852
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2632800220654406e-08,
+      "loss": 0.7215,
+      "step": 2853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.174534461953593e-08,
+      "loss": 0.642,
+      "step": 2854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0870104034498306e-08,
+      "loss": 0.7454,
+      "step": 2855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.000707953805249e-08,
+      "loss": 0.7075,
+      "step": 2856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.915627218774142e-08,
+      "loss": 0.734,
+      "step": 2857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8317683026138908e-08,
+      "loss": 0.7625,
+      "step": 2858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7491313080844072e-08,
+      "loss": 0.712,
+      "step": 2859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.667716336448356e-08,
+      "loss": 0.7158,
+      "step": 2860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5875234874709332e-08,
+      "loss": 0.6639,
+      "step": 2861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5085528594199772e-08,
+      "loss": 0.7383,
+      "step": 2862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.430804549065302e-08,
+      "loss": 0.7639,
+      "step": 2863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3542786516789206e-08,
+      "loss": 0.6637,
+      "step": 2864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2789752610352657e-08,
+      "loss": 0.7652,
+      "step": 2865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2048944694104123e-08,
+      "loss": 0.7321,
+      "step": 2866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1320363675821908e-08,
+      "loss": 0.7856,
+      "step": 2867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0604010448305178e-08,
+      "loss": 0.7212,
+      "step": 2868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.989988588936509e-08,
+      "loss": 0.7916,
+      "step": 2869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.920799086183256e-08,
+      "loss": 0.7333,
+      "step": 2870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8528326213548276e-08,
+      "loss": 0.7753,
+      "step": 2871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7860892777367133e-08,
+      "loss": 0.7341,
+      "step": 2872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.720569137115824e-08,
+      "loss": 0.7597,
+      "step": 2873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.656272279779825e-08,
+      "loss": 0.7166,
+      "step": 2874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5931987845176912e-08,
+      "loss": 0.633,
+      "step": 2875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.531348728619042e-08,
+      "loss": 0.736,
+      "step": 2876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4707221878745847e-08,
+      "loss": 0.7366,
+      "step": 2877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.411319236575337e-08,
+      "loss": 0.7016,
+      "step": 2878
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3531399475132933e-08,
+      "loss": 0.7466,
+      "step": 2879
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2961843919809813e-08,
+      "loss": 0.7126,
+      "step": 2880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2404526397711281e-08,
+      "loss": 0.7612,
+      "step": 2881
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1859447591769934e-08,
+      "loss": 0.8206,
+      "step": 2882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1326608169920373e-08,
+      "loss": 0.6939,
+      "step": 2883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0806008785100297e-08,
+      "loss": 0.7454,
+      "step": 2884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.029765007524719e-08,
+      "loss": 0.7039,
+      "step": 2885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801532663300528e-09,
+      "loss": 0.762,
+      "step": 2886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.317657157197347e-09,
+      "loss": 0.6866,
+      "step": 2887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.846024149875564e-09,
+      "loss": 0.6782,
+      "step": 2888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.386634219269551e-09,
+      "loss": 0.7231,
+      "step": 2889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.93948792831234e-09,
+      "loss": 0.6985,
+      "step": 2890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.504585824933408e-09,
+      "loss": 0.7172,
+      "step": 2891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.081928442057573e-09,
+      "loss": 0.7107,
+      "step": 2892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.671516297606095e-09,
+      "loss": 0.7072,
+      "step": 2893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.273349894494463e-09,
+      "loss": 0.7481,
+      "step": 2894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.887429720633497e-09,
+      "loss": 0.7247,
+      "step": 2895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.513756248924917e-09,
+      "loss": 0.7991,
+      "step": 2896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.152329937265777e-09,
+      "loss": 0.6802,
+      "step": 2897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.803151228544023e-09,
+      "loss": 0.727,
+      "step": 2898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.466220550641831e-09,
+      "loss": 0.6831,
+      "step": 2899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1415383164311596e-09,
+      "loss": 0.7435,
+      "step": 2900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.829104923773752e-09,
+      "loss": 0.6586,
+      "step": 2901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.528920755523357e-09,
+      "loss": 0.7425,
+      "step": 2902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2409861795235087e-09,
+      "loss": 0.7475,
+      "step": 2903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9653015486064143e-09,
+      "loss": 0.7684,
+      "step": 2904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.701867200592956e-09,
+      "loss": 0.7532,
+      "step": 2905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4506834582960213e-09,
+      "loss": 0.6582,
+      "step": 2906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.211750629511622e-09,
+      "loss": 0.6682,
+      "step": 2907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9850690070266633e-09,
+      "loss": 0.6454,
+      "step": 2908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7706388686145049e-09,
+      "loss": 0.7645,
+      "step": 2909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5684604770371814e-09,
+      "loss": 0.7268,
+      "step": 2910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.378534080042071e-09,
+      "loss": 0.7737,
+      "step": 2911
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2008599103618956e-09,
+      "loss": 0.6463,
+      "step": 2912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.035438185719162e-09,
+      "loss": 0.6644,
+      "step": 2913
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.822691088195001e-10,
+      "loss": 0.6984,
+      "step": 2914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.413528673549941e-10,
+      "loss": 0.7002,
+      "step": 2915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.126896340030719e-10,
+      "loss": 0.6308,
+      "step": 2916
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.962795664265052e-10,
+      "loss": 0.7575,
+      "step": 2917
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.921228072745198e-10,
+      "loss": 0.6975,
+      "step": 2918
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0021948417724434e-10,
+      "loss": 0.7345,
+      "step": 2919
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2056970975459223e-10,
+      "loss": 0.727,
+      "step": 2920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5317358160849005e-10,
+      "loss": 0.6888,
+      "step": 2921
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.803118232398767e-11,
+      "loss": 0.7429,
+      "step": 2922
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.514257947369928e-11,
+      "loss": 0.7098,
+      "step": 2923
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.450782561225218e-11,
+      "loss": 0.752,
+      "step": 2924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.126958279617512e-12,
+      "loss": 0.7236,
+      "step": 2925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.7174,
+      "step": 2926
+    },
+    {
+      "epoch": 1.0,
+      "step": 2926,
+      "total_flos": 3323334993575936.0,
+      "train_loss": 0.8017947865282112,
+      "train_runtime": 54932.954,
+      "train_samples_per_second": 10.227,
+      "train_steps_per_second": 0.053
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2926,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 3323334993575936.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}