diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,36894 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 6144,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-07,
+      "loss": 0.7712,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1621621621621625e-07,
+      "loss": 0.8513,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2432432432432436e-07,
+      "loss": 0.9517,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.324324324324325e-07,
+      "loss": 0.7664,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 0.7875,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.486486486486487e-07,
+      "loss": 0.8069,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.567567567567569e-07,
+      "loss": 0.7127,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.64864864864865e-07,
+      "loss": 0.7129,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.72972972972973e-07,
+      "loss": 0.6557,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 0.7586,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1891891891891893e-06,
+      "loss": 0.729,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2972972972972974e-06,
+      "loss": 0.8003,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4054054054054056e-06,
+      "loss": 0.7228,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5135135135135137e-06,
+      "loss": 0.6714,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 0.6628,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.72972972972973e-06,
+      "loss": 0.6665,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8378378378378381e-06,
+      "loss": 0.6613,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.945945945945946e-06,
+      "loss": 0.6408,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.054054054054054e-06,
+      "loss": 0.6591,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 0.6391,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2702702702702705e-06,
+      "loss": 0.6929,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3783783783783786e-06,
+      "loss": 0.7011,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4864864864864867e-06,
+      "loss": 0.7434,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.594594594594595e-06,
+      "loss": 0.605,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 0.7307,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.810810810810811e-06,
+      "loss": 0.6887,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9189189189189193e-06,
+      "loss": 0.725,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0270270270270274e-06,
+      "loss": 0.6561,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1351351351351356e-06,
+      "loss": 0.6513,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 0.6597,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.351351351351352e-06,
+      "loss": 0.6498,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.45945945945946e-06,
+      "loss": 0.6341,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.567567567567568e-06,
+      "loss": 0.6772,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6756756756756763e-06,
+      "loss": 0.5817,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 0.6184,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.891891891891892e-06,
+      "loss": 0.6391,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.6556,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.108108108108108e-06,
+      "loss": 0.5859,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.216216216216217e-06,
+      "loss": 0.6649,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 0.6803,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.432432432432433e-06,
+      "loss": 0.6888,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.540540540540541e-06,
+      "loss": 0.6834,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6486486486486495e-06,
+      "loss": 0.621,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.756756756756757e-06,
+      "loss": 0.6763,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 0.7321,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9729729729729735e-06,
+      "loss": 0.6755,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.081081081081082e-06,
+      "loss": 0.674,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.18918918918919e-06,
+      "loss": 0.6304,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.297297297297298e-06,
+      "loss": 0.6529,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 0.6142,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.513513513513515e-06,
+      "loss": 0.73,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.621621621621622e-06,
+      "loss": 0.651,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.729729729729731e-06,
+      "loss": 0.6138,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.837837837837839e-06,
+      "loss": 0.6541,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 0.6665,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.054054054054055e-06,
+      "loss": 0.5945,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.162162162162163e-06,
+      "loss": 0.6753,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.270270270270271e-06,
+      "loss": 0.6275,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.378378378378379e-06,
+      "loss": 0.6458,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 0.6096,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.594594594594595e-06,
+      "loss": 0.6131,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.702702702702704e-06,
+      "loss": 0.6235,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.810810810810811e-06,
+      "loss": 0.5928,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.91891891891892e-06,
+      "loss": 0.6265,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 0.5966,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.135135135135136e-06,
+      "loss": 0.5535,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.243243243243244e-06,
+      "loss": 0.7179,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.3513513513513525e-06,
+      "loss": 0.539,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.45945945945946e-06,
+      "loss": 0.6449,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 0.6005,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.675675675675676e-06,
+      "loss": 0.649,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.783783783783784e-06,
+      "loss": 0.6597,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.891891891891894e-06,
+      "loss": 0.6108,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6489,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 0.6562,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.216216216216217e-06,
+      "loss": 0.5913,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.324324324324326e-06,
+      "loss": 0.6117,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.432432432432434e-06,
+      "loss": 0.6198,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.540540540540542e-06,
+      "loss": 0.6186,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 0.6637,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.756756756756759e-06,
+      "loss": 0.5974,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.864864864864866e-06,
+      "loss": 0.6383,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.972972972972974e-06,
+      "loss": 0.5569,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.081081081081082e-06,
+      "loss": 0.625,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 0.6177,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.297297297297299e-06,
+      "loss": 0.6009,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.405405405405407e-06,
+      "loss": 0.6191,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.513513513513514e-06,
+      "loss": 0.5543,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.621621621621622e-06,
+      "loss": 0.5993,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 0.603,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.83783783783784e-06,
+      "loss": 0.5968,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.945945945945947e-06,
+      "loss": 0.5936,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0054054054054055e-05,
+      "loss": 0.6724,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0162162162162164e-05,
+      "loss": 0.6522,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 0.5563,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.037837837837838e-05,
+      "loss": 0.5983,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0486486486486487e-05,
+      "loss": 0.6375,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0594594594594597e-05,
+      "loss": 0.6621,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0702702702702703e-05,
+      "loss": 0.6481,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 0.5901,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.091891891891892e-05,
+      "loss": 0.6087,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.102702702702703e-05,
+      "loss": 0.5886,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1135135135135135e-05,
+      "loss": 0.5765,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1243243243243245e-05,
+      "loss": 0.6473,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 0.6125,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1459459459459462e-05,
+      "loss": 0.5808,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1567567567567568e-05,
+      "loss": 0.605,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1675675675675677e-05,
+      "loss": 0.618,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1783783783783785e-05,
+      "loss": 0.6574,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 0.6391,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6557,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.210810810810811e-05,
+      "loss": 0.5789,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2216216216216217e-05,
+      "loss": 0.593,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2324324324324327e-05,
+      "loss": 0.5919,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 0.6212,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2540540540540542e-05,
+      "loss": 0.6514,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.264864864864865e-05,
+      "loss": 0.6545,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2756756756756758e-05,
+      "loss": 0.5513,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2864864864864865e-05,
+      "loss": 0.6612,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 0.6045,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3081081081081083e-05,
+      "loss": 0.6516,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.318918918918919e-05,
+      "loss": 0.641,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3297297297297298e-05,
+      "loss": 0.5456,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3405405405405407e-05,
+      "loss": 0.6303,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 0.6441,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3621621621621623e-05,
+      "loss": 0.6654,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.372972972972973e-05,
+      "loss": 0.557,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.383783783783784e-05,
+      "loss": 0.6175,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3945945945945946e-05,
+      "loss": 0.5824,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 0.6618,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4162162162162163e-05,
+      "loss": 0.6337,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4270270270270272e-05,
+      "loss": 0.5948,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4378378378378378e-05,
+      "loss": 0.6772,
+      "step": 133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4486486486486488e-05,
+      "loss": 0.6818,
+      "step": 134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 0.6026,
+      "step": 135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4702702702702705e-05,
+      "loss": 0.6053,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4810810810810811e-05,
+      "loss": 0.6064,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.491891891891892e-05,
+      "loss": 0.6147,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5027027027027028e-05,
+      "loss": 0.6584,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 0.564,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5243243243243244e-05,
+      "loss": 0.6585,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5351351351351353e-05,
+      "loss": 0.533,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.545945945945946e-05,
+      "loss": 0.6361,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.556756756756757e-05,
+      "loss": 0.6519,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 0.5594,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5783783783783787e-05,
+      "loss": 0.5197,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.589189189189189e-05,
+      "loss": 0.6175,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.5585,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.610810810810811e-05,
+      "loss": 0.5524,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.5909,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6324324324324326e-05,
+      "loss": 0.5515,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6432432432432434e-05,
+      "loss": 0.4734,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.654054054054054e-05,
+      "loss": 0.6018,
+      "step": 153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6648648648648652e-05,
+      "loss": 0.5753,
+      "step": 154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 0.6427,
+      "step": 155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6864864864864868e-05,
+      "loss": 0.5972,
+      "step": 156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6972972972972975e-05,
+      "loss": 0.5956,
+      "step": 157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7081081081081083e-05,
+      "loss": 0.5588,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.718918918918919e-05,
+      "loss": 0.6473,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 0.5932,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7405405405405406e-05,
+      "loss": 0.6356,
+      "step": 161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7513513513513517e-05,
+      "loss": 0.5065,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7621621621621622e-05,
+      "loss": 0.5544,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7729729729729733e-05,
+      "loss": 0.5424,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 0.5673,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7945945945945948e-05,
+      "loss": 0.5957,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8054054054054056e-05,
+      "loss": 0.597,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8162162162162164e-05,
+      "loss": 0.5922,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.827027027027027e-05,
+      "loss": 0.66,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 0.5721,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8486486486486487e-05,
+      "loss": 0.5723,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8594594594594598e-05,
+      "loss": 0.5272,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8702702702702706e-05,
+      "loss": 0.5559,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8810810810810813e-05,
+      "loss": 0.5802,
+      "step": 174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 0.5852,
+      "step": 175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.902702702702703e-05,
+      "loss": 0.5158,
+      "step": 176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9135135135135137e-05,
+      "loss": 0.5681,
+      "step": 177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9243243243243244e-05,
+      "loss": 0.5491,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9351351351351352e-05,
+      "loss": 0.5323,
+      "step": 179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 0.7047,
+      "step": 180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956756756756757e-05,
+      "loss": 0.5613,
+      "step": 181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967567567567568e-05,
+      "loss": 0.617,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783783783783786e-05,
+      "loss": 0.5841,
+      "step": 183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9891891891891894e-05,
+      "loss": 0.5942,
+      "step": 184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.6044,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999998610293885e-05,
+      "loss": 0.6454,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999994441175927e-05,
+      "loss": 0.6021,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999987492647274e-05,
+      "loss": 0.573,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999997776470987e-05,
+      "loss": 0.6273,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999965257366413e-05,
+      "loss": 0.6055,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999949970620383e-05,
+      "loss": 0.5062,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999931904476024e-05,
+      "loss": 0.6059,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999911058938363e-05,
+      "loss": 0.5734,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999887434013186e-05,
+      "loss": 0.5353,
+      "step": 194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999861029707063e-05,
+      "loss": 0.5634,
+      "step": 195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999831846027335e-05,
+      "loss": 0.5265,
+      "step": 196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999799882982115e-05,
+      "loss": 0.5747,
+      "step": 197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999976514058028e-05,
+      "loss": 0.5438,
+      "step": 198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999727618831492e-05,
+      "loss": 0.5558,
+      "step": 199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999968731774618e-05,
+      "loss": 0.5505,
+      "step": 200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999964423733554e-05,
+      "loss": 0.5685,
+      "step": 201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999959837761155e-05,
+      "loss": 0.5864,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999549738586954e-05,
+      "loss": 0.5873,
+      "step": 203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999498320275275e-05,
+      "loss": 0.5549,
+      "step": 204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999444122690802e-05,
+      "loss": 0.5756,
+      "step": 205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999387145848595e-05,
+      "loss": 0.5569,
+      "step": 206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999327389764495e-05,
+      "loss": 0.5498,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999926485445511e-05,
+      "loss": 0.5967,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999919953993782e-05,
+      "loss": 0.6037,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999131446230778e-05,
+      "loss": 0.6495,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999906057335291e-05,
+      "loss": 0.6337,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998986921323918e-05,
+      "loss": 0.5847,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998910490164268e-05,
+      "loss": 0.5756,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998831279895204e-05,
+      "loss": 0.6542,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999874929053875e-05,
+      "loss": 0.5803,
+      "step": 215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998664522117682e-05,
+      "loss": 0.612,
+      "step": 216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999857697465557e-05,
+      "loss": 0.6861,
+      "step": 217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999848664817674e-05,
+      "loss": 0.6454,
+      "step": 218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998393542706304e-05,
+      "loss": 0.5834,
+      "step": 219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998297658270137e-05,
+      "loss": 0.6296,
+      "step": 220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998198994894887e-05,
+      "loss": 0.5973,
+      "step": 221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998097552607977e-05,
+      "loss": 0.5904,
+      "step": 222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997993331437607e-05,
+      "loss": 0.6446,
+      "step": 223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999788633141274e-05,
+      "loss": 0.5913,
+      "step": 224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997776552563117e-05,
+      "loss": 0.5481,
+      "step": 225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997663994919247e-05,
+      "loss": 0.5591,
+      "step": 226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997548658512416e-05,
+      "loss": 0.5892,
+      "step": 227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999743054337469e-05,
+      "loss": 0.5253,
+      "step": 228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997309649538885e-05,
+      "loss": 0.5638,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997185977038604e-05,
+      "loss": 0.6335,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999705952590823e-05,
+      "loss": 0.5879,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99969302961829e-05,
+      "loss": 0.5996,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996798287898532e-05,
+      "loss": 0.5397,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996663501091827e-05,
+      "loss": 0.5902,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996525935800234e-05,
+      "loss": 0.518,
+      "step": 235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996385592061996e-05,
+      "loss": 0.6126,
+      "step": 236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996242469916124e-05,
+      "loss": 0.5483,
+      "step": 237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996096569402387e-05,
+      "loss": 0.6116,
+      "step": 238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995947890561345e-05,
+      "loss": 0.5324,
+      "step": 239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999579643343432e-05,
+      "loss": 0.6082,
+      "step": 240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995642198063405e-05,
+      "loss": 0.5828,
+      "step": 241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999548518449147e-05,
+      "loss": 0.5629,
+      "step": 242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999532539276216e-05,
+      "loss": 0.6017,
+      "step": 243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999516282291988e-05,
+      "loss": 0.6103,
+      "step": 244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999499747500982e-05,
+      "loss": 0.5494,
+      "step": 245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999482934907794e-05,
+      "loss": 0.5029,
+      "step": 246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999465844517096e-05,
+      "loss": 0.6462,
+      "step": 247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999448476333639e-05,
+      "loss": 0.579,
+      "step": 248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994308303622502e-05,
+      "loss": 0.5885,
+      "step": 249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994129066078333e-05,
+      "loss": 0.5762,
+      "step": 250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999394705075371e-05,
+      "loss": 0.5989,
+      "step": 251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993762257699218e-05,
+      "loss": 0.5863,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999357468696622e-05,
+      "loss": 0.5575,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999338433860685e-05,
+      "loss": 0.6468,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999319121267401e-05,
+      "loss": 0.541,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9992995309221384e-05,
+      "loss": 0.5533,
+      "step": 256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992796628303418e-05,
+      "loss": 0.5536,
+      "step": 257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992595169975333e-05,
+      "loss": 0.6366,
+      "step": 258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999239093429312e-05,
+      "loss": 0.5474,
+      "step": 259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992183921313554e-05,
+      "loss": 0.5996,
+      "step": 260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991974131094164e-05,
+      "loss": 0.5402,
+      "step": 261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991761563693264e-05,
+      "loss": 0.646,
+      "step": 262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999154621916993e-05,
+      "loss": 0.5739,
+      "step": 263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999132809758402e-05,
+      "loss": 0.614,
+      "step": 264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991107198996156e-05,
+      "loss": 0.5183,
+      "step": 265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990883523467736e-05,
+      "loss": 0.6664,
+      "step": 266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990657071060932e-05,
+      "loss": 0.5551,
+      "step": 267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999042784183868e-05,
+      "loss": 0.5906,
+      "step": 268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990195835864686e-05,
+      "loss": 0.5537,
+      "step": 269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998996105320345e-05,
+      "loss": 0.5696,
+      "step": 270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989723493920218e-05,
+      "loss": 0.5893,
+      "step": 271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989483158081016e-05,
+      "loss": 0.5178,
+      "step": 272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998924004575265e-05,
+      "loss": 0.5838,
+      "step": 273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998899415700268e-05,
+      "loss": 0.5703,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998874549189946e-05,
+      "loss": 0.6117,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99884940505121e-05,
+      "loss": 0.5552,
+      "step": 276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988239832910485e-05,
+      "loss": 0.5864,
+      "step": 277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998798283916527e-05,
+      "loss": 0.6146,
+      "step": 278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987723069347893e-05,
+      "loss": 0.6815,
+      "step": 279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987460523530545e-05,
+      "loss": 0.6106,
+      "step": 280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987195201786206e-05,
+      "loss": 0.6579,
+      "step": 281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986927104188614e-05,
+      "loss": 0.5857,
+      "step": 282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998665623081229e-05,
+      "loss": 0.5343,
+      "step": 283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998638258173251e-05,
+      "loss": 0.5565,
+      "step": 284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986106157025348e-05,
+      "loss": 0.5878,
+      "step": 285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998582695676762e-05,
+      "loss": 0.52,
+      "step": 286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985544981036935e-05,
+      "loss": 0.5712,
+      "step": 287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985260229911664e-05,
+      "loss": 0.608,
+      "step": 288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984972703470952e-05,
+      "loss": 0.586,
+      "step": 289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984682401794713e-05,
+      "loss": 0.6098,
+      "step": 290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984389324963632e-05,
+      "loss": 0.5474,
+      "step": 291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998409347305917e-05,
+      "loss": 0.5925,
+      "step": 292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983794846163556e-05,
+      "loss": 0.5329,
+      "step": 293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983493444359793e-05,
+      "loss": 0.5756,
+      "step": 294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983189267731643e-05,
+      "loss": 0.5278,
+      "step": 295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982882316363664e-05,
+      "loss": 0.5522,
+      "step": 296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998257259034116e-05,
+      "loss": 0.59,
+      "step": 297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982260089750216e-05,
+      "loss": 0.6005,
+      "step": 298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981944814677698e-05,
+      "loss": 0.5908,
+      "step": 299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981626765211228e-05,
+      "loss": 0.5765,
+      "step": 300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981305941439203e-05,
+      "loss": 0.6087,
+      "step": 301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998098234345079e-05,
+      "loss": 0.5662,
+      "step": 302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980655971335944e-05,
+      "loss": 0.6147,
+      "step": 303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998032682518537e-05,
+      "loss": 0.6566,
+      "step": 304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979994905090546e-05,
+      "loss": 0.5517,
+      "step": 305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997966021114373e-05,
+      "loss": 0.5822,
+      "step": 306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997932274343795e-05,
+      "loss": 0.6395,
+      "step": 307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978982502067e-05,
+      "loss": 0.5351,
+      "step": 308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978639487125445e-05,
+      "loss": 0.5088,
+      "step": 309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978293698708627e-05,
+      "loss": 0.5484,
+      "step": 310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977945136912654e-05,
+      "loss": 0.5821,
+      "step": 311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977593801834403e-05,
+      "loss": 0.544,
+      "step": 312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977239693571525e-05,
+      "loss": 0.5775,
+      "step": 313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976882812222444e-05,
+      "loss": 0.5793,
+      "step": 314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997652315788635e-05,
+      "loss": 0.5351,
+      "step": 315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976160730663208e-05,
+      "loss": 0.5517,
+      "step": 316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975795530653747e-05,
+      "loss": 0.5783,
+      "step": 317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975427557959473e-05,
+      "loss": 0.5985,
+      "step": 318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997505681268267e-05,
+      "loss": 0.5847,
+      "step": 319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974683294926366e-05,
+      "loss": 0.6107,
+      "step": 320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974307004794388e-05,
+      "loss": 0.6243,
+      "step": 321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997392794239132e-05,
+      "loss": 0.596,
+      "step": 322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997354610782252e-05,
+      "loss": 0.5594,
+      "step": 323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997316150119412e-05,
+      "loss": 0.5493,
+      "step": 324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972774122613008e-05,
+      "loss": 0.5695,
+      "step": 325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972383972186858e-05,
+      "loss": 0.5342,
+      "step": 326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971991050024112e-05,
+      "loss": 0.5765,
+      "step": 327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997159535623397e-05,
+      "loss": 0.535,
+      "step": 328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971196890926423e-05,
+      "loss": 0.5834,
+      "step": 329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997079565421221e-05,
+      "loss": 0.5925,
+      "step": 330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997039164620286e-05,
+      "loss": 0.5717,
+      "step": 331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969984867010663e-05,
+      "loss": 0.5535,
+      "step": 332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969575316748673e-05,
+      "loss": 0.5787,
+      "step": 333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969162995530723e-05,
+      "loss": 0.5674,
+      "step": 334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968747903471423e-05,
+      "loss": 0.5798,
+      "step": 335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996833004068613e-05,
+      "loss": 0.5346,
+      "step": 336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967909407290995e-05,
+      "loss": 0.6281,
+      "step": 337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967486003402924e-05,
+      "loss": 0.6502,
+      "step": 338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967059829139602e-05,
+      "loss": 0.5807,
+      "step": 339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966630884619484e-05,
+      "loss": 0.6203,
+      "step": 340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996619916996178e-05,
+      "loss": 0.5454,
+      "step": 341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965764685286493e-05,
+      "loss": 0.5577,
+      "step": 342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965327430714377e-05,
+      "loss": 0.4962,
+      "step": 343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964887406366967e-05,
+      "loss": 0.5816,
+      "step": 344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996444461236656e-05,
+      "loss": 0.574,
+      "step": 345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996399904883623e-05,
+      "loss": 0.5295,
+      "step": 346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963550715899816e-05,
+      "loss": 0.5674,
+      "step": 347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996309961368193e-05,
+      "loss": 0.5468,
+      "step": 348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962645742307953e-05,
+      "loss": 0.6046,
+      "step": 349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962189101904027e-05,
+      "loss": 0.5957,
+      "step": 350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961729692597077e-05,
+      "loss": 0.5467,
+      "step": 351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961267514514793e-05,
+      "loss": 0.5378,
+      "step": 352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996080256778563e-05,
+      "loss": 0.5433,
+      "step": 353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996033485253882e-05,
+      "loss": 0.5745,
+      "step": 354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959864368904353e-05,
+      "loss": 0.5661,
+      "step": 355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959391117013003e-05,
+      "loss": 0.5881,
+      "step": 356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99589150969963e-05,
+      "loss": 0.6048,
+      "step": 357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995843630898656e-05,
+      "loss": 0.6533,
+      "step": 358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995795475311685e-05,
+      "loss": 0.6525,
+      "step": 359
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957470429521013e-05,
+      "loss": 0.5386,
+      "step": 360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956983338333666e-05,
+      "loss": 0.5864,
+      "step": 361
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995649347969019e-05,
+      "loss": 0.5476,
+      "step": 362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956000853726737e-05,
+      "loss": 0.5271,
+      "step": 363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955505460580226e-05,
+      "loss": 0.5736,
+      "step": 364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955007300388353e-05,
+      "loss": 0.592,
+      "step": 365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995450637328957e-05,
+      "loss": 0.5268,
+      "step": 366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954002679423116e-05,
+      "loss": 0.5292,
+      "step": 367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953496218928978e-05,
+      "loss": 0.5863,
+      "step": 368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952986991947927e-05,
+      "loss": 0.5983,
+      "step": 369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952474998621495e-05,
+      "loss": 0.5603,
+      "step": 370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951960239091992e-05,
+      "loss": 0.4998,
+      "step": 371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951442713502484e-05,
+      "loss": 0.505,
+      "step": 372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950922421996814e-05,
+      "loss": 0.5899,
+      "step": 373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950399364719595e-05,
+      "loss": 0.5215,
+      "step": 374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994987354181621e-05,
+      "loss": 0.5793,
+      "step": 375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949344953432797e-05,
+      "loss": 0.587,
+      "step": 376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948813599716278e-05,
+      "loss": 0.5877,
+      "step": 377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994827948081434e-05,
+      "loss": 0.5639,
+      "step": 378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994774259687543e-05,
+      "loss": 0.5636,
+      "step": 379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947202948048782e-05,
+      "loss": 0.5474,
+      "step": 380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946660534484373e-05,
+      "loss": 0.5844,
+      "step": 381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946115356332968e-05,
+      "loss": 0.6345,
+      "step": 382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99455674137461e-05,
+      "loss": 0.5681,
+      "step": 383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945016706876056e-05,
+      "loss": 0.5669,
+      "step": 384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99444632358759e-05,
+      "loss": 0.5583,
+      "step": 385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943907000899478e-05,
+      "loss": 0.5976,
+      "step": 386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943348002101374e-05,
+      "loss": 0.533,
+      "step": 387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942786239636964e-05,
+      "loss": 0.6126,
+      "step": 388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942221713662383e-05,
+      "loss": 0.6454,
+      "step": 389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994165442433454e-05,
+      "loss": 0.5689,
+      "step": 390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941084371811107e-05,
+      "loss": 0.5236,
+      "step": 391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940511556250524e-05,
+      "loss": 0.5675,
+      "step": 392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939935977812e-05,
+      "loss": 0.5906,
+      "step": 393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993935763665551e-05,
+      "loss": 0.5685,
+      "step": 394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938776532941798e-05,
+      "loss": 0.5897,
+      "step": 395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938192666832386e-05,
+      "loss": 0.6118,
+      "step": 396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993760603848954e-05,
+      "loss": 0.5336,
+      "step": 397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937016648076325e-05,
+      "loss": 0.5851,
+      "step": 398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936424495756543e-05,
+      "loss": 0.5496,
+      "step": 399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993582958169478e-05,
+      "loss": 0.5445,
+      "step": 400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935231906056392e-05,
+      "loss": 0.5209,
+      "step": 401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934631469007495e-05,
+      "loss": 0.5211,
+      "step": 402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934028270714975e-05,
+      "loss": 0.5713,
+      "step": 403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933422311346487e-05,
+      "loss": 0.5115,
+      "step": 404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932813591070452e-05,
+      "loss": 0.5674,
+      "step": 405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932202110056056e-05,
+      "loss": 0.5465,
+      "step": 406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993158786847326e-05,
+      "loss": 0.5122,
+      "step": 407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930970866492778e-05,
+      "loss": 0.5028,
+      "step": 408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993035110428611e-05,
+      "loss": 0.5853,
+      "step": 409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992972858202551e-05,
+      "loss": 0.5797,
+      "step": 410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929103299884e-05,
+      "loss": 0.5849,
+      "step": 411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928475258035373e-05,
+      "loss": 0.5535,
+      "step": 412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927844456654188e-05,
+      "loss": 0.5254,
+      "step": 413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927210895915773e-05,
+      "loss": 0.5643,
+      "step": 414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926574575996216e-05,
+      "loss": 0.6423,
+      "step": 415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992593549707238e-05,
+      "loss": 0.5657,
+      "step": 416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925293659321892e-05,
+      "loss": 0.5232,
+      "step": 417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924649062923143e-05,
+      "loss": 0.6044,
+      "step": 418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924001708055295e-05,
+      "loss": 0.5788,
+      "step": 419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9923351594898272e-05,
+      "loss": 0.5214,
+      "step": 420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992269872363277e-05,
+      "loss": 0.5438,
+      "step": 421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992204309444024e-05,
+      "loss": 0.5462,
+      "step": 422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921384707502926e-05,
+      "loss": 0.5625,
+      "step": 423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920723563003803e-05,
+      "loss": 0.5672,
+      "step": 424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992005966112664e-05,
+      "loss": 0.5291,
+      "step": 425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919393002055964e-05,
+      "loss": 0.5683,
+      "step": 426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991872358597706e-05,
+      "loss": 0.5498,
+      "step": 427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991805141307599e-05,
+      "loss": 0.559,
+      "step": 428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991737648353958e-05,
+      "loss": 0.5891,
+      "step": 429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916698797555418e-05,
+      "loss": 0.5646,
+      "step": 430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916018355311862e-05,
+      "loss": 0.5509,
+      "step": 431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991533515699804e-05,
+      "loss": 0.5255,
+      "step": 432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991464920280383e-05,
+      "loss": 0.5729,
+      "step": 433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913960492919897e-05,
+      "loss": 0.5628,
+      "step": 434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913269027537658e-05,
+      "loss": 0.493,
+      "step": 435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912574806849293e-05,
+      "loss": 0.5506,
+      "step": 436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911877831047767e-05,
+      "loss": 0.5577,
+      "step": 437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991117810032679e-05,
+      "loss": 0.5619,
+      "step": 438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991047561488085e-05,
+      "loss": 0.5963,
+      "step": 439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909770374905197e-05,
+      "loss": 0.5694,
+      "step": 440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909062380595842e-05,
+      "loss": 0.5229,
+      "step": 441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990835163214957e-05,
+      "loss": 0.5275,
+      "step": 442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990763812976392e-05,
+      "loss": 0.5209,
+      "step": 443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906921873637214e-05,
+      "loss": 0.5668,
+      "step": 444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906202863968525e-05,
+      "loss": 0.595,
+      "step": 445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990548110095769e-05,
+      "loss": 0.5231,
+      "step": 446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904756584805326e-05,
+      "loss": 0.5257,
+      "step": 447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904029315712802e-05,
+      "loss": 0.5553,
+      "step": 448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903299293882252e-05,
+      "loss": 0.5937,
+      "step": 449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902566519516584e-05,
+      "loss": 0.5549,
+      "step": 450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901830992819466e-05,
+      "loss": 0.5584,
+      "step": 451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901092713995326e-05,
+      "loss": 0.6454,
+      "step": 452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900351683249372e-05,
+      "loss": 0.4934,
+      "step": 453
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899607900787553e-05,
+      "loss": 0.639,
+      "step": 454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989886136681661e-05,
+      "loss": 0.4594,
+      "step": 455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898112081544033e-05,
+      "loss": 0.566,
+      "step": 456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897360045178075e-05,
+      "loss": 0.5506,
+      "step": 457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896605257927755e-05,
+      "loss": 0.4991,
+      "step": 458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989584772000287e-05,
+      "loss": 0.5731,
+      "step": 459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895087431613963e-05,
+      "loss": 0.5727,
+      "step": 460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989432439297235e-05,
+      "loss": 0.542,
+      "step": 461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893558604290114e-05,
+      "loss": 0.5005,
+      "step": 462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892790065780097e-05,
+      "loss": 0.5598,
+      "step": 463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892018777655912e-05,
+      "loss": 0.5489,
+      "step": 464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989124474013192e-05,
+      "loss": 0.5601,
+      "step": 465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890467953423272e-05,
+      "loss": 0.5419,
+      "step": 466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889688417745863e-05,
+      "loss": 0.5101,
+      "step": 467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888906133316357e-05,
+      "loss": 0.5682,
+      "step": 468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888121100352182e-05,
+      "loss": 0.5904,
+      "step": 469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887333319071537e-05,
+      "loss": 0.5658,
+      "step": 470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886542789693375e-05,
+      "loss": 0.6163,
+      "step": 471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885749512437416e-05,
+      "loss": 0.5861,
+      "step": 472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884953487524143e-05,
+      "loss": 0.6148,
+      "step": 473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988415471517481e-05,
+      "loss": 0.5036,
+      "step": 474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883353195611422e-05,
+      "loss": 0.5509,
+      "step": 475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882548929056758e-05,
+      "loss": 0.5059,
+      "step": 476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988174191573436e-05,
+      "loss": 0.6074,
+      "step": 477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880932155868522e-05,
+      "loss": 0.6268,
+      "step": 478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880119649684315e-05,
+      "loss": 0.5218,
+      "step": 479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987930439740757e-05,
+      "loss": 0.5435,
+      "step": 480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878486399264875e-05,
+      "loss": 0.6003,
+      "step": 481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877665655483587e-05,
+      "loss": 0.5869,
+      "step": 482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987684216629182e-05,
+      "loss": 0.563,
+      "step": 483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876015931918468e-05,
+      "loss": 0.4723,
+      "step": 484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875186952593162e-05,
+      "loss": 0.5465,
+      "step": 485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874355228546316e-05,
+      "loss": 0.5713,
+      "step": 486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98735207600091e-05,
+      "loss": 0.6356,
+      "step": 487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872683547213446e-05,
+      "loss": 0.539,
+      "step": 488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871843590392053e-05,
+      "loss": 0.526,
+      "step": 489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871000889778377e-05,
+      "loss": 0.574,
+      "step": 490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870155445606637e-05,
+      "loss": 0.5312,
+      "step": 491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869307258111822e-05,
+      "loss": 0.5353,
+      "step": 492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868456327529674e-05,
+      "loss": 0.5726,
+      "step": 493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867602654096702e-05,
+      "loss": 0.5105,
+      "step": 494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986674623805018e-05,
+      "loss": 0.5268,
+      "step": 495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865887079628138e-05,
+      "loss": 0.5671,
+      "step": 496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865025179069374e-05,
+      "loss": 0.5056,
+      "step": 497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864160536613447e-05,
+      "loss": 0.5435,
+      "step": 498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986329315250067e-05,
+      "loss": 0.5959,
+      "step": 499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862423026972136e-05,
+      "loss": 0.5094,
+      "step": 500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861550160269675e-05,
+      "loss": 0.6154,
+      "step": 501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860674552635906e-05,
+      "loss": 0.516,
+      "step": 502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859796204314188e-05,
+      "loss": 0.5642,
+      "step": 503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985891511554865e-05,
+      "loss": 0.5034,
+      "step": 504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985803128658419e-05,
+      "loss": 0.5563,
+      "step": 505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857144717666453e-05,
+      "loss": 0.5209,
+      "step": 506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856255409041855e-05,
+      "loss": 0.5285,
+      "step": 507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855363360957573e-05,
+      "loss": 0.5376,
+      "step": 508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854468573661548e-05,
+      "loss": 0.589,
+      "step": 509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853571047402467e-05,
+      "loss": 0.5779,
+      "step": 510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98526707824298e-05,
+      "loss": 0.6068,
+      "step": 511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851767778993764e-05,
+      "loss": 0.5151,
+      "step": 512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985086203734534e-05,
+      "loss": 0.5261,
+      "step": 513
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849953557736274e-05,
+      "loss": 0.5729,
+      "step": 514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849042340419063e-05,
+      "loss": 0.487,
+      "step": 515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984812838564698e-05,
+      "loss": 0.5158,
+      "step": 516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847211693674046e-05,
+      "loss": 0.4992,
+      "step": 517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846292264755047e-05,
+      "loss": 0.6199,
+      "step": 518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845370099145534e-05,
+      "loss": 0.5439,
+      "step": 519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844445197101815e-05,
+      "loss": 0.5486,
+      "step": 520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843517558880955e-05,
+      "loss": 0.6331,
+      "step": 521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842587184740782e-05,
+      "loss": 0.4629,
+      "step": 522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984165407493989e-05,
+      "loss": 0.6042,
+      "step": 523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840718229737623e-05,
+      "loss": 0.5957,
+      "step": 524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839779649394098e-05,
+      "loss": 0.5418,
+      "step": 525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983883833417018e-05,
+      "loss": 0.5372,
+      "step": 526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837894284327495e-05,
+      "loss": 0.5534,
+      "step": 527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836947500128447e-05,
+      "loss": 0.5778,
+      "step": 528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835997981836174e-05,
+      "loss": 0.5407,
+      "step": 529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835045729714593e-05,
+      "loss": 0.6094,
+      "step": 530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983409074402837e-05,
+      "loss": 0.5351,
+      "step": 531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833133025042938e-05,
+      "loss": 0.5332,
+      "step": 532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832172573024483e-05,
+      "loss": 0.5934,
+      "step": 533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831209388239956e-05,
+      "loss": 0.5174,
+      "step": 534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830243470957067e-05,
+      "loss": 0.5218,
+      "step": 535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982927482144428e-05,
+      "loss": 0.6092,
+      "step": 536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828303439970833e-05,
+      "loss": 0.4764,
+      "step": 537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827329326806702e-05,
+      "loss": 0.5083,
+      "step": 538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982635248222264e-05,
+      "loss": 0.5356,
+      "step": 539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825372906490145e-05,
+      "loss": 0.5475,
+      "step": 540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982439059988149e-05,
+      "loss": 0.6307,
+      "step": 541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823405562669693e-05,
+      "loss": 0.5647,
+      "step": 542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822417795128535e-05,
+      "loss": 0.5872,
+      "step": 543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821427297532564e-05,
+      "loss": 0.4688,
+      "step": 544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820434070157076e-05,
+      "loss": 0.5338,
+      "step": 545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981943811327813e-05,
+      "loss": 0.5271,
+      "step": 546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818439427172542e-05,
+      "loss": 0.4768,
+      "step": 547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817438012117892e-05,
+      "loss": 0.561,
+      "step": 548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816433868392512e-05,
+      "loss": 0.5221,
+      "step": 549
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815426996275495e-05,
+      "loss": 0.4892,
+      "step": 550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981441739604669e-05,
+      "loss": 0.5326,
+      "step": 551
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981340506798671e-05,
+      "loss": 0.5152,
+      "step": 552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812390012376923e-05,
+      "loss": 0.5442,
+      "step": 553
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811372229499455e-05,
+      "loss": 0.5632,
+      "step": 554
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810351719637185e-05,
+      "loss": 0.5515,
+      "step": 555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980932848307376e-05,
+      "loss": 0.5572,
+      "step": 556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808302520093575e-05,
+      "loss": 0.5349,
+      "step": 557
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807273830981794e-05,
+      "loss": 0.5435,
+      "step": 558
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806242416024327e-05,
+      "loss": 0.6023,
+      "step": 559
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805208275507847e-05,
+      "loss": 0.5715,
+      "step": 560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804171409719784e-05,
+      "loss": 0.4667,
+      "step": 561
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803131818948328e-05,
+      "loss": 0.5469,
+      "step": 562
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802089503482426e-05,
+      "loss": 0.5434,
+      "step": 563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801044463611774e-05,
+      "loss": 0.5175,
+      "step": 564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799996699626835e-05,
+      "loss": 0.5819,
+      "step": 565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798946211818827e-05,
+      "loss": 0.4527,
+      "step": 566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797893000479725e-05,
+      "loss": 0.5835,
+      "step": 567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796837065902254e-05,
+      "loss": 0.5263,
+      "step": 568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795778408379908e-05,
+      "loss": 0.5222,
+      "step": 569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794717028206928e-05,
+      "loss": 0.5359,
+      "step": 570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793652925678315e-05,
+      "loss": 0.5534,
+      "step": 571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979258610108983e-05,
+      "loss": 0.5356,
+      "step": 572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791516554737987e-05,
+      "loss": 0.5848,
+      "step": 573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979044428692005e-05,
+      "loss": 0.5837,
+      "step": 574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978936929793406e-05,
+      "loss": 0.5695,
+      "step": 575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978829158807879e-05,
+      "loss": 0.5551,
+      "step": 576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978721115765378e-05,
+      "loss": 0.5064,
+      "step": 577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786128006959335e-05,
+      "loss": 0.5902,
+      "step": 578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785042136296498e-05,
+      "loss": 0.5337,
+      "step": 579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978395354596708e-05,
+      "loss": 0.5797,
+      "step": 580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782862236273646e-05,
+      "loss": 0.5233,
+      "step": 581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781768207519515e-05,
+      "loss": 0.5425,
+      "step": 582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780671460008763e-05,
+      "loss": 0.4765,
+      "step": 583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779571994046225e-05,
+      "loss": 0.4822,
+      "step": 584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977846980993748e-05,
+      "loss": 0.4923,
+      "step": 585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977736490798888e-05,
+      "loss": 0.4896,
+      "step": 586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776257288507518e-05,
+      "loss": 0.556,
+      "step": 587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775146951801242e-05,
+      "loss": 0.5681,
+      "step": 588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774033898178668e-05,
+      "loss": 0.5394,
+      "step": 589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772918127949156e-05,
+      "loss": 0.4863,
+      "step": 590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771799641422828e-05,
+      "loss": 0.455,
+      "step": 591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770678438910552e-05,
+      "loss": 0.5734,
+      "step": 592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976955452072396e-05,
+      "loss": 0.5398,
+      "step": 593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768427887175434e-05,
+      "loss": 0.5416,
+      "step": 594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976729853857811e-05,
+      "loss": 0.6178,
+      "step": 595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766166475245888e-05,
+      "loss": 0.6415,
+      "step": 596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765031697493407e-05,
+      "loss": 0.6074,
+      "step": 597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976389420563607e-05,
+      "loss": 0.5274,
+      "step": 598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976275399999004e-05,
+      "loss": 0.5508,
+      "step": 599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761611080872216e-05,
+      "loss": 0.5323,
+      "step": 600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976046544860027e-05,
+      "loss": 0.5768,
+      "step": 601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759317103492615e-05,
+      "loss": 0.5017,
+      "step": 602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975816604586843e-05,
+      "loss": 0.5249,
+      "step": 603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975701227604764e-05,
+      "loss": 0.4715,
+      "step": 604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975585579435092e-05,
+      "loss": 0.6069,
+      "step": 605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754696601099707e-05,
+      "loss": 0.5629,
+      "step": 606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975353469661619e-05,
+      "loss": 0.5764,
+      "step": 607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975237008122331e-05,
+      "loss": 0.4682,
+      "step": 608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751202755244758e-05,
+      "loss": 0.5366,
+      "step": 609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750032719004984e-05,
+      "loss": 0.4821,
+      "step": 610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748859972829193e-05,
+      "loss": 0.5217,
+      "step": 611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747684517043335e-05,
+      "loss": 0.5575,
+      "step": 612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746506351974117e-05,
+      "loss": 0.5759,
+      "step": 613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745325477949003e-05,
+      "loss": 0.5595,
+      "step": 614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974414189529621e-05,
+      "loss": 0.4947,
+      "step": 615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742955604344693e-05,
+      "loss": 0.4822,
+      "step": 616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974176660542418e-05,
+      "loss": 0.5446,
+      "step": 617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974057489886514e-05,
+      "loss": 0.5421,
+      "step": 618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97393804849988e-05,
+      "loss": 0.5383,
+      "step": 619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738183364157133e-05,
+      "loss": 0.5568,
+      "step": 620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973698353667287e-05,
+      "loss": 0.4939,
+      "step": 621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973578100287949e-05,
+      "loss": 0.5147,
+      "step": 622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734575763111238e-05,
+      "loss": 0.4984,
+      "step": 623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733367817703083e-05,
+      "loss": 0.6306,
+      "step": 624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9732157166990773e-05,
+      "loss": 0.5776,
+      "step": 625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730943811310795e-05,
+      "loss": 0.5026,
+      "step": 626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729727751000393e-05,
+      "loss": 0.6025,
+      "step": 627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972850898639756e-05,
+      "loss": 0.6064,
+      "step": 628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727287517841037e-05,
+      "loss": 0.5467,
+      "step": 629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972606334567032e-05,
+      "loss": 0.5433,
+      "step": 630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724836470225665e-05,
+      "loss": 0.5018,
+      "step": 631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723606891848063e-05,
+      "loss": 0.5507,
+      "step": 632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972237461087927e-05,
+      "loss": 0.5146,
+      "step": 633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972113962766178e-05,
+      "loss": 0.5363,
+      "step": 634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719901942538856e-05,
+      "loss": 0.5455,
+      "step": 635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718661555854497e-05,
+      "loss": 0.5225,
+      "step": 636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717418467953456e-05,
+      "loss": 0.4833,
+      "step": 637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971617267918124e-05,
+      "loss": 0.5543,
+      "step": 638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714924189884104e-05,
+      "loss": 0.5921,
+      "step": 639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713673000409056e-05,
+      "loss": 0.5365,
+      "step": 640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971241911110385e-05,
+      "loss": 0.531,
+      "step": 641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711162522317e-05,
+      "loss": 0.5278,
+      "step": 642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709903234397758e-05,
+      "loss": 0.4952,
+      "step": 643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708641247696136e-05,
+      "loss": 0.4889,
+      "step": 644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707376562562886e-05,
+      "loss": 0.5934,
+      "step": 645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706109179349522e-05,
+      "loss": 0.509,
+      "step": 646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704839098408303e-05,
+      "loss": 0.6238,
+      "step": 647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970356632009223e-05,
+      "loss": 0.5536,
+      "step": 648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702290844755064e-05,
+      "loss": 0.5122,
+      "step": 649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701012672751317e-05,
+      "loss": 0.5441,
+      "step": 650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699731804436238e-05,
+      "loss": 0.5318,
+      "step": 651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698448240165837e-05,
+      "loss": 0.4911,
+      "step": 652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969716198029687e-05,
+      "loss": 0.5542,
+      "step": 653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969587302518684e-05,
+      "loss": 0.5777,
+      "step": 654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694581375194004e-05,
+      "loss": 0.5682,
+      "step": 655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693287030677363e-05,
+      "loss": 0.5431,
+      "step": 656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691989991996663e-05,
+      "loss": 0.6155,
+      "step": 657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969069025951241e-05,
+      "loss": 0.5509,
+      "step": 658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689387833585855e-05,
+      "loss": 0.5312,
+      "step": 659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688082714578993e-05,
+      "loss": 0.5437,
+      "step": 660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968677490285457e-05,
+      "loss": 0.5677,
+      "step": 661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685464398776083e-05,
+      "loss": 0.5659,
+      "step": 662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684151202707774e-05,
+      "loss": 0.5475,
+      "step": 663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682835315014633e-05,
+      "loss": 0.5576,
+      "step": 664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96815167360624e-05,
+      "loss": 0.4869,
+      "step": 665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9680195466217564e-05,
+      "loss": 0.4726,
+      "step": 666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678871505847357e-05,
+      "loss": 0.5018,
+      "step": 667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967754485531977e-05,
+      "loss": 0.5382,
+      "step": 668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676215515003524e-05,
+      "loss": 0.5347,
+      "step": 669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96748834852681e-05,
+      "loss": 0.485,
+      "step": 670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967354876648373e-05,
+      "loss": 0.5303,
+      "step": 671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967221135902138e-05,
+      "loss": 0.5244,
+      "step": 672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670871263252777e-05,
+      "loss": 0.5796,
+      "step": 673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669528479550385e-05,
+      "loss": 0.5025,
+      "step": 674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668183008287417e-05,
+      "loss": 0.52,
+      "step": 675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966683484983784e-05,
+      "loss": 0.5092,
+      "step": 676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966548400457636e-05,
+      "loss": 0.5154,
+      "step": 677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664130472878432e-05,
+      "loss": 0.5277,
+      "step": 678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662774255120262e-05,
+      "loss": 0.5547,
+      "step": 679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966141535167879e-05,
+      "loss": 0.488,
+      "step": 680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660053762931723e-05,
+      "loss": 0.5587,
+      "step": 681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658689489257498e-05,
+      "loss": 0.499,
+      "step": 682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657322531035298e-05,
+      "loss": 0.5271,
+      "step": 683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655952888645062e-05,
+      "loss": 0.5072,
+      "step": 684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965458056246747e-05,
+      "loss": 0.5273,
+      "step": 685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653205552883945e-05,
+      "loss": 0.5435,
+      "step": 686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965182786027666e-05,
+      "loss": 0.4866,
+      "step": 687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650447485028538e-05,
+      "loss": 0.5759,
+      "step": 688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649064427523234e-05,
+      "loss": 0.5177,
+      "step": 689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647678688145163e-05,
+      "loss": 0.5469,
+      "step": 690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646290267279472e-05,
+      "loss": 0.5431,
+      "step": 691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964489916531207e-05,
+      "loss": 0.5593,
+      "step": 692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643505382629593e-05,
+      "loss": 0.5345,
+      "step": 693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642108919619434e-05,
+      "loss": 0.5607,
+      "step": 694
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964070977666973e-05,
+      "loss": 0.4588,
+      "step": 695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639307954169357e-05,
+      "loss": 0.6692,
+      "step": 696
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637903452507937e-05,
+      "loss": 0.6389,
+      "step": 697
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963649627207585e-05,
+      "loss": 0.5455,
+      "step": 698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635086413264196e-05,
+      "loss": 0.52,
+      "step": 699
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963367387646484e-05,
+      "loss": 0.6011,
+      "step": 700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963225866207038e-05,
+      "loss": 0.561,
+      "step": 701
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963084077047417e-05,
+      "loss": 0.5046,
+      "step": 702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629420202070294e-05,
+      "loss": 0.5814,
+      "step": 703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962799695725359e-05,
+      "loss": 0.5586,
+      "step": 704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626571036419632e-05,
+      "loss": 0.5314,
+      "step": 705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625142439964747e-05,
+      "loss": 0.5071,
+      "step": 706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9623711168286e-05,
+      "loss": 0.5599,
+      "step": 707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622277221781195e-05,
+      "loss": 0.5663,
+      "step": 708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620840600848892e-05,
+      "loss": 0.501,
+      "step": 709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961940130588838e-05,
+      "loss": 0.48,
+      "step": 710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961795933729971e-05,
+      "loss": 0.4867,
+      "step": 711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961651469548365e-05,
+      "loss": 0.5338,
+      "step": 712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615067380841735e-05,
+      "loss": 0.4884,
+      "step": 713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613617393776233e-05,
+      "loss": 0.5653,
+      "step": 714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961216473469015e-05,
+      "loss": 0.5191,
+      "step": 715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610709403987248e-05,
+      "loss": 0.473,
+      "step": 716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609251402072017e-05,
+      "loss": 0.5412,
+      "step": 717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607790729349694e-05,
+      "loss": 0.552,
+      "step": 718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606327386226264e-05,
+      "loss": 0.4935,
+      "step": 719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960486137310845e-05,
+      "loss": 0.58,
+      "step": 720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603392690403722e-05,
+      "loss": 0.5667,
+      "step": 721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960192133852028e-05,
+      "loss": 0.5972,
+      "step": 722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600447317867072e-05,
+      "loss": 0.5695,
+      "step": 723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.95989706288538e-05,
+      "loss": 0.5493,
+      "step": 724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597491271890883e-05,
+      "loss": 0.5373,
+      "step": 725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596009247389506e-05,
+      "loss": 0.514,
+      "step": 726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959452455576158e-05,
+      "loss": 0.5568,
+      "step": 727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593037197419766e-05,
+      "loss": 0.482,
+      "step": 728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591547172777458e-05,
+      "loss": 0.5294,
+      "step": 729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590054482248794e-05,
+      "loss": 0.5535,
+      "step": 730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958855912624866e-05,
+      "loss": 0.4445,
+      "step": 731
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958706110519267e-05,
+      "loss": 0.5153,
+      "step": 732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585560419497192e-05,
+      "loss": 0.5275,
+      "step": 733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958405706957932e-05,
+      "loss": 0.5651,
+      "step": 734
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582551055856914e-05,
+      "loss": 0.5836,
+      "step": 735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958104237874854e-05,
+      "loss": 0.6032,
+      "step": 736
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579531038673528e-05,
+      "loss": 0.5525,
+      "step": 737
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578017036051943e-05,
+      "loss": 0.5603,
+      "step": 738
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957650037130458e-05,
+      "loss": 0.55,
+      "step": 739
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574981044853e-05,
+      "loss": 0.5562,
+      "step": 740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957345905711947e-05,
+      "loss": 0.5105,
+      "step": 741
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571934408527024e-05,
+      "loss": 0.5411,
+      "step": 742
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570407099499416e-05,
+      "loss": 0.5973,
+      "step": 743
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956887713046115e-05,
+      "loss": 0.5309,
+      "step": 744
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956734450183748e-05,
+      "loss": 0.5174,
+      "step": 745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565809214054366e-05,
+      "loss": 0.5077,
+      "step": 746
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564271267538544e-05,
+      "loss": 0.6376,
+      "step": 747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562730662717464e-05,
+      "loss": 0.503,
+      "step": 748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956118740001933e-05,
+      "loss": 0.5314,
+      "step": 749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955964147987307e-05,
+      "loss": 0.5897,
+      "step": 750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558092902708365e-05,
+      "loss": 0.4183,
+      "step": 751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556541668955627e-05,
+      "loss": 0.5608,
+      "step": 752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554987779046013e-05,
+      "loss": 0.576,
+      "step": 753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95534312334114e-05,
+      "loss": 0.4511,
+      "step": 754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955187203248443e-05,
+      "loss": 0.5473,
+      "step": 755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550310176698465e-05,
+      "loss": 0.499,
+      "step": 756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548745666487605e-05,
+      "loss": 0.5441,
+      "step": 757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547178502286694e-05,
+      "loss": 0.5971,
+      "step": 758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954560868453131e-05,
+      "loss": 0.4966,
+      "step": 759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544036213657772e-05,
+      "loss": 0.5739,
+      "step": 760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542461090103138e-05,
+      "loss": 0.5336,
+      "step": 761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540883314305197e-05,
+      "loss": 0.4975,
+      "step": 762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539302886702475e-05,
+      "loss": 0.5453,
+      "step": 763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953771980773424e-05,
+      "loss": 0.5258,
+      "step": 764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536134077840495e-05,
+      "loss": 0.5688,
+      "step": 765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953454569746198e-05,
+      "loss": 0.503,
+      "step": 766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532954667040172e-05,
+      "loss": 0.5122,
+      "step": 767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9531360987017283e-05,
+      "loss": 0.5588,
+      "step": 768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529764657836262e-05,
+      "loss": 0.5072,
+      "step": 769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528165679940792e-05,
+      "loss": 0.5477,
+      "step": 770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526564053775305e-05,
+      "loss": 0.5058,
+      "step": 771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524959779784946e-05,
+      "loss": 0.5465,
+      "step": 772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952335285841562e-05,
+      "loss": 0.5815,
+      "step": 773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952174329011395e-05,
+      "loss": 0.6165,
+      "step": 774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95201310753273e-05,
+      "loss": 0.59,
+      "step": 775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518516214503776e-05,
+      "loss": 0.5503,
+      "step": 776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516898708092212e-05,
+      "loss": 0.4739,
+      "step": 777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951527855654218e-05,
+      "loss": 0.5117,
+      "step": 778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951365576030399e-05,
+      "loss": 0.5198,
+      "step": 779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512030319828678e-05,
+      "loss": 0.5265,
+      "step": 780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510402235568022e-05,
+      "loss": 0.5323,
+      "step": 781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508771507974538e-05,
+      "loss": 0.5083,
+      "step": 782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950713813750147e-05,
+      "loss": 0.5022,
+      "step": 783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95055021246028e-05,
+      "loss": 0.5211,
+      "step": 784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950386346973324e-05,
+      "loss": 0.5427,
+      "step": 785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950222217334824e-05,
+      "loss": 0.5246,
+      "step": 786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950057823590399e-05,
+      "loss": 0.5014,
+      "step": 787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498931657857404e-05,
+      "loss": 0.4897,
+      "step": 788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949728243966613e-05,
+      "loss": 0.4926,
+      "step": 789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495630581788563e-05,
+      "loss": 0.5694,
+      "step": 790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493976084683814e-05,
+      "loss": 0.5399,
+      "step": 791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949231894881174e-05,
+      "loss": 0.5214,
+      "step": 792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490659174632926e-05,
+      "loss": 0.5358,
+      "step": 793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948899676260869e-05,
+      "loss": 0.5511,
+      "step": 794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948733171320109e-05,
+      "loss": 0.5415,
+      "step": 795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485664026872907e-05,
+      "loss": 0.5883,
+      "step": 796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483993704087666e-05,
+      "loss": 0.5052,
+      "step": 797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948232074530961e-05,
+      "loss": 0.5281,
+      "step": 798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948064515100373e-05,
+      "loss": 0.542,
+      "step": 799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478966921635735e-05,
+      "loss": 0.4998,
+      "step": 800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477286057672084e-05,
+      "loss": 0.5064,
+      "step": 801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947560255957995e-05,
+      "loss": 0.5589,
+      "step": 802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473916427827252e-05,
+      "loss": 0.5628,
+      "step": 803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947222766288263e-05,
+      "loss": 0.5321,
+      "step": 804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947053626521547e-05,
+      "loss": 0.5375,
+      "step": 805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468842235295874e-05,
+      "loss": 0.5335,
+      "step": 806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467145573594684e-05,
+      "loss": 0.5547,
+      "step": 807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465446280583476e-05,
+      "loss": 0.5152,
+      "step": 808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9463744356734548e-05,
+      "loss": 0.5339,
+      "step": 809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9462039802520937e-05,
+      "loss": 0.5584,
+      "step": 810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460332618416412e-05,
+      "loss": 0.5489,
+      "step": 811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458622804895464e-05,
+      "loss": 0.511,
+      "step": 812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456910362433327e-05,
+      "loss": 0.4961,
+      "step": 813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455195291505954e-05,
+      "loss": 0.5626,
+      "step": 814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453477592590037e-05,
+      "loss": 0.5159,
+      "step": 815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451757266162995e-05,
+      "loss": 0.4839,
+      "step": 816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450034312702976e-05,
+      "loss": 0.5397,
+      "step": 817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944830873268886e-05,
+      "loss": 0.5234,
+      "step": 818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446580526600258e-05,
+      "loss": 0.573,
+      "step": 819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944484969491751e-05,
+      "loss": 0.5238,
+      "step": 820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443116238121687e-05,
+      "loss": 0.5498,
+      "step": 821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441380156694585e-05,
+      "loss": 0.5848,
+      "step": 822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943964145111873e-05,
+      "loss": 0.4682,
+      "step": 823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437900121877388e-05,
+      "loss": 0.5607,
+      "step": 824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943615616945454e-05,
+      "loss": 0.5315,
+      "step": 825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434409594334904e-05,
+      "loss": 0.4707,
+      "step": 826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432660397003924e-05,
+      "loss": 0.4917,
+      "step": 827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430908577947777e-05,
+      "loss": 0.5593,
+      "step": 828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9429154137653365e-05,
+      "loss": 0.5138,
+      "step": 829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427397076608318e-05,
+      "loss": 0.5154,
+      "step": 830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425637395300993e-05,
+      "loss": 0.5621,
+      "step": 831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423875094220483e-05,
+      "loss": 0.5491,
+      "step": 832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94221101738566e-05,
+      "loss": 0.5089,
+      "step": 833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420342634699893e-05,
+      "loss": 0.5244,
+      "step": 834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941857247724163e-05,
+      "loss": 0.5025,
+      "step": 835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416799701973812e-05,
+      "loss": 0.5551,
+      "step": 836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415024309389164e-05,
+      "loss": 0.5718,
+      "step": 837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413246299981148e-05,
+      "loss": 0.5358,
+      "step": 838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411465674243938e-05,
+      "loss": 0.6045,
+      "step": 839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409682432672445e-05,
+      "loss": 0.5014,
+      "step": 840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940789657576231e-05,
+      "loss": 0.5564,
+      "step": 841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406108104009892e-05,
+      "loss": 0.5009,
+      "step": 842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404317017912283e-05,
+      "loss": 0.4822,
+      "step": 843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402523317967294e-05,
+      "loss": 0.4948,
+      "step": 844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400727004673476e-05,
+      "loss": 0.5265,
+      "step": 845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9398928078530096e-05,
+      "loss": 0.5651,
+      "step": 846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939712654003715e-05,
+      "loss": 0.5407,
+      "step": 847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9395322389695357e-05,
+      "loss": 0.5411,
+      "step": 848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939351562800617e-05,
+      "loss": 0.5365,
+      "step": 849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9391706255471753e-05,
+      "loss": 0.5602,
+      "step": 850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389894272595015e-05,
+      "loss": 0.448,
+      "step": 851
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388079679879578e-05,
+      "loss": 0.5068,
+      "step": 852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938626247782979e-05,
+      "loss": 0.4859,
+      "step": 853
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384442666950726e-05,
+      "loss": 0.4989,
+      "step": 854
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938262024774819e-05,
+      "loss": 0.4988,
+      "step": 855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380795220728704e-05,
+      "loss": 0.5109,
+      "step": 856
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378967586399525e-05,
+      "loss": 0.4486,
+      "step": 857
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377137345268616e-05,
+      "loss": 0.5148,
+      "step": 858
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375304497844687e-05,
+      "loss": 0.4984,
+      "step": 859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373469044637158e-05,
+      "loss": 0.556,
+      "step": 860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371630986156176e-05,
+      "loss": 0.5321,
+      "step": 861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369790322912617e-05,
+      "loss": 0.5504,
+      "step": 862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367947055418072e-05,
+      "loss": 0.567,
+      "step": 863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936610118418486e-05,
+      "loss": 0.5813,
+      "step": 864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9364252709726036e-05,
+      "loss": 0.5467,
+      "step": 865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362401632555357e-05,
+      "loss": 0.4786,
+      "step": 866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9360547953187315e-05,
+      "loss": 0.572,
+      "step": 867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358691672137124e-05,
+      "loss": 0.5916,
+      "step": 868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356832789920722e-05,
+      "loss": 0.5657,
+      "step": 869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935497130705477e-05,
+      "loss": 0.4741,
+      "step": 870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9353107224056652e-05,
+      "loss": 0.5018,
+      "step": 871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351240541444468e-05,
+      "loss": 0.5068,
+      "step": 872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934937125973705e-05,
+      "loss": 0.4635,
+      "step": 873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934749937945395e-05,
+      "loss": 0.4849,
+      "step": 874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345624901115437e-05,
+      "loss": 0.5122,
+      "step": 875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934374782524251e-05,
+      "loss": 0.5154,
+      "step": 876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934186815235688e-05,
+      "loss": 0.5174,
+      "step": 877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9339985882980992e-05,
+      "loss": 0.4736,
+      "step": 878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338101017638e-05,
+      "loss": 0.5162,
+      "step": 879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336213556851787e-05,
+      "loss": 0.5861,
+      "step": 880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9334323501146964e-05,
+      "loss": 0.6289,
+      "step": 881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9332430851048846e-05,
+      "loss": 0.5059,
+      "step": 882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330535607083482e-05,
+      "loss": 0.5871,
+      "step": 883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9328637769777638e-05,
+      "loss": 0.5033,
+      "step": 884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93267373396588e-05,
+      "loss": 0.5094,
+      "step": 885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324834317255182e-05,
+      "loss": 0.5001,
+      "step": 886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322928703095704e-05,
+      "loss": 0.5122,
+      "step": 887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9321020497710022e-05,
+      "loss": 0.5213,
+      "step": 888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319109701628494e-05,
+      "loss": 0.5917,
+      "step": 889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317196315382226e-05,
+      "loss": 0.5549,
+      "step": 890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9315280339503013e-05,
+      "loss": 0.5601,
+      "step": 891
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313361774523387e-05,
+      "loss": 0.5214,
+      "step": 892
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311440620976597e-05,
+      "loss": 0.4979,
+      "step": 893
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930951687939661e-05,
+      "loss": 0.5301,
+      "step": 894
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930759055031812e-05,
+      "loss": 0.5596,
+      "step": 895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9305661634276523e-05,
+      "loss": 0.5234,
+      "step": 896
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303730131807947e-05,
+      "loss": 0.5434,
+      "step": 897
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301796043449244e-05,
+      "loss": 0.5215,
+      "step": 898
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299859369737965e-05,
+      "loss": 0.5278,
+      "step": 899
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92979201112124e-05,
+      "loss": 0.5066,
+      "step": 900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295978268411547e-05,
+      "loss": 0.5373,
+      "step": 901
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929403384187512e-05,
+      "loss": 0.4611,
+      "step": 902
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929208683214356e-05,
+      "loss": 0.6304,
+      "step": 903
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929013723975802e-05,
+      "loss": 0.4647,
+      "step": 904
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928818506526037e-05,
+      "loss": 0.4928,
+      "step": 905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286230309193206e-05,
+      "loss": 0.4641,
+      "step": 906
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284272972099826e-05,
+      "loss": 0.5354,
+      "step": 907
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9282313054524265e-05,
+      "loss": 0.5161,
+      "step": 908
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9280350557011257e-05,
+      "loss": 0.5571,
+      "step": 909
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927838548010626e-05,
+      "loss": 0.482,
+      "step": 910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9276417824355457e-05,
+      "loss": 0.5356,
+      "step": 911
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9274447590305736e-05,
+      "loss": 0.5449,
+      "step": 912
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927247477850471e-05,
+      "loss": 0.4675,
+      "step": 913
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92704993895007e-05,
+      "loss": 0.6092,
+      "step": 914
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926852142384275e-05,
+      "loss": 0.4996,
+      "step": 915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266540882080615e-05,
+      "loss": 0.4964,
+      "step": 916
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264557764764778e-05,
+      "loss": 0.4809,
+      "step": 917
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926257207244642e-05,
+      "loss": 0.5322,
+      "step": 918
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926058380567745e-05,
+      "loss": 0.5079,
+      "step": 919
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258592965010492e-05,
+      "loss": 0.6026,
+      "step": 920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9256599550998878e-05,
+      "loss": 0.4691,
+      "step": 921
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254603564196664e-05,
+      "loss": 0.535,
+      "step": 922
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9252605005158613e-05,
+      "loss": 0.5408,
+      "step": 923
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250603874440213e-05,
+      "loss": 0.5404,
+      "step": 924
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924860017259765e-05,
+      "loss": 0.4578,
+      "step": 925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924659390018785e-05,
+      "loss": 0.4886,
+      "step": 926
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9244585057768427e-05,
+      "loss": 0.6105,
+      "step": 927
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9242573645897724e-05,
+      "loss": 0.4648,
+      "step": 928
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240559665134798e-05,
+      "loss": 0.4999,
+      "step": 929
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238543116039414e-05,
+      "loss": 0.4875,
+      "step": 930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923652399917206e-05,
+      "loss": 0.4684,
+      "step": 931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9234502315093924e-05,
+      "loss": 0.4723,
+      "step": 932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232478064366917e-05,
+      "loss": 0.5311,
+      "step": 933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230451247553667e-05,
+      "loss": 0.5656,
+      "step": 934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922842186521751e-05,
+      "loss": 0.5575,
+      "step": 935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226389917922486e-05,
+      "loss": 0.4834,
+      "step": 936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922435540623336e-05,
+      "loss": 0.5616,
+      "step": 937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222318330715618e-05,
+      "loss": 0.48,
+      "step": 938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922027869193543e-05,
+      "loss": 0.482,
+      "step": 939
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921823649045971e-05,
+      "loss": 0.5482,
+      "step": 940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921619172685606e-05,
+      "loss": 0.5069,
+      "step": 941
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214144401692812e-05,
+      "loss": 0.5644,
+      "step": 942
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9212094515538996e-05,
+      "loss": 0.4811,
+      "step": 943
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921004206896436e-05,
+      "loss": 0.537,
+      "step": 944
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920798706253937e-05,
+      "loss": 0.5211,
+      "step": 945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205929496835193e-05,
+      "loss": 0.5457,
+      "step": 946
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9203869372423707e-05,
+      "loss": 0.5471,
+      "step": 947
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920180668987751e-05,
+      "loss": 0.469,
+      "step": 948
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199741449769908e-05,
+      "loss": 0.502,
+      "step": 949
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9197673652674912e-05,
+      "loss": 0.4468,
+      "step": 950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9195603299167254e-05,
+      "loss": 0.5381,
+      "step": 951
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9193530389822364e-05,
+      "loss": 0.5362,
+      "step": 952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9191454925216394e-05,
+      "loss": 0.4987,
+      "step": 953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189376905926197e-05,
+      "loss": 0.5604,
+      "step": 954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187296332529347e-05,
+      "loss": 0.4854,
+      "step": 955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9185213205604108e-05,
+      "loss": 0.5073,
+      "step": 956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183127525729484e-05,
+      "loss": 0.5648,
+      "step": 957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181039293485158e-05,
+      "loss": 0.557,
+      "step": 958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917894850945154e-05,
+      "loss": 0.4702,
+      "step": 959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176855174209747e-05,
+      "loss": 0.5071,
+      "step": 960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174759288341605e-05,
+      "loss": 0.5879,
+      "step": 961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917266085242964e-05,
+      "loss": 0.4748,
+      "step": 962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170559867057098e-05,
+      "loss": 0.5402,
+      "step": 963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168456332807926e-05,
+      "loss": 0.5772,
+      "step": 964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916635025026679e-05,
+      "loss": 0.5723,
+      "step": 965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164241620019055e-05,
+      "loss": 0.5746,
+      "step": 966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916213044265079e-05,
+      "loss": 0.5683,
+      "step": 967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160016718748788e-05,
+      "loss": 0.5143,
+      "step": 968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915790044890053e-05,
+      "loss": 0.5353,
+      "step": 969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915578163369422e-05,
+      "loss": 0.5662,
+      "step": 970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153660273718765e-05,
+      "loss": 0.595,
+      "step": 971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151536369563778e-05,
+      "loss": 0.5051,
+      "step": 972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9149409921819575e-05,
+      "loss": 0.4738,
+      "step": 973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147280931077187e-05,
+      "loss": 0.4911,
+      "step": 974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145149397928352e-05,
+      "loss": 0.5411,
+      "step": 975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143015322965505e-05,
+      "loss": 0.5326,
+      "step": 976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140878706781792e-05,
+      "loss": 0.4992,
+      "step": 977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138739549971076e-05,
+      "loss": 0.4756,
+      "step": 978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136597853127906e-05,
+      "loss": 0.4857,
+      "step": 979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134453616847555e-05,
+      "loss": 0.5233,
+      "step": 980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913230684172599e-05,
+      "loss": 0.5065,
+      "step": 981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9130157528359895e-05,
+      "loss": 0.4943,
+      "step": 982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128005677346645e-05,
+      "loss": 0.5523,
+      "step": 983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125851289284335e-05,
+      "loss": 0.5209,
+      "step": 984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123694364771752e-05,
+      "loss": 0.4912,
+      "step": 985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121534904408397e-05,
+      "loss": 0.5476,
+      "step": 986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911937290879448e-05,
+      "loss": 0.4804,
+      "step": 987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911720837853089e-05,
+      "loss": 0.4969,
+      "step": 988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911504131421926e-05,
+      "loss": 0.6182,
+      "step": 989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9112871716461895e-05,
+      "loss": 0.5069,
+      "step": 990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911069958586182e-05,
+      "loss": 0.5705,
+      "step": 991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9108524923022754e-05,
+      "loss": 0.4484,
+      "step": 992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9106347728549134e-05,
+      "loss": 0.5292,
+      "step": 993
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9104168003046088e-05,
+      "loss": 0.5364,
+      "step": 994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101985747119445e-05,
+      "loss": 0.5131,
+      "step": 995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099800961375754e-05,
+      "loss": 0.5336,
+      "step": 996
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909761364642225e-05,
+      "loss": 0.6229,
+      "step": 997
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095423802866887e-05,
+      "loss": 0.4158,
+      "step": 998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90932314313183e-05,
+      "loss": 0.516,
+      "step": 999
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091036532385848e-05,
+      "loss": 0.5716,
+      "step": 1000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088839106679584e-05,
+      "loss": 0.5168,
+      "step": 1001
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908663915481026e-05,
+      "loss": 0.5268,
+      "step": 1002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084436677389336e-05,
+      "loss": 0.5588,
+      "step": 1003
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908223167502897e-05,
+      "loss": 0.5114,
+      "step": 1004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908002414834202e-05,
+      "loss": 0.4953,
+      "step": 1005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077814097942054e-05,
+      "loss": 0.4546,
+      "step": 1006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9075601524443333e-05,
+      "loss": 0.4954,
+      "step": 1007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907338642846082e-05,
+      "loss": 0.5192,
+      "step": 1008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907116881061019e-05,
+      "loss": 0.4993,
+      "step": 1009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90689486715078e-05,
+      "loss": 0.5282,
+      "step": 1010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066726011770725e-05,
+      "loss": 0.5756,
+      "step": 1011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9064500832016734e-05,
+      "loss": 0.5527,
+      "step": 1012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9062273132864292e-05,
+      "loss": 0.4198,
+      "step": 1013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.906004291493257e-05,
+      "loss": 0.4927,
+      "step": 1014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057810178841442e-05,
+      "loss": 0.547,
+      "step": 1015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905557492521147e-05,
+      "loss": 0.5365,
+      "step": 1016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053337154663927e-05,
+      "loss": 0.4973,
+      "step": 1017
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905109686782078e-05,
+      "loss": 0.4522,
+      "step": 1018
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90488540653047e-05,
+      "loss": 0.5438,
+      "step": 1019
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9046608747739053e-05,
+      "loss": 0.5442,
+      "step": 1020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044360915747903e-05,
+      "loss": 0.4572,
+      "step": 1021
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9042110569956018e-05,
+      "loss": 0.5354,
+      "step": 1022
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039857710988857e-05,
+      "loss": 0.5243,
+      "step": 1023
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903760233947259e-05,
+      "loss": 0.5246,
+      "step": 1024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035344456034076e-05,
+      "loss": 0.4476,
+      "step": 1025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903308406130087e-05,
+      "loss": 0.4763,
+      "step": 1026
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903082115590123e-05,
+      "loss": 0.561,
+      "step": 1027
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9028555740464113e-05,
+      "loss": 0.5907,
+      "step": 1028
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026287815619167e-05,
+      "loss": 0.5224,
+      "step": 1029
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024017381996747e-05,
+      "loss": 0.4696,
+      "step": 1030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90217444402279e-05,
+      "loss": 0.5469,
+      "step": 1031
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019468990944366e-05,
+      "loss": 0.6443,
+      "step": 1032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9017191034778583e-05,
+      "loss": 0.5293,
+      "step": 1033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90149105723637e-05,
+      "loss": 0.5727,
+      "step": 1034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901262760433354e-05,
+      "loss": 0.485,
+      "step": 1035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010342131322645e-05,
+      "loss": 0.4933,
+      "step": 1036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9008054153966235e-05,
+      "loss": 0.4979,
+      "step": 1037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005763672900235e-05,
+      "loss": 0.543,
+      "step": 1038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003470688761263e-05,
+      "loss": 0.4774,
+      "step": 1039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001175202186638e-05,
+      "loss": 0.5274,
+      "step": 1040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899887721381436e-05,
+      "loss": 0.5414,
+      "step": 1041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899657672428315e-05,
+      "loss": 0.4999,
+      "step": 1042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8994273734232396e-05,
+      "loss": 0.5082,
+      "step": 1043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991968244302198e-05,
+      "loss": 0.4947,
+      "step": 1044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898966025513335e-05,
+      "loss": 0.4833,
+      "step": 1045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898734976736733e-05,
+      "loss": 0.5,
+      "step": 1046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985036781646327e-05,
+      "loss": 0.4858,
+      "step": 1047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898272129861321e-05,
+      "loss": 0.4773,
+      "step": 1048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980403318911547e-05,
+      "loss": 0.5683,
+      "step": 1049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89780828431856e-05,
+      "loss": 0.4979,
+      "step": 1050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8975759872080325e-05,
+      "loss": 0.4678,
+      "step": 1051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8973434406241372e-05,
+      "loss": 0.56,
+      "step": 1052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971106446315085e-05,
+      "loss": 0.5287,
+      "step": 1053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89687759929485e-05,
+      "loss": 0.5313,
+      "step": 1054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896644304678934e-05,
+      "loss": 0.5076,
+      "step": 1055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896410760848604e-05,
+      "loss": 0.478,
+      "step": 1056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8961769678687697e-05,
+      "loss": 0.5339,
+      "step": 1057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895942925804413e-05,
+      "loss": 0.5535,
+      "step": 1058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957086347205842e-05,
+      "loss": 0.4988,
+      "step": 1059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8954740946824012e-05,
+      "loss": 0.4992,
+      "step": 1060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8952393057550535e-05,
+      "loss": 0.5923,
+      "step": 1061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950042680037977e-05,
+      "loss": 0.4946,
+      "step": 1062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894768981493961e-05,
+      "loss": 0.4904,
+      "step": 1063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894533446290939e-05,
+      "loss": 0.4879,
+      "step": 1064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894297662460197e-05,
+      "loss": 0.4585,
+      "step": 1065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8940616300672685e-05,
+      "loss": 0.5899,
+      "step": 1066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893825349177757e-05,
+      "loss": 0.5212,
+      "step": 1067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893588819857335e-05,
+      "loss": 0.5132,
+      "step": 1068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933520421717432e-05,
+      "loss": 0.4663,
+      "step": 1069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931150161867917e-05,
+      "loss": 0.487,
+      "step": 1070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928777419683603e-05,
+      "loss": 0.5331,
+      "step": 1071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926402195823973e-05,
+      "loss": 0.584,
+      "step": 1072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924024490949196e-05,
+      "loss": 0.4961,
+      "step": 1073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892164430572014e-05,
+      "loss": 0.546,
+      "step": 1074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891926164079835e-05,
+      "loss": 0.53,
+      "step": 1075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8916876496846067e-05,
+      "loss": 0.4556,
+      "step": 1076
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914488874526228e-05,
+      "loss": 0.4879,
+      "step": 1077
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8912098774502444e-05,
+      "loss": 0.4875,
+      "step": 1078
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8909706197439026e-05,
+      "loss": 0.4887,
+      "step": 1079
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890731114400097e-05,
+      "loss": 0.4848,
+      "step": 1080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890491361485396e-05,
+      "loss": 0.5555,
+      "step": 1081
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8902513610664364e-05,
+      "loss": 0.4804,
+      "step": 1082
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900111132099246e-05,
+      "loss": 0.4808,
+      "step": 1083
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897706179826355e-05,
+      "loss": 0.4903,
+      "step": 1084
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895298754514123e-05,
+      "loss": 0.5279,
+      "step": 1085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889288885683168e-05,
+      "loss": 0.478,
+      "step": 1086
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8890476487448826e-05,
+      "loss": 0.5141,
+      "step": 1087
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888806164703606e-05,
+      "loss": 0.4956,
+      "step": 1088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885644336264573e-05,
+      "loss": 0.5508,
+      "step": 1089
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883224555806226e-05,
+      "loss": 0.4902,
+      "step": 1090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880802306333585e-05,
+      "loss": 0.531,
+      "step": 1091
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8878377588519883e-05,
+      "loss": 0.5502,
+      "step": 1092
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875950403039056e-05,
+      "loss": 0.5814,
+      "step": 1093
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873520750565716e-05,
+      "loss": 0.5003,
+      "step": 1094
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8871088631775167e-05,
+      "loss": 0.5018,
+      "step": 1095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886865404734339e-05,
+      "loss": 0.5438,
+      "step": 1096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866216997947062e-05,
+      "loss": 0.5063,
+      "step": 1097
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863777484263536e-05,
+      "loss": 0.5216,
+      "step": 1098
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886133550697085e-05,
+      "loss": 0.5444,
+      "step": 1099
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8858891066747737e-05,
+      "loss": 0.5392,
+      "step": 1100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8856444164273605e-05,
+      "loss": 0.4679,
+      "step": 1101
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8853994800228546e-05,
+      "loss": 0.5228,
+      "step": 1102
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8851542975293345e-05,
+      "loss": 0.531,
+      "step": 1103
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884908869014946e-05,
+      "loss": 0.4918,
+      "step": 1104
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884663194547904e-05,
+      "loss": 0.5263,
+      "step": 1105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884417274196492e-05,
+      "loss": 0.5735,
+      "step": 1106
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884171108029061e-05,
+      "loss": 0.5114,
+      "step": 1107
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88392469611403e-05,
+      "loss": 0.5451,
+      "step": 1108
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8836780385198883e-05,
+      "loss": 0.5186,
+      "step": 1109
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8834311353151916e-05,
+      "loss": 0.5067,
+      "step": 1110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8831839865685648e-05,
+      "loss": 0.4654,
+      "step": 1111
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829365923487e-05,
+      "loss": 0.5229,
+      "step": 1112
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8826889527243587e-05,
+      "loss": 0.4427,
+      "step": 1113
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824410677643705e-05,
+      "loss": 0.523,
+      "step": 1114
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882192937537632e-05,
+      "loss": 0.5372,
+      "step": 1115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88194456211311e-05,
+      "loss": 0.5428,
+      "step": 1116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816959415598373e-05,
+      "loss": 0.5454,
+      "step": 1117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881447075946916e-05,
+      "loss": 0.4946,
+      "step": 1118
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8811979653435166e-05,
+      "loss": 0.4584,
+      "step": 1119
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809486098188766e-05,
+      "loss": 0.534,
+      "step": 1120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8806990094423026e-05,
+      "loss": 0.579,
+      "step": 1121
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8804491642831682e-05,
+      "loss": 0.4788,
+      "step": 1122
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801990744109164e-05,
+      "loss": 0.4675,
+      "step": 1123
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799487398950573e-05,
+      "loss": 0.5082,
+      "step": 1124
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879698160805169e-05,
+      "loss": 0.423,
+      "step": 1125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8794473372108977e-05,
+      "loss": 0.4922,
+      "step": 1126
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791962691819582e-05,
+      "loss": 0.4901,
+      "step": 1127
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8789449567881317e-05,
+      "loss": 0.6041,
+      "step": 1128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878693400099269e-05,
+      "loss": 0.554,
+      "step": 1129
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784415991852878e-05,
+      "loss": 0.5379,
+      "step": 1130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8781895541161737e-05,
+      "loss": 0.4707,
+      "step": 1131
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779372649619815e-05,
+      "loss": 0.4729,
+      "step": 1132
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8776847317928316e-05,
+      "loss": 0.4662,
+      "step": 1133
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8774319546789135e-05,
+      "loss": 0.5024,
+      "step": 1134
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877178933690485e-05,
+      "loss": 0.5796,
+      "step": 1135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769256688978706e-05,
+      "loss": 0.4942,
+      "step": 1136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876672160371463e-05,
+      "loss": 0.4901,
+      "step": 1137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876418408181723e-05,
+      "loss": 0.5295,
+      "step": 1138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8761644123991787e-05,
+      "loss": 0.5172,
+      "step": 1139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8759101730944257e-05,
+      "loss": 0.5574,
+      "step": 1140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756556903381278e-05,
+      "loss": 0.4434,
+      "step": 1141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875400964201016e-05,
+      "loss": 0.5808,
+      "step": 1142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8751459947538898e-05,
+      "loss": 0.5425,
+      "step": 1143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874890782067615e-05,
+      "loss": 0.5221,
+      "step": 1144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874635326213126e-05,
+      "loss": 0.4987,
+      "step": 1145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874379627261425e-05,
+      "loss": 0.5145,
+      "step": 1146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874123685283581e-05,
+      "loss": 0.5295,
+      "step": 1147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87386750035073e-05,
+      "loss": 0.464,
+      "step": 1148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736110725340777e-05,
+      "loss": 0.4597,
+      "step": 1149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733544019048947e-05,
+      "loss": 0.5644,
+      "step": 1150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730974885345216e-05,
+      "loss": 0.5258,
+      "step": 1151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872840332494364e-05,
+      "loss": 0.5059,
+      "step": 1152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8725829338558973e-05,
+      "loss": 0.5202,
+      "step": 1153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8723252926906622e-05,
+      "loss": 0.5394,
+      "step": 1154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720674090702684e-05,
+      "loss": 0.545,
+      "step": 1155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871809283066392e-05,
+      "loss": 0.5828,
+      "step": 1156
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871550914750777e-05,
+      "loss": 0.5043,
+      "step": 1157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871292304195235e-05,
+      "loss": 0.5106,
+      "step": 1158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871033451471644e-05,
+      "loss": 0.5364,
+      "step": 1159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8707743566519503e-05,
+      "loss": 0.5388,
+      "step": 1160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8705150198081665e-05,
+      "loss": 0.5247,
+      "step": 1161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8702554410123734e-05,
+      "loss": 0.6227,
+      "step": 1162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8699956203367183e-05,
+      "loss": 0.4689,
+      "step": 1163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697355578534167e-05,
+      "loss": 0.5367,
+      "step": 1164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86947525363475e-05,
+      "loss": 0.4954,
+      "step": 1165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8692147077530683e-05,
+      "loss": 0.5533,
+      "step": 1166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8689539202807868e-05,
+      "loss": 0.5005,
+      "step": 1167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8686928912903903e-05,
+      "loss": 0.5191,
+      "step": 1168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684316208544286e-05,
+      "loss": 0.4733,
+      "step": 1169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86817010904552e-05,
+      "loss": 0.5026,
+      "step": 1170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8679083559363498e-05,
+      "loss": 0.4803,
+      "step": 1171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8676463615996693e-05,
+      "loss": 0.5422,
+      "step": 1172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8673841261082977e-05,
+      "loss": 0.511,
+      "step": 1173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867121649535121e-05,
+      "loss": 0.4764,
+      "step": 1174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8668589319530924e-05,
+      "loss": 0.5827,
+      "step": 1175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866595973435232e-05,
+      "loss": 0.5929,
+      "step": 1176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8663327740546268e-05,
+      "loss": 0.5081,
+      "step": 1177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8660693338844306e-05,
+      "loss": 0.5696,
+      "step": 1178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8658056529978642e-05,
+      "loss": 0.5516,
+      "step": 1179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655417314682156e-05,
+      "loss": 0.4611,
+      "step": 1180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8652775693688393e-05,
+      "loss": 0.477,
+      "step": 1181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650131667731574e-05,
+      "loss": 0.5161,
+      "step": 1182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647485237546578e-05,
+      "loss": 0.5314,
+      "step": 1183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8644836403868955e-05,
+      "loss": 0.4934,
+      "step": 1184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864218516743493e-05,
+      "loss": 0.5223,
+      "step": 1185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8639531528981384e-05,
+      "loss": 0.4868,
+      "step": 1186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8636875489245877e-05,
+      "loss": 0.5688,
+      "step": 1187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8634217048966638e-05,
+      "loss": 0.5203,
+      "step": 1188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631556208882546e-05,
+      "loss": 0.3975,
+      "step": 1189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8628892969733166e-05,
+      "loss": 0.4859,
+      "step": 1190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8626227332258718e-05,
+      "loss": 0.5099,
+      "step": 1191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8623559297200094e-05,
+      "loss": 0.5256,
+      "step": 1192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862088886529885e-05,
+      "loss": 0.4681,
+      "step": 1193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861821603729721e-05,
+      "loss": 0.569,
+      "step": 1194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615540813938063e-05,
+      "loss": 0.4802,
+      "step": 1195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8612863195964963e-05,
+      "loss": 0.5341,
+      "step": 1196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610183184122134e-05,
+      "loss": 0.5307,
+      "step": 1197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8607500779154452e-05,
+      "loss": 0.5503,
+      "step": 1198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8604815981807483e-05,
+      "loss": 0.5696,
+      "step": 1199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860212879282743e-05,
+      "loss": 0.5262,
+      "step": 1200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859943921296118e-05,
+      "loss": 0.4929,
+      "step": 1201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8596747242956276e-05,
+      "loss": 0.4835,
+      "step": 1202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8594052883560925e-05,
+      "loss": 0.4996,
+      "step": 1203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591356135524007e-05,
+      "loss": 0.52,
+      "step": 1204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588656999595053e-05,
+      "loss": 0.549,
+      "step": 1205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8585955476524268e-05,
+      "loss": 0.5353,
+      "step": 1206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858325156706252e-05,
+      "loss": 0.4541,
+      "step": 1207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8580545271961323e-05,
+      "loss": 0.5539,
+      "step": 1208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577836591972884e-05,
+      "loss": 0.5206,
+      "step": 1209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575125527850048e-05,
+      "loss": 0.5576,
+      "step": 1210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8572412080346335e-05,
+      "loss": 0.6064,
+      "step": 1211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8569696250215923e-05,
+      "loss": 0.5074,
+      "step": 1212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856697803821365e-05,
+      "loss": 0.5149,
+      "step": 1213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564257445095028e-05,
+      "loss": 0.5765,
+      "step": 1214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856153447161621e-05,
+      "loss": 0.5468,
+      "step": 1215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8558809118534032e-05,
+      "loss": 0.4975,
+      "step": 1216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8556081386605974e-05,
+      "loss": 0.5321,
+      "step": 1217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8553351276590193e-05,
+      "loss": 0.5074,
+      "step": 1218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8550618789245497e-05,
+      "loss": 0.5362,
+      "step": 1219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8547883925331354e-05,
+      "loss": 0.5121,
+      "step": 1220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8545146685607894e-05,
+      "loss": 0.4621,
+      "step": 1221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854240707083592e-05,
+      "loss": 0.5141,
+      "step": 1222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8539665081776866e-05,
+      "loss": 0.5305,
+      "step": 1223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853692071919286e-05,
+      "loss": 0.4904,
+      "step": 1224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534173983846663e-05,
+      "loss": 0.5532,
+      "step": 1225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853142487650171e-05,
+      "loss": 0.5019,
+      "step": 1226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8528673397922093e-05,
+      "loss": 0.4877,
+      "step": 1227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8525919548872563e-05,
+      "loss": 0.4875,
+      "step": 1228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8523163330118522e-05,
+      "loss": 0.4772,
+      "step": 1229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8520404742426036e-05,
+      "loss": 0.5227,
+      "step": 1230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517643786561837e-05,
+      "loss": 0.5203,
+      "step": 1231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8514880463293307e-05,
+      "loss": 0.5679,
+      "step": 1232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512114773388484e-05,
+      "loss": 0.5063,
+      "step": 1233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8509346717616066e-05,
+      "loss": 0.5236,
+      "step": 1234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506576296745414e-05,
+      "loss": 0.5418,
+      "step": 1235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8503803511546543e-05,
+      "loss": 0.4758,
+      "step": 1236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850102836279012e-05,
+      "loss": 0.5085,
+      "step": 1237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498250851247477e-05,
+      "loss": 0.5321,
+      "step": 1238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8495470977690596e-05,
+      "loss": 0.5598,
+      "step": 1239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849268874289212e-05,
+      "loss": 0.5348,
+      "step": 1240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848990414762535e-05,
+      "loss": 0.4697,
+      "step": 1241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848711719266423e-05,
+      "loss": 0.495,
+      "step": 1242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848432787878338e-05,
+      "loss": 0.5091,
+      "step": 1243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8481536206758063e-05,
+      "loss": 0.4604,
+      "step": 1244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8478742177364194e-05,
+      "loss": 0.6007,
+      "step": 1245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8475945791378356e-05,
+      "loss": 0.477,
+      "step": 1246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8473147049577777e-05,
+      "loss": 0.4859,
+      "step": 1247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847034595274034e-05,
+      "loss": 0.4915,
+      "step": 1248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8467542501644594e-05,
+      "loss": 0.4332,
+      "step": 1249
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846473669706972e-05,
+      "loss": 0.4642,
+      "step": 1250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8461928539795578e-05,
+      "loss": 0.5011,
+      "step": 1251
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459118030602667e-05,
+      "loss": 0.5856,
+      "step": 1252
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8456305170272144e-05,
+      "loss": 0.4477,
+      "step": 1253
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8453489959585817e-05,
+      "loss": 0.5811,
+      "step": 1254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845067239932615e-05,
+      "loss": 0.5152,
+      "step": 1255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8447852490276262e-05,
+      "loss": 0.4934,
+      "step": 1256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8445030233219916e-05,
+      "loss": 0.5555,
+      "step": 1257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844220562894154e-05,
+      "loss": 0.525,
+      "step": 1258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8439378678226204e-05,
+      "loss": 0.4237,
+      "step": 1259
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8436549381859633e-05,
+      "loss": 0.4707,
+      "step": 1260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843371774062821e-05,
+      "loss": 0.4542,
+      "step": 1261
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843088375531896e-05,
+      "loss": 0.5791,
+      "step": 1262
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842804742671957e-05,
+      "loss": 0.4971,
+      "step": 1263
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8425208755618366e-05,
+      "loss": 0.5531,
+      "step": 1264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8422367742804335e-05,
+      "loss": 0.4996,
+      "step": 1265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8419524389067113e-05,
+      "loss": 0.4902,
+      "step": 1266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416678695196985e-05,
+      "loss": 0.5023,
+      "step": 1267
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841383066198488e-05,
+      "loss": 0.5235,
+      "step": 1268
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841098029022239e-05,
+      "loss": 0.5345,
+      "step": 1269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408127580701755e-05,
+      "loss": 0.5383,
+      "step": 1270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840527253421585e-05,
+      "loss": 0.4794,
+      "step": 1271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8402415151558216e-05,
+      "loss": 0.5881,
+      "step": 1272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839955543352304e-05,
+      "loss": 0.5202,
+      "step": 1273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839669338090515e-05,
+      "loss": 0.5925,
+      "step": 1274
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839382899450003e-05,
+      "loss": 0.463,
+      "step": 1275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390962275103813e-05,
+      "loss": 0.441,
+      "step": 1276
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838809322351328e-05,
+      "loss": 0.5235,
+      "step": 1277
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8385221840525856e-05,
+      "loss": 0.4879,
+      "step": 1278
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8382348126939617e-05,
+      "loss": 0.5516,
+      "step": 1279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.837947208355328e-05,
+      "loss": 0.4837,
+      "step": 1280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8376593711166227e-05,
+      "loss": 0.4777,
+      "step": 1281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373713010578475e-05,
+      "loss": 0.4836,
+      "step": 1282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837082998259068e-05,
+      "loss": 0.5528,
+      "step": 1283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8367944628004164e-05,
+      "loss": 0.4832,
+      "step": 1284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365056947620882e-05,
+      "loss": 0.5218,
+      "step": 1285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362166942243437e-05,
+      "loss": 0.4793,
+      "step": 1286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8359274612675086e-05,
+      "loss": 0.5429,
+      "step": 1287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356379959719726e-05,
+      "loss": 0.587,
+      "step": 1288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353482984181897e-05,
+      "loss": 0.4382,
+      "step": 1289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350583686866786e-05,
+      "loss": 0.4637,
+      "step": 1290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347682068580237e-05,
+      "loss": 0.4974,
+      "step": 1291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834477813012872e-05,
+      "loss": 0.5562,
+      "step": 1292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834187187231936e-05,
+      "loss": 0.5555,
+      "step": 1293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833896329595993e-05,
+      "loss": 0.5043,
+      "step": 1294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833605240185884e-05,
+      "loss": 0.517,
+      "step": 1295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833313919082515e-05,
+      "loss": 0.5943,
+      "step": 1296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833022366366856e-05,
+      "loss": 0.5706,
+      "step": 1297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832730582119942e-05,
+      "loss": 0.418,
+      "step": 1298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8324385664228708e-05,
+      "loss": 0.5342,
+      "step": 1299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8321463193568065e-05,
+      "loss": 0.5741,
+      "step": 1300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831853841002976e-05,
+      "loss": 0.4977,
+      "step": 1301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8315611314426715e-05,
+      "loss": 0.5481,
+      "step": 1302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8312681907572492e-05,
+      "loss": 0.577,
+      "step": 1303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830975019028129e-05,
+      "loss": 0.5209,
+      "step": 1304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830681616336796e-05,
+      "loss": 0.5171,
+      "step": 1305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8303879827647977e-05,
+      "loss": 0.5215,
+      "step": 1306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8300941183937482e-05,
+      "loss": 0.4773,
+      "step": 1307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8298000233053238e-05,
+      "loss": 0.5309,
+      "step": 1308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8295056975812662e-05,
+      "loss": 0.5289,
+      "step": 1309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82921114130338e-05,
+      "loss": 0.4898,
+      "step": 1310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8289163545535354e-05,
+      "loss": 0.4986,
+      "step": 1311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828621337413665e-05,
+      "loss": 0.519,
+      "step": 1312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8283260899657667e-05,
+      "loss": 0.4925,
+      "step": 1313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8280306122919014e-05,
+      "loss": 0.4555,
+      "step": 1314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277349044741953e-05,
+      "loss": 0.5063,
+      "step": 1315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8274389665948368e-05,
+      "loss": 0.5015,
+      "step": 1316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271427987360804e-05,
+      "loss": 0.5252,
+      "step": 1317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8268464009802422e-05,
+      "loss": 0.5199,
+      "step": 1318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8265497734097043e-05,
+      "loss": 0.5213,
+      "step": 1319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8262529161069115e-05,
+      "loss": 0.4955,
+      "step": 1320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825955829154372e-05,
+      "loss": 0.5151,
+      "step": 1321
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8256585126346595e-05,
+      "loss": 0.527,
+      "step": 1322
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82536096663041e-05,
+      "loss": 0.5213,
+      "step": 1323
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8250631912243235e-05,
+      "loss": 0.5186,
+      "step": 1324
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824765186499165e-05,
+      "loss": 0.497,
+      "step": 1325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824466952537761e-05,
+      "loss": 0.4952,
+      "step": 1326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824168489423004e-05,
+      "loss": 0.4755,
+      "step": 1327
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8238697972378493e-05,
+      "loss": 0.5171,
+      "step": 1328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823570876065315e-05,
+      "loss": 0.5222,
+      "step": 1329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232717259884842e-05,
+      "loss": 0.5065,
+      "step": 1330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822972347090503e-05,
+      "loss": 0.5016,
+      "step": 1331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8226727394545806e-05,
+      "loss": 0.4993,
+      "step": 1332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822372903163991e-05,
+      "loss": 0.5531,
+      "step": 1333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8220728383020706e-05,
+      "loss": 0.5993,
+      "step": 1334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82177254495222e-05,
+      "loss": 0.5051,
+      "step": 1335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821472023197903e-05,
+      "loss": 0.5518,
+      "step": 1336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821171273122647e-05,
+      "loss": 0.5295,
+      "step": 1337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8208702948100433e-05,
+      "loss": 0.5497,
+      "step": 1338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820569088343745e-05,
+      "loss": 0.49,
+      "step": 1339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8202676538074714e-05,
+      "loss": 0.5269,
+      "step": 1340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8199659912850025e-05,
+      "loss": 0.494,
+      "step": 1341
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819664100860183e-05,
+      "loss": 0.4683,
+      "step": 1342
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8193619826169208e-05,
+      "loss": 0.4758,
+      "step": 1343
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8190596366391868e-05,
+      "loss": 0.4456,
+      "step": 1344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818757063011015e-05,
+      "loss": 0.4897,
+      "step": 1345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818454261816504e-05,
+      "loss": 0.4927,
+      "step": 1346
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181512331398146e-05,
+      "loss": 0.477,
+      "step": 1347
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8178479770651705e-05,
+      "loss": 0.577,
+      "step": 1348
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8175444936768593e-05,
+      "loss": 0.5344,
+      "step": 1349
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8172407830592314e-05,
+      "loss": 0.5287,
+      "step": 1350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8169368452967008e-05,
+      "loss": 0.557,
+      "step": 1351
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8166326804737438e-05,
+      "loss": 0.4405,
+      "step": 1352
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816328288674901e-05,
+      "loss": 0.5654,
+      "step": 1353
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816023669984775e-05,
+      "loss": 0.4411,
+      "step": 1354
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8157188244880322e-05,
+      "loss": 0.4926,
+      "step": 1355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8154137522694014e-05,
+      "loss": 0.519,
+      "step": 1356
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8151084534136746e-05,
+      "loss": 0.4541,
+      "step": 1357
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8148029280057077e-05,
+      "loss": 0.4964,
+      "step": 1358
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814497176130418e-05,
+      "loss": 0.5116,
+      "step": 1359
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814191197872787e-05,
+      "loss": 0.4399,
+      "step": 1360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8138849933178587e-05,
+      "loss": 0.5697,
+      "step": 1361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8135785625507396e-05,
+      "loss": 0.4993,
+      "step": 1362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8132719056565994e-05,
+      "loss": 0.5321,
+      "step": 1363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8129650227206715e-05,
+      "loss": 0.5218,
+      "step": 1364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8126579138282502e-05,
+      "loss": 0.4346,
+      "step": 1365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812350579064695e-05,
+      "loss": 0.5382,
+      "step": 1366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8120430185154256e-05,
+      "loss": 0.5294,
+      "step": 1367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8117352322659264e-05,
+      "loss": 0.5244,
+      "step": 1368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8114272204017444e-05,
+      "loss": 0.5249,
+      "step": 1369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8111189830084875e-05,
+      "loss": 0.4518,
+      "step": 1370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8108105201718285e-05,
+      "loss": 0.4671,
+      "step": 1371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8105018319775022e-05,
+      "loss": 0.5052,
+      "step": 1372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8101929185113046e-05,
+      "loss": 0.4916,
+      "step": 1373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809883779859097e-05,
+      "loss": 0.4581,
+      "step": 1374
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8095744161068e-05,
+      "loss": 0.5013,
+      "step": 1375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8092648273404002e-05,
+      "loss": 0.4265,
+      "step": 1376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808955013645944e-05,
+      "loss": 0.4482,
+      "step": 1377
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808644975109542e-05,
+      "loss": 0.4755,
+      "step": 1378
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8083347118173668e-05,
+      "loss": 0.4849,
+      "step": 1379
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8080242238556528e-05,
+      "loss": 0.4893,
+      "step": 1380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8077135113106975e-05,
+      "loss": 0.5391,
+      "step": 1381
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8074025742688614e-05,
+      "loss": 0.5071,
+      "step": 1382
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8070914128165655e-05,
+      "loss": 0.555,
+      "step": 1383
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067800270402956e-05,
+      "loss": 0.5193,
+      "step": 1384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806468417026598e-05,
+      "loss": 0.4581,
+      "step": 1385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061565828620817e-05,
+      "loss": 0.5072,
+      "step": 1386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8058445246334192e-05,
+      "loss": 0.4749,
+      "step": 1387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8055322424273436e-05,
+      "loss": 0.5465,
+      "step": 1388
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805219736330651e-05,
+      "loss": 0.4615,
+      "step": 1389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8049070064302002e-05,
+      "loss": 0.5213,
+      "step": 1390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8045940528129114e-05,
+      "loss": 0.4933,
+      "step": 1391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8042808755657674e-05,
+      "loss": 0.5063,
+      "step": 1392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8039674747758133e-05,
+      "loss": 0.4957,
+      "step": 1393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8036538505301557e-05,
+      "loss": 0.5337,
+      "step": 1394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8033400029159637e-05,
+      "loss": 0.4818,
+      "step": 1395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030259320204685e-05,
+      "loss": 0.5569,
+      "step": 1396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8027116379309637e-05,
+      "loss": 0.5156,
+      "step": 1397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8023971207348044e-05,
+      "loss": 0.4699,
+      "step": 1398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802082380519408e-05,
+      "loss": 0.4941,
+      "step": 1399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8017674173722535e-05,
+      "loss": 0.4862,
+      "step": 1400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8014522313808826e-05,
+      "loss": 0.5235,
+      "step": 1401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8011368226328976e-05,
+      "loss": 0.6012,
+      "step": 1402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8008211912159645e-05,
+      "loss": 0.4755,
+      "step": 1403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8005053372178098e-05,
+      "loss": 0.496,
+      "step": 1404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8001892607262226e-05,
+      "loss": 0.5841,
+      "step": 1405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7998729618290535e-05,
+      "loss": 0.4439,
+      "step": 1406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799556440614215e-05,
+      "loss": 0.4336,
+      "step": 1407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7992396971696813e-05,
+      "loss": 0.512,
+      "step": 1408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7989227315834887e-05,
+      "loss": 0.4658,
+      "step": 1409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798605543943735e-05,
+      "loss": 0.4786,
+      "step": 1410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798288134338579e-05,
+      "loss": 0.4594,
+      "step": 1411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797970502856243e-05,
+      "loss": 0.4492,
+      "step": 1412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797652649585009e-05,
+      "loss": 0.4755,
+      "step": 1413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7973345746132224e-05,
+      "loss": 0.5195,
+      "step": 1414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970162780292882e-05,
+      "loss": 0.4889,
+      "step": 1415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796697759921675e-05,
+      "loss": 0.5389,
+      "step": 1416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796379020378912e-05,
+      "loss": 0.5061,
+      "step": 1417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79606005948959e-05,
+      "loss": 0.4823,
+      "step": 1418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795740877342361e-05,
+      "loss": 0.5433,
+      "step": 1419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7954214740259396e-05,
+      "loss": 0.4821,
+      "step": 1420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7951018496291004e-05,
+      "loss": 0.5307,
+      "step": 1421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794782004240681e-05,
+      "loss": 0.5439,
+      "step": 1422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7944619379495788e-05,
+      "loss": 0.4671,
+      "step": 1423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7941416508447537e-05,
+      "loss": 0.5314,
+      "step": 1424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793821143015227e-05,
+      "loss": 0.5303,
+      "step": 1425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935004145500803e-05,
+      "loss": 0.5474,
+      "step": 1426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931794655384584e-05,
+      "loss": 0.5193,
+      "step": 1427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7928582960695655e-05,
+      "loss": 0.588,
+      "step": 1428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7925369062326678e-05,
+      "loss": 0.5153,
+      "step": 1429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7922152961170933e-05,
+      "loss": 0.4799,
+      "step": 1430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79189346581223e-05,
+      "loss": 0.5099,
+      "step": 1431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7915714154075283e-05,
+      "loss": 0.535,
+      "step": 1432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7912491449924993e-05,
+      "loss": 0.4801,
+      "step": 1433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790926654656715e-05,
+      "loss": 0.4716,
+      "step": 1434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7906039444898087e-05,
+      "loss": 0.5306,
+      "step": 1435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790281014581475e-05,
+      "loss": 0.4962,
+      "step": 1436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7899578650214695e-05,
+      "loss": 0.5093,
+      "step": 1437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7896344958996087e-05,
+      "loss": 0.4779,
+      "step": 1438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7893109073057704e-05,
+      "loss": 0.4853,
+      "step": 1439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788987099329893e-05,
+      "loss": 0.5251,
+      "step": 1440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788663072061976e-05,
+      "loss": 0.5259,
+      "step": 1441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7883388255920803e-05,
+      "loss": 0.4536,
+      "step": 1442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7880143600103266e-05,
+      "loss": 0.527,
+      "step": 1443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787689675406898e-05,
+      "loss": 0.5771,
+      "step": 1444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7873647718720378e-05,
+      "loss": 0.4971,
+      "step": 1445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7870396494960494e-05,
+      "loss": 0.4655,
+      "step": 1446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7867143083692986e-05,
+      "loss": 0.5171,
+      "step": 1447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78638874858221e-05,
+      "loss": 0.5209,
+      "step": 1448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7860629702252715e-05,
+      "loss": 0.4651,
+      "step": 1449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7857369733890292e-05,
+      "loss": 0.4886,
+      "step": 1450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785410758164091e-05,
+      "loss": 0.5833,
+      "step": 1451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7850843246411266e-05,
+      "loss": 0.5109,
+      "step": 1452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7847576729108644e-05,
+      "loss": 0.4768,
+      "step": 1453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7844308030640946e-05,
+      "loss": 0.4342,
+      "step": 1454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784103715191668e-05,
+      "loss": 0.5803,
+      "step": 1455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7837764093844954e-05,
+      "loss": 0.4522,
+      "step": 1456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783448885733549e-05,
+      "loss": 0.51,
+      "step": 1457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7831211443298613e-05,
+      "loss": 0.4761,
+      "step": 1458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7827931852645246e-05,
+      "loss": 0.56,
+      "step": 1459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7824650086286927e-05,
+      "loss": 0.5501,
+      "step": 1460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7821366145135788e-05,
+      "loss": 0.538,
+      "step": 1461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7818080030104578e-05,
+      "loss": 0.5581,
+      "step": 1462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781479174210664e-05,
+      "loss": 0.5363,
+      "step": 1463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7811501282055927e-05,
+      "loss": 0.5198,
+      "step": 1464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7808208650866993e-05,
+      "loss": 0.4784,
+      "step": 1465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7804913849454996e-05,
+      "loss": 0.4988,
+      "step": 1466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7801616878735696e-05,
+      "loss": 0.5702,
+      "step": 1467
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7798317739625455e-05,
+      "loss": 0.5197,
+      "step": 1468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7795016433041246e-05,
+      "loss": 0.5854,
+      "step": 1469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7791712959900637e-05,
+      "loss": 0.5021,
+      "step": 1470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778840732112179e-05,
+      "loss": 0.5196,
+      "step": 1471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778509951762349e-05,
+      "loss": 0.5048,
+      "step": 1472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778178955032511e-05,
+      "loss": 0.4864,
+      "step": 1473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7778477420146617e-05,
+      "loss": 0.505,
+      "step": 1474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77751631280086e-05,
+      "loss": 0.463,
+      "step": 1475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777184667483223e-05,
+      "loss": 0.5432,
+      "step": 1476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7768528061539293e-05,
+      "loss": 0.5003,
+      "step": 1477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776520728905216e-05,
+      "loss": 0.4998,
+      "step": 1478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776188435829382e-05,
+      "loss": 0.4601,
+      "step": 1479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775855927018784e-05,
+      "loss": 0.5035,
+      "step": 1480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775523202565841e-05,
+      "loss": 0.5781,
+      "step": 1481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7751902625630303e-05,
+      "loss": 0.4131,
+      "step": 1482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77485710710289e-05,
+      "loss": 0.4921,
+      "step": 1483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7745237362780175e-05,
+      "loss": 0.4441,
+      "step": 1484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7741901501810703e-05,
+      "loss": 0.4975,
+      "step": 1485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7738563489047654e-05,
+      "loss": 0.4481,
+      "step": 1486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735223325418812e-05,
+      "loss": 0.4896,
+      "step": 1487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773188101185253e-05,
+      "loss": 0.4763,
+      "step": 1488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7728536549277783e-05,
+      "loss": 0.4984,
+      "step": 1489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7725189938624135e-05,
+      "loss": 0.5575,
+      "step": 1490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7721841180821747e-05,
+      "loss": 0.4759,
+      "step": 1491
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7718490276801374e-05,
+      "loss": 0.5916,
+      "step": 1492
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7715137227494375e-05,
+      "loss": 0.4964,
+      "step": 1493
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7711782033832694e-05,
+      "loss": 0.4478,
+      "step": 1494
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7708424696748886e-05,
+      "loss": 0.4806,
+      "step": 1495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7705065217176085e-05,
+      "loss": 0.5517,
+      "step": 1496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7701703596048034e-05,
+      "loss": 0.5014,
+      "step": 1497
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7698339834299064e-05,
+      "loss": 0.441,
+      "step": 1498
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7694973932864106e-05,
+      "loss": 0.5609,
+      "step": 1499
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7691605892678677e-05,
+      "loss": 0.506,
+      "step": 1500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76882357146789e-05,
+      "loss": 0.5523,
+      "step": 1501
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684863399801485e-05,
+      "loss": 0.4705,
+      "step": 1502
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7681488948983737e-05,
+      "loss": 0.4897,
+      "step": 1503
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678112363163552e-05,
+      "loss": 0.4654,
+      "step": 1504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7674733643279428e-05,
+      "loss": 0.5327,
+      "step": 1505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7671352790270444e-05,
+      "loss": 0.4373,
+      "step": 1506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7667969805076284e-05,
+      "loss": 0.4616,
+      "step": 1507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766458468863722e-05,
+      "loss": 0.4962,
+      "step": 1508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766119744189411e-05,
+      "loss": 0.5722,
+      "step": 1509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7657808065788406e-05,
+      "loss": 0.5515,
+      "step": 1510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7654416561262166e-05,
+      "loss": 0.5146,
+      "step": 1511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7651022929258025e-05,
+      "loss": 0.4708,
+      "step": 1512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7647627170719213e-05,
+      "loss": 0.4698,
+      "step": 1513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7644229286589548e-05,
+      "loss": 0.5075,
+      "step": 1514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7640829277813445e-05,
+      "loss": 0.4621,
+      "step": 1515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7637427145335907e-05,
+      "loss": 0.5187,
+      "step": 1516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7634022890102527e-05,
+      "loss": 0.4994,
+      "step": 1517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630616513059485e-05,
+      "loss": 0.4521,
+      "step": 1518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762720801515356e-05,
+      "loss": 0.5244,
+      "step": 1519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7623797397332104e-05,
+      "loss": 0.5161,
+      "step": 1520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7620384660543077e-05,
+      "loss": 0.5768,
+      "step": 1521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616969805735015e-05,
+      "loss": 0.4539,
+      "step": 1522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761355283385705e-05,
+      "loss": 0.5145,
+      "step": 1523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7610133745858897e-05,
+      "loss": 0.4734,
+      "step": 1524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760671254269086e-05,
+      "loss": 0.4592,
+      "step": 1525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7603289225303837e-05,
+      "loss": 0.4735,
+      "step": 1526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7599863794649302e-05,
+      "loss": 0.5263,
+      "step": 1527
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7596436251679335e-05,
+      "loss": 0.4562,
+      "step": 1528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759300659734658e-05,
+      "loss": 0.5238,
+      "step": 1529
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758957483260428e-05,
+      "loss": 0.5461,
+      "step": 1530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7586140958406274e-05,
+      "loss": 0.5475,
+      "step": 1531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7582704975706967e-05,
+      "loss": 0.4496,
+      "step": 1532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7579266885461365e-05,
+      "loss": 0.524,
+      "step": 1533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575826688625053e-05,
+      "loss": 0.4922,
+      "step": 1534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572384386154207e-05,
+      "loss": 0.4387,
+      "step": 1535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7568939979005577e-05,
+      "loss": 0.5144,
+      "step": 1536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7565493468136517e-05,
+      "loss": 0.4704,
+      "step": 1537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7562044854504946e-05,
+      "loss": 0.4252,
+      "step": 1538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7558594139069374e-05,
+      "loss": 0.4941,
+      "step": 1539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555141322788907e-05,
+      "loss": 0.5102,
+      "step": 1540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755168640662322e-05,
+      "loss": 0.4603,
+      "step": 1541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7548229391532572e-05,
+      "loss": 0.4883,
+      "step": 1542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7544770278477816e-05,
+      "loss": 0.483,
+      "step": 1543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754130906842038e-05,
+      "loss": 0.5658,
+      "step": 1544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7537845762322273e-05,
+      "loss": 0.4614,
+      "step": 1545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75343803611461e-05,
+      "loss": 0.4661,
+      "step": 1546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.753091286585503e-05,
+      "loss": 0.4243,
+      "step": 1547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752744327741283e-05,
+      "loss": 0.5085,
+      "step": 1548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7523971596783837e-05,
+      "loss": 0.4986,
+      "step": 1549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752049782493297e-05,
+      "loss": 0.5009,
+      "step": 1550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7517021962825742e-05,
+      "loss": 0.4622,
+      "step": 1551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7513544011428238e-05,
+      "loss": 0.4694,
+      "step": 1552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7510063971707113e-05,
+      "loss": 0.4591,
+      "step": 1553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7506581844629628e-05,
+      "loss": 0.5022,
+      "step": 1554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503097631163602e-05,
+      "loss": 0.5477,
+      "step": 1555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7499611332277443e-05,
+      "loss": 0.5307,
+      "step": 1556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749612294894014e-05,
+      "loss": 0.4787,
+      "step": 1557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7492632482121247e-05,
+      "loss": 0.495,
+      "step": 1558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748913993279092e-05,
+      "loss": 0.4836,
+      "step": 1559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7485645301919884e-05,
+      "loss": 0.4751,
+      "step": 1560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7482148590479432e-05,
+      "loss": 0.5175,
+      "step": 1561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747864979944145e-05,
+      "loss": 0.4614,
+      "step": 1562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7475148929778392e-05,
+      "loss": 0.4959,
+      "step": 1563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74716459824633e-05,
+      "loss": 0.5637,
+      "step": 1564
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7468140958469783e-05,
+      "loss": 0.4518,
+      "step": 1565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464633858772037e-05,
+      "loss": 0.5025,
+      "step": 1566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.746112468434482e-05,
+      "loss": 0.501,
+      "step": 1567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7457613436163485e-05,
+      "loss": 0.5061,
+      "step": 1568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7454100115203948e-05,
+      "loss": 0.514,
+      "step": 1569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7450584722442707e-05,
+      "loss": 0.5524,
+      "step": 1570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7447067258856834e-05,
+      "loss": 0.457,
+      "step": 1571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7443547725423978e-05,
+      "loss": 0.4566,
+      "step": 1572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7440026123122365e-05,
+      "loss": 0.5457,
+      "step": 1573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7436502452930787e-05,
+      "loss": 0.4445,
+      "step": 1574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7432976715828623e-05,
+      "loss": 0.4683,
+      "step": 1575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742944891279582e-05,
+      "loss": 0.4782,
+      "step": 1576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7425919044812895e-05,
+      "loss": 0.4292,
+      "step": 1577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742238711286095e-05,
+      "loss": 0.4621,
+      "step": 1578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7418853117921652e-05,
+      "loss": 0.4199,
+      "step": 1579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7415317060977244e-05,
+      "loss": 0.505,
+      "step": 1580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741177894301054e-05,
+      "loss": 0.4863,
+      "step": 1581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740823876500493e-05,
+      "loss": 0.4668,
+      "step": 1582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740469652794438e-05,
+      "loss": 0.5106,
+      "step": 1583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401152232813413e-05,
+      "loss": 0.4721,
+      "step": 1584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7397605880597147e-05,
+      "loss": 0.5037,
+      "step": 1585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739405747228125e-05,
+      "loss": 0.4964,
+      "step": 1586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7390507008851977e-05,
+      "loss": 0.4723,
+      "step": 1587
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7386954491296144e-05,
+      "loss": 0.4651,
+      "step": 1588
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7383399920601148e-05,
+      "loss": 0.4055,
+      "step": 1589
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7379843297754944e-05,
+      "loss": 0.5551,
+      "step": 1590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7376284623746068e-05,
+      "loss": 0.4899,
+      "step": 1591
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737272389956362e-05,
+      "loss": 0.5148,
+      "step": 1592
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736916112619727e-05,
+      "loss": 0.4629,
+      "step": 1593
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365596304637267e-05,
+      "loss": 0.5707,
+      "step": 1594
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7362029435874417e-05,
+      "loss": 0.4552,
+      "step": 1595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73584605209001e-05,
+      "loss": 0.503,
+      "step": 1596
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7354889560706262e-05,
+      "loss": 0.5479,
+      "step": 1597
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7351316556285427e-05,
+      "loss": 0.4532,
+      "step": 1598
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7347741508630673e-05,
+      "loss": 0.5026,
+      "step": 1599
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7344164418735658e-05,
+      "loss": 0.4379,
+      "step": 1600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7340585287594605e-05,
+      "loss": 0.4813,
+      "step": 1601
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7337004116202293e-05,
+      "loss": 0.4693,
+      "step": 1602
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733342090555408e-05,
+      "loss": 0.4539,
+      "step": 1603
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7329835656645898e-05,
+      "loss": 0.4849,
+      "step": 1604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7326248370474222e-05,
+      "loss": 0.4561,
+      "step": 1605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7322659048036113e-05,
+      "loss": 0.4893,
+      "step": 1606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7319067690329194e-05,
+      "loss": 0.5419,
+      "step": 1607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7315474298351648e-05,
+      "loss": 0.4739,
+      "step": 1608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7311878873102225e-05,
+      "loss": 0.514,
+      "step": 1609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7308281415580242e-05,
+      "loss": 0.4799,
+      "step": 1610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730468192678559e-05,
+      "loss": 0.4924,
+      "step": 1611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7301080407718704e-05,
+      "loss": 0.4982,
+      "step": 1612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72974768593806e-05,
+      "loss": 0.4143,
+      "step": 1613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729387128277285e-05,
+      "loss": 0.5297,
+      "step": 1614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7290263678897593e-05,
+      "loss": 0.5215,
+      "step": 1615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7286654048757533e-05,
+      "loss": 0.5484,
+      "step": 1616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7283042393355933e-05,
+      "loss": 0.5185,
+      "step": 1617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279428713696625e-05,
+      "loss": 0.5092,
+      "step": 1618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7275813010783994e-05,
+      "loss": 0.565,
+      "step": 1619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7272195285622993e-05,
+      "loss": 0.5225,
+      "step": 1620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7268575539219142e-05,
+      "loss": 0.5249,
+      "step": 1621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7264953772578514e-05,
+      "loss": 0.5893,
+      "step": 1622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261329986707745e-05,
+      "loss": 0.4411,
+      "step": 1623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725770418261404e-05,
+      "loss": 0.4972,
+      "step": 1624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7254076361305155e-05,
+      "loss": 0.4903,
+      "step": 1625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7250446523789416e-05,
+      "loss": 0.5341,
+      "step": 1626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72468146710757e-05,
+      "loss": 0.4829,
+      "step": 1627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724318080417345e-05,
+      "loss": 0.4628,
+      "step": 1628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7239544924092664e-05,
+      "loss": 0.504,
+      "step": 1629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7235907031843914e-05,
+      "loss": 0.4837,
+      "step": 1630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7232267128438307e-05,
+      "loss": 0.5688,
+      "step": 1631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722862521488753e-05,
+      "loss": 0.5278,
+      "step": 1632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7224981292203815e-05,
+      "loss": 0.4883,
+      "step": 1633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7221335361399966e-05,
+      "loss": 0.4954,
+      "step": 1634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7217687423489332e-05,
+      "loss": 0.4461,
+      "step": 1635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721403747948583e-05,
+      "loss": 0.4922,
+      "step": 1636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7210385530403923e-05,
+      "loss": 0.5242,
+      "step": 1637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7206731577258643e-05,
+      "loss": 0.517,
+      "step": 1638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7203075621065573e-05,
+      "loss": 0.4636,
+      "step": 1639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7199417662840855e-05,
+      "loss": 0.4818,
+      "step": 1640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7195757703601183e-05,
+      "loss": 0.4958,
+      "step": 1641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7192095744363817e-05,
+      "loss": 0.4863,
+      "step": 1642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7188431786146563e-05,
+      "loss": 0.4445,
+      "step": 1643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718476582996778e-05,
+      "loss": 0.4895,
+      "step": 1644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.71810978768464e-05,
+      "loss": 0.5486,
+      "step": 1645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7177427927801885e-05,
+      "loss": 0.5412,
+      "step": 1646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7173755983854276e-05,
+      "loss": 0.4629,
+      "step": 1647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7170082046024156e-05,
+      "loss": 0.5262,
+      "step": 1648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716640611533266e-05,
+      "loss": 0.5369,
+      "step": 1649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716272819280148e-05,
+      "loss": 0.5385,
+      "step": 1650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7159048279452867e-05,
+      "loss": 0.5166,
+      "step": 1651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715536637630962e-05,
+      "loss": 0.4838,
+      "step": 1652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7151682484395088e-05,
+      "loss": 0.4675,
+      "step": 1653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7147996604733177e-05,
+      "loss": 0.4463,
+      "step": 1654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714430873834835e-05,
+      "loss": 0.5042,
+      "step": 1655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714061888626561e-05,
+      "loss": 0.4882,
+      "step": 1656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7136927049510525e-05,
+      "loss": 0.5332,
+      "step": 1657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7133233229109203e-05,
+      "loss": 0.5208,
+      "step": 1658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7129537426088313e-05,
+      "loss": 0.4734,
+      "step": 1659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7125839641475074e-05,
+      "loss": 0.4452,
+      "step": 1660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712213987629724e-05,
+      "loss": 0.5227,
+      "step": 1661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7118438131583142e-05,
+      "loss": 0.5601,
+      "step": 1662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7114734408361645e-05,
+      "loss": 0.5071,
+      "step": 1663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7111028707662162e-05,
+      "loss": 0.5105,
+      "step": 1664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710732103051466e-05,
+      "loss": 0.4829,
+      "step": 1665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7103611377949662e-05,
+      "loss": 0.4283,
+      "step": 1666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709989975099822e-05,
+      "loss": 0.6422,
+      "step": 1667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7096186150691965e-05,
+      "loss": 0.4441,
+      "step": 1668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7092470578063046e-05,
+      "loss": 0.5486,
+      "step": 1669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088753034144182e-05,
+      "loss": 0.5083,
+      "step": 1670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085033519968627e-05,
+      "loss": 0.4871,
+      "step": 1671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7081312036570188e-05,
+      "loss": 0.5251,
+      "step": 1672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707758858498322e-05,
+      "loss": 0.5134,
+      "step": 1673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7073863166242625e-05,
+      "loss": 0.5053,
+      "step": 1674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707013578138385e-05,
+      "loss": 0.4934,
+      "step": 1675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066406431442885e-05,
+      "loss": 0.5012,
+      "step": 1676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7062675117456277e-05,
+      "loss": 0.4947,
+      "step": 1677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7058941840461102e-05,
+      "loss": 0.5148,
+      "step": 1678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7055206601495e-05,
+      "loss": 0.4934,
+      "step": 1679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7051469401596148e-05,
+      "loss": 0.4538,
+      "step": 1680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7047730241803263e-05,
+      "loss": 0.5125,
+      "step": 1681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043989123155614e-05,
+      "loss": 0.5314,
+      "step": 1682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704024604669301e-05,
+      "loss": 0.5146,
+      "step": 1683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7036501013455808e-05,
+      "loss": 0.4219,
+      "step": 1684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703275402448491e-05,
+      "loss": 0.4617,
+      "step": 1685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029005080821756e-05,
+      "loss": 0.4794,
+      "step": 1686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7025254183508328e-05,
+      "loss": 0.4634,
+      "step": 1687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7021501333587164e-05,
+      "loss": 0.4699,
+      "step": 1688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7017746532101328e-05,
+      "loss": 0.4924,
+      "step": 1689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7013989780094437e-05,
+      "loss": 0.489,
+      "step": 1690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7010231078610648e-05,
+      "loss": 0.4541,
+      "step": 1691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700647042869466e-05,
+      "loss": 0.4718,
+      "step": 1692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7002707831391706e-05,
+      "loss": 0.5208,
+      "step": 1693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6998943287747573e-05,
+      "loss": 0.4694,
+      "step": 1694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6995176798808584e-05,
+      "loss": 0.4992,
+      "step": 1695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6991408365621595e-05,
+      "loss": 0.4592,
+      "step": 1696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6987637989234015e-05,
+      "loss": 0.5168,
+      "step": 1697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698386567069379e-05,
+      "loss": 0.5679,
+      "step": 1698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6980091411049388e-05,
+      "loss": 0.49,
+      "step": 1699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6976315211349848e-05,
+      "loss": 0.5132,
+      "step": 1700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6972537072644722e-05,
+      "loss": 0.4515,
+      "step": 1701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6968756995984117e-05,
+      "loss": 0.4717,
+      "step": 1702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6964974982418665e-05,
+      "loss": 0.5364,
+      "step": 1703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6961191032999548e-05,
+      "loss": 0.5531,
+      "step": 1704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695740514877848e-05,
+      "loss": 0.5256,
+      "step": 1705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6953617330807714e-05,
+      "loss": 0.5291,
+      "step": 1706
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694982758014004e-05,
+      "loss": 0.4752,
+      "step": 1707
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6946035897828794e-05,
+      "loss": 0.4257,
+      "step": 1708
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6942242284927826e-05,
+      "loss": 0.4486,
+      "step": 1709
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693844674249155e-05,
+      "loss": 0.5016,
+      "step": 1710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69346492715749e-05,
+      "loss": 0.4074,
+      "step": 1711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6930849873233345e-05,
+      "loss": 0.4605,
+      "step": 1712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6927048548522904e-05,
+      "loss": 0.4569,
+      "step": 1713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6923245298500113e-05,
+      "loss": 0.4359,
+      "step": 1714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6919440124222056e-05,
+      "loss": 0.4632,
+      "step": 1715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6915633026746344e-05,
+      "loss": 0.4145,
+      "step": 1716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6911824007131135e-05,
+      "loss": 0.4751,
+      "step": 1717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69080130664351e-05,
+      "loss": 0.4991,
+      "step": 1718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690420020571747e-05,
+      "loss": 0.5272,
+      "step": 1719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6900385426037986e-05,
+      "loss": 0.5407,
+      "step": 1720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6896568728456936e-05,
+      "loss": 0.5049,
+      "step": 1721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6892750114035143e-05,
+      "loss": 0.5095,
+      "step": 1722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6888929583833945e-05,
+      "loss": 0.472,
+      "step": 1723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6885107138915234e-05,
+      "loss": 0.4899,
+      "step": 1724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6881282780341427e-05,
+      "loss": 0.4556,
+      "step": 1725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687745650917546e-05,
+      "loss": 0.4883,
+      "step": 1726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6873628326480823e-05,
+      "loss": 0.4929,
+      "step": 1727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686979823332152e-05,
+      "loss": 0.4983,
+      "step": 1728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865966230762093e-05,
+      "loss": 0.5058,
+      "step": 1729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6862132319867613e-05,
+      "loss": 0.4871,
+      "step": 1730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6858296501703683e-05,
+      "loss": 0.5353,
+      "step": 1731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6854458777336435e-05,
+      "loss": 0.495,
+      "step": 1732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685061914783253e-05,
+      "loss": 0.5294,
+      "step": 1733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684677761425916e-05,
+      "loss": 0.4372,
+      "step": 1734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6842934177684043e-05,
+      "loss": 0.4821,
+      "step": 1735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683908883917543e-05,
+      "loss": 0.4793,
+      "step": 1736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68352415998021e-05,
+      "loss": 0.4341,
+      "step": 1737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6831392460633357e-05,
+      "loss": 0.5152,
+      "step": 1738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682754142273904e-05,
+      "loss": 0.4869,
+      "step": 1739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682368848718951e-05,
+      "loss": 0.5095,
+      "step": 1740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6819833655055652e-05,
+      "loss": 0.4682,
+      "step": 1741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6815976927408887e-05,
+      "loss": 0.5589,
+      "step": 1742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6812118305321158e-05,
+      "loss": 0.5157,
+      "step": 1743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6808257789864934e-05,
+      "loss": 0.4973,
+      "step": 1744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6804395382113213e-05,
+      "loss": 0.4842,
+      "step": 1745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6800531083139514e-05,
+      "loss": 0.4471,
+      "step": 1746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679666489401789e-05,
+      "loss": 0.5085,
+      "step": 1747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679279681582291e-05,
+      "loss": 0.4581,
+      "step": 1748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788926849629678e-05,
+      "loss": 0.5428,
+      "step": 1749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678505499651381e-05,
+      "loss": 0.502,
+      "step": 1750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6781181257551456e-05,
+      "loss": 0.4668,
+      "step": 1751
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6777305633819287e-05,
+      "loss": 0.5185,
+      "step": 1752
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6773428126394505e-05,
+      "loss": 0.481,
+      "step": 1753
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6769548736354822e-05,
+      "loss": 0.5472,
+      "step": 1754
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6765667464778482e-05,
+      "loss": 0.5193,
+      "step": 1755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676178431274425e-05,
+      "loss": 0.5349,
+      "step": 1756
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675789928133142e-05,
+      "loss": 0.4917,
+      "step": 1757
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675401237161979e-05,
+      "loss": 0.5002,
+      "step": 1758
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750123584689707e-05,
+      "loss": 0.5036,
+      "step": 1759
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6746232921622018e-05,
+      "loss": 0.4757,
+      "step": 1760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6742340383498094e-05,
+      "loss": 0.4773,
+      "step": 1761
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6738445971399836e-05,
+      "loss": 0.4535,
+      "step": 1762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6734549686409667e-05,
+      "loss": 0.5041,
+      "step": 1763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6730651529610516e-05,
+      "loss": 0.4491,
+      "step": 1764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672675150208585e-05,
+      "loss": 0.4739,
+      "step": 1765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672284960491964e-05,
+      "loss": 0.4944,
+      "step": 1766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671894583919639e-05,
+      "loss": 0.5112,
+      "step": 1767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715040206001118e-05,
+      "loss": 0.5183,
+      "step": 1768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6711132706419353e-05,
+      "loss": 0.5209,
+      "step": 1769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6707223341537157e-05,
+      "loss": 0.4809,
+      "step": 1770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6703312112441095e-05,
+      "loss": 0.4113,
+      "step": 1771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669939902021827e-05,
+      "loss": 0.4875,
+      "step": 1772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6695484065956286e-05,
+      "loss": 0.467,
+      "step": 1773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669156725074327e-05,
+      "loss": 0.5191,
+      "step": 1774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687648575667868e-05,
+      "loss": 0.4855,
+      "step": 1775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668372804181924e-05,
+      "loss": 0.4817,
+      "step": 1776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6679805650287068e-05,
+      "loss": 0.5536,
+      "step": 1777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667588140216154e-05,
+      "loss": 0.4648,
+      "step": 1778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6671955298533367e-05,
+      "loss": 0.4834,
+      "step": 1779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6668027340493778e-05,
+      "loss": 0.5043,
+      "step": 1780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6664097529134517e-05,
+      "loss": 0.5181,
+      "step": 1781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6660165865547835e-05,
+      "loss": 0.4585,
+      "step": 1782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6656232350826504e-05,
+      "loss": 0.5363,
+      "step": 1783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6652296986063813e-05,
+      "loss": 0.5056,
+      "step": 1784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6648359772353558e-05,
+      "loss": 0.4533,
+      "step": 1785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664442071079006e-05,
+      "loss": 0.4796,
+      "step": 1786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640479802468142e-05,
+      "loss": 0.5143,
+      "step": 1787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6636537048483143e-05,
+      "loss": 0.4597,
+      "step": 1788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6632592449930924e-05,
+      "loss": 0.4568,
+      "step": 1789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662864600790784e-05,
+      "loss": 0.4803,
+      "step": 1790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6624697723510775e-05,
+      "loss": 0.4424,
+      "step": 1791
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6620747597837128e-05,
+      "loss": 0.5241,
+      "step": 1792
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661679563198479e-05,
+      "loss": 0.4092,
+      "step": 1793
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6612841827052185e-05,
+      "loss": 0.437,
+      "step": 1794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660888618413823e-05,
+      "loss": 0.5221,
+      "step": 1795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6604928704342367e-05,
+      "loss": 0.4429,
+      "step": 1796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660096938876454e-05,
+      "loss": 0.3947,
+      "step": 1797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659700823850521e-05,
+      "loss": 0.3961,
+      "step": 1798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6593045254665336e-05,
+      "loss": 0.4537,
+      "step": 1799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65890804383464e-05,
+      "loss": 0.4342,
+      "step": 1800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658511379065039e-05,
+      "loss": 0.4315,
+      "step": 1801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6581145312679793e-05,
+      "loss": 0.5121,
+      "step": 1802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6577175005537623e-05,
+      "loss": 0.4675,
+      "step": 1803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6573202870327385e-05,
+      "loss": 0.4769,
+      "step": 1804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65692289081531e-05,
+      "loss": 0.5108,
+      "step": 1805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6565253120119298e-05,
+      "loss": 0.607,
+      "step": 1806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6561275507331012e-05,
+      "loss": 0.5148,
+      "step": 1807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655729607089379e-05,
+      "loss": 0.528,
+      "step": 1808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655331481191367e-05,
+      "loss": 0.4993,
+      "step": 1809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654933173149722e-05,
+      "loss": 0.4582,
+      "step": 1810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6545346830751495e-05,
+      "loss": 0.4216,
+      "step": 1811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6541360110784067e-05,
+      "loss": 0.5067,
+      "step": 1812
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653737157270301e-05,
+      "loss": 0.4413,
+      "step": 1813
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65333812176169e-05,
+      "loss": 0.4696,
+      "step": 1814
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6529389046634823e-05,
+      "loss": 0.5509,
+      "step": 1815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6525395060866367e-05,
+      "loss": 0.4706,
+      "step": 1816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6521399261421626e-05,
+      "loss": 0.4807,
+      "step": 1817
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6517401649411197e-05,
+      "loss": 0.4649,
+      "step": 1818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6513402225946183e-05,
+      "loss": 0.4925,
+      "step": 1819
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6509400992138183e-05,
+      "loss": 0.532,
+      "step": 1820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6505397949099314e-05,
+      "loss": 0.4668,
+      "step": 1821
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650139309794218e-05,
+      "loss": 0.553,
+      "step": 1822
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6497386439779897e-05,
+      "loss": 0.5132,
+      "step": 1823
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6493377975726076e-05,
+      "loss": 0.4689,
+      "step": 1824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648936770689484e-05,
+      "loss": 0.5301,
+      "step": 1825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648535563440081e-05,
+      "loss": 0.5068,
+      "step": 1826
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6481341759359094e-05,
+      "loss": 0.4528,
+      "step": 1827
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6477326082885325e-05,
+      "loss": 0.4765,
+      "step": 1828
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647330860609562e-05,
+      "loss": 0.4846,
+      "step": 1829
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469289330106607e-05,
+      "loss": 0.5684,
+      "step": 1830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6465268256035402e-05,
+      "loss": 0.493,
+      "step": 1831
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646124538499963e-05,
+      "loss": 0.5467,
+      "step": 1832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6457220718117413e-05,
+      "loss": 0.4537,
+      "step": 1833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6453194256507367e-05,
+      "loss": 0.4679,
+      "step": 1834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644916600128862e-05,
+      "loss": 0.4369,
+      "step": 1835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6445135953580783e-05,
+      "loss": 0.6017,
+      "step": 1836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644110411450398e-05,
+      "loss": 0.481,
+      "step": 1837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6437070485178816e-05,
+      "loss": 0.5181,
+      "step": 1838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6433035066726412e-05,
+      "loss": 0.4666,
+      "step": 1839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642899786026837e-05,
+      "loss": 0.4989,
+      "step": 1840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6424958866926798e-05,
+      "loss": 0.4721,
+      "step": 1841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64209180878243e-05,
+      "loss": 0.5139,
+      "step": 1842
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6416875524083974e-05,
+      "loss": 0.5419,
+      "step": 1843
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6412831176829418e-05,
+      "loss": 0.4826,
+      "step": 1844
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640878504718472e-05,
+      "loss": 0.4628,
+      "step": 1845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6404737136274466e-05,
+      "loss": 0.5282,
+      "step": 1846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6400687445223734e-05,
+      "loss": 0.4816,
+      "step": 1847
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6396635975158113e-05,
+      "loss": 0.5365,
+      "step": 1848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6392582727203657e-05,
+      "loss": 0.5052,
+      "step": 1849
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388527702486942e-05,
+      "loss": 0.4106,
+      "step": 1850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6384470902135022e-05,
+      "loss": 0.4454,
+      "step": 1851
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638041232727545e-05,
+      "loss": 0.4542,
+      "step": 1852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6376351979036273e-05,
+      "loss": 0.5278,
+      "step": 1853
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6372289858546028e-05,
+      "loss": 0.4917,
+      "step": 1854
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636822596693374e-05,
+      "loss": 0.4608,
+      "step": 1855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6364160305328942e-05,
+      "loss": 0.4362,
+      "step": 1856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6360092874861643e-05,
+      "loss": 0.4969,
+      "step": 1857
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635602367666235e-05,
+      "loss": 0.5152,
+      "step": 1858
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635195271186206e-05,
+      "loss": 0.461,
+      "step": 1859
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6347879981592268e-05,
+      "loss": 0.5409,
+      "step": 1860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343805486984945e-05,
+      "loss": 0.5357,
+      "step": 1861
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6339729229172565e-05,
+      "loss": 0.4542,
+      "step": 1862
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633565120928809e-05,
+      "loss": 0.4928,
+      "step": 1863
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633157142846497e-05,
+      "loss": 0.5203,
+      "step": 1864
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632748988783714e-05,
+      "loss": 0.4677,
+      "step": 1865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6323406588539034e-05,
+      "loss": 0.4919,
+      "step": 1866
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631932153170556e-05,
+      "loss": 0.5242,
+      "step": 1867
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6315234718472135e-05,
+      "loss": 0.4851,
+      "step": 1868
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6311146149974647e-05,
+      "loss": 0.5478,
+      "step": 1869
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6307055827349472e-05,
+      "loss": 0.4495,
+      "step": 1870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6302963751733492e-05,
+      "loss": 0.452,
+      "step": 1871
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6298869924264055e-05,
+      "loss": 0.5071,
+      "step": 1872
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629477434607901e-05,
+      "loss": 0.4664,
+      "step": 1873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.629067701831668e-05,
+      "loss": 0.5392,
+      "step": 1874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6286577942115887e-05,
+      "loss": 0.4796,
+      "step": 1875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6282477118615928e-05,
+      "loss": 0.5403,
+      "step": 1876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6278374548956594e-05,
+      "loss": 0.4934,
+      "step": 1877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6274270234278162e-05,
+      "loss": 0.5065,
+      "step": 1878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6270164175721383e-05,
+      "loss": 0.4616,
+      "step": 1879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6266056374427502e-05,
+      "loss": 0.4603,
+      "step": 1880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6261946831538255e-05,
+      "loss": 0.552,
+      "step": 1881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625783554819584e-05,
+      "loss": 0.4878,
+      "step": 1882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6253722525542958e-05,
+      "loss": 0.503,
+      "step": 1883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624960776472279e-05,
+      "loss": 0.4957,
+      "step": 1884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245491266878995e-05,
+      "loss": 0.5204,
+      "step": 1885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6241373033155715e-05,
+      "loss": 0.526,
+      "step": 1886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6237253064697582e-05,
+      "loss": 0.4511,
+      "step": 1887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6233131362649703e-05,
+      "loss": 0.4243,
+      "step": 1888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622900792815767e-05,
+      "loss": 0.5215,
+      "step": 1889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622488276236755e-05,
+      "loss": 0.5129,
+      "step": 1890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6220755866425905e-05,
+      "loss": 0.544,
+      "step": 1891
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6216627241479763e-05,
+      "loss": 0.4303,
+      "step": 1892
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621249688867664e-05,
+      "loss": 0.5897,
+      "step": 1893
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6208364809164537e-05,
+      "loss": 0.5628,
+      "step": 1894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6204231004091923e-05,
+      "loss": 0.428,
+      "step": 1895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6200095474607753e-05,
+      "loss": 0.4985,
+      "step": 1896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6195958221861463e-05,
+      "loss": 0.5187,
+      "step": 1897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6191819247002968e-05,
+      "loss": 0.4981,
+      "step": 1898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6187678551182656e-05,
+      "loss": 0.5189,
+      "step": 1899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61835361355514e-05,
+      "loss": 0.4852,
+      "step": 1900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617939200126055e-05,
+      "loss": 0.4899,
+      "step": 1901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617524614946192e-05,
+      "loss": 0.4921,
+      "step": 1902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617109858130783e-05,
+      "loss": 0.4285,
+      "step": 1903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6166949297951046e-05,
+      "loss": 0.4999,
+      "step": 1904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6162798300544832e-05,
+      "loss": 0.5316,
+      "step": 1905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615864559024292e-05,
+      "loss": 0.4557,
+      "step": 1906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615449116819952e-05,
+      "loss": 0.4401,
+      "step": 1907
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6150335035569318e-05,
+      "loss": 0.4777,
+      "step": 1908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6146177193507474e-05,
+      "loss": 0.4117,
+      "step": 1909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614201764316962e-05,
+      "loss": 0.4839,
+      "step": 1910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613785638571187e-05,
+      "loss": 0.4706,
+      "step": 1911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613369342229081e-05,
+      "loss": 0.5058,
+      "step": 1912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6129528754063494e-05,
+      "loss": 0.4783,
+      "step": 1913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6125362382187462e-05,
+      "loss": 0.5298,
+      "step": 1914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6121194307820715e-05,
+      "loss": 0.4467,
+      "step": 1915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6117024532121734e-05,
+      "loss": 0.5126,
+      "step": 1916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6112853056249474e-05,
+      "loss": 0.5188,
+      "step": 1917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6108679881363358e-05,
+      "loss": 0.4832,
+      "step": 1918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6104505008623285e-05,
+      "loss": 0.5343,
+      "step": 1919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6100328439189618e-05,
+      "loss": 0.5179,
+      "step": 1920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609615017422321e-05,
+      "loss": 0.4576,
+      "step": 1921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609197021488536e-05,
+      "loss": 0.4588,
+      "step": 1922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6087788562337858e-05,
+      "loss": 0.4771,
+      "step": 1923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6083605217742958e-05,
+      "loss": 0.4836,
+      "step": 1924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607942018226338e-05,
+      "loss": 0.4476,
+      "step": 1925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607523345706232e-05,
+      "loss": 0.4825,
+      "step": 1926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6071045043303444e-05,
+      "loss": 0.4907,
+      "step": 1927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6066854942150882e-05,
+      "loss": 0.4667,
+      "step": 1928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6062663154769237e-05,
+      "loss": 0.4622,
+      "step": 1929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605846968232358e-05,
+      "loss": 0.4809,
+      "step": 1930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6054274525979442e-05,
+      "loss": 0.4712,
+      "step": 1931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6050077686902844e-05,
+      "loss": 0.4592,
+      "step": 1932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6045879166260253e-05,
+      "loss": 0.4321,
+      "step": 1933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6041678965218606e-05,
+      "loss": 0.5133,
+      "step": 1934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6037477084945323e-05,
+      "loss": 0.4047,
+      "step": 1935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603327352660827e-05,
+      "loss": 0.4263,
+      "step": 1936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6029068291375794e-05,
+      "loss": 0.4922,
+      "step": 1937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6024861380416706e-05,
+      "loss": 0.4883,
+      "step": 1938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602065279490027e-05,
+      "loss": 0.458,
+      "step": 1939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6016442535996235e-05,
+      "loss": 0.4717,
+      "step": 1940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012230604874804e-05,
+      "loss": 0.4839,
+      "step": 1941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6008017002706643e-05,
+      "loss": 0.4781,
+      "step": 1942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6003801730662886e-05,
+      "loss": 0.5763,
+      "step": 1943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5999584789915132e-05,
+      "loss": 0.4424,
+      "step": 1944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5995366181635444e-05,
+      "loss": 0.4418,
+      "step": 1945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5991145906996347e-05,
+      "loss": 0.4597,
+      "step": 1946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5986923967170826e-05,
+      "loss": 0.471,
+      "step": 1947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598270036333234e-05,
+      "loss": 0.4332,
+      "step": 1948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597847509665479e-05,
+      "loss": 0.4922,
+      "step": 1949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597424816831256e-05,
+      "loss": 0.5391,
+      "step": 1950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5970019579480487e-05,
+      "loss": 0.5149,
+      "step": 1951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965789331333873e-05,
+      "loss": 0.518,
+      "step": 1952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596155742504847e-05,
+      "loss": 0.4446,
+      "step": 1953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5957323861800506e-05,
+      "loss": 0.5375,
+      "step": 1954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595308864276666e-05,
+      "loss": 0.5,
+      "step": 1955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5948851769124075e-05,
+      "loss": 0.4316,
+      "step": 1956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5944613242050356e-05,
+      "loss": 0.4683,
+      "step": 1957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594037306272356e-05,
+      "loss": 0.4871,
+      "step": 1958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5936131232322204e-05,
+      "loss": 0.493,
+      "step": 1959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593188775202528e-05,
+      "loss": 0.4254,
+      "step": 1960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5927642623012212e-05,
+      "loss": 0.5385,
+      "step": 1961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923395846462905e-05,
+      "loss": 0.4851,
+      "step": 1962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5919147423557714e-05,
+      "loss": 0.4278,
+      "step": 1963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5914897355477438e-05,
+      "loss": 0.3925,
+      "step": 1964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5910645643403367e-05,
+      "loss": 0.4936,
+      "step": 1965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590639228851721e-05,
+      "loss": 0.5501,
+      "step": 1966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5902137292001157e-05,
+      "loss": 0.4383,
+      "step": 1967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5897880655037843e-05,
+      "loss": 0.5074,
+      "step": 1968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5893622378810373e-05,
+      "loss": 0.5762,
+      "step": 1969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5889362464502284e-05,
+      "loss": 0.5013,
+      "step": 1970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5885100913297586e-05,
+      "loss": 0.4706,
+      "step": 1971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5880837726380748e-05,
+      "loss": 0.4917,
+      "step": 1972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587657290493667e-05,
+      "loss": 0.5878,
+      "step": 1973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5872306450150737e-05,
+      "loss": 0.5393,
+      "step": 1974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5868038363208766e-05,
+      "loss": 0.5678,
+      "step": 1975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5863768645297036e-05,
+      "loss": 0.5028,
+      "step": 1976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5859497297602273e-05,
+      "loss": 0.5433,
+      "step": 1977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585522432131166e-05,
+      "loss": 0.535,
+      "step": 1978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5850949717612844e-05,
+      "loss": 0.5496,
+      "step": 1979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5846673487693905e-05,
+      "loss": 0.465,
+      "step": 1980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5842395632743385e-05,
+      "loss": 0.4414,
+      "step": 1981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5838116153950275e-05,
+      "loss": 0.5059,
+      "step": 1982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583383505250402e-05,
+      "loss": 0.4202,
+      "step": 1983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829552329594517e-05,
+      "loss": 0.5129,
+      "step": 1984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5825267986412106e-05,
+      "loss": 0.5374,
+      "step": 1985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5820982024147585e-05,
+      "loss": 0.494,
+      "step": 1986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5816694443992205e-05,
+      "loss": 0.4806,
+      "step": 1987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5812405247137653e-05,
+      "loss": 0.4596,
+      "step": 1988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5808114434776076e-05,
+      "loss": 0.4268,
+      "step": 1989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580382200810007e-05,
+      "loss": 0.4779,
+      "step": 1990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5799527968302675e-05,
+      "loss": 0.5124,
+      "step": 1991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5795232316577383e-05,
+      "loss": 0.4852,
+      "step": 1992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579093505411813e-05,
+      "loss": 0.5265,
+      "step": 1993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5786636182119308e-05,
+      "loss": 0.4748,
+      "step": 1994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5782335701775747e-05,
+      "loss": 0.5229,
+      "step": 1995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5778033614282727e-05,
+      "loss": 0.5002,
+      "step": 1996
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5773729920835975e-05,
+      "loss": 0.4578,
+      "step": 1997
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5769424622631667e-05,
+      "loss": 0.4939,
+      "step": 1998
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5765117720866423e-05,
+      "loss": 0.459,
+      "step": 1999
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5760809216737308e-05,
+      "loss": 0.4802,
+      "step": 2000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5756499111441832e-05,
+      "loss": 0.5286,
+      "step": 2001
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5752187406177956e-05,
+      "loss": 0.5086,
+      "step": 2002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5747874102144073e-05,
+      "loss": 0.5257,
+      "step": 2003
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574355920053903e-05,
+      "loss": 0.4957,
+      "step": 2004
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5739242702562115e-05,
+      "loss": 0.4336,
+      "step": 2005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573492460941307e-05,
+      "loss": 0.5915,
+      "step": 2006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5730604922292057e-05,
+      "loss": 0.5174,
+      "step": 2007
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5726283642399707e-05,
+      "loss": 0.4927,
+      "step": 2008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5721960770937076e-05,
+      "loss": 0.4483,
+      "step": 2009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571763630910567e-05,
+      "loss": 0.4461,
+      "step": 2010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571331025810743e-05,
+      "loss": 0.4435,
+      "step": 2011
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5708982619144752e-05,
+      "loss": 0.4576,
+      "step": 2012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570465339342046e-05,
+      "loss": 0.5153,
+      "step": 2013
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570032258213783e-05,
+      "loss": 0.4646,
+      "step": 2014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5695990186500564e-05,
+      "loss": 0.4499,
+      "step": 2015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691656207712818e-05,
+      "loss": 0.5192,
+      "step": 2016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568732064697919e-05,
+      "loss": 0.4388,
+      "step": 2017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5682983505504698e-05,
+      "loss": 0.5189,
+      "step": 2018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5678644784494824e-05,
+      "loss": 0.4886,
+      "step": 2019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567430448515547e-05,
+      "loss": 0.4527,
+      "step": 2020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5669962608692987e-05,
+      "loss": 0.4767,
+      "step": 2021
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566561915631416e-05,
+      "loss": 0.459,
+      "step": 2022
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5661274129226214e-05,
+      "loss": 0.5217,
+      "step": 2023
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5656927528636813e-05,
+      "loss": 0.462,
+      "step": 2024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5652579355754057e-05,
+      "loss": 0.5073,
+      "step": 2025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564822961178648e-05,
+      "loss": 0.444,
+      "step": 2026
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5643878297943054e-05,
+      "loss": 0.4838,
+      "step": 2027
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5639525415433187e-05,
+      "loss": 0.5103,
+      "step": 2028
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5635170965466734e-05,
+      "loss": 0.4621,
+      "step": 2029
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563081494925397e-05,
+      "loss": 0.5031,
+      "step": 2030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5626457368005606e-05,
+      "loss": 0.4825,
+      "step": 2031
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5622098222932802e-05,
+      "loss": 0.433,
+      "step": 2032
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561773751524714e-05,
+      "loss": 0.4482,
+      "step": 2033
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5613375246160643e-05,
+      "loss": 0.44,
+      "step": 2034
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5609011416885763e-05,
+      "loss": 0.4514,
+      "step": 2035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5604646028635388e-05,
+      "loss": 0.4902,
+      "step": 2036
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5600279082622842e-05,
+      "loss": 0.4907,
+      "step": 2037
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5595910580061876e-05,
+      "loss": 0.5137,
+      "step": 2038
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591540522166678e-05,
+      "loss": 0.4498,
+      "step": 2039
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5587168910151872e-05,
+      "loss": 0.4578,
+      "step": 2040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55827957452325e-05,
+      "loss": 0.4482,
+      "step": 2041
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5578421028624053e-05,
+      "loss": 0.4233,
+      "step": 2042
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5574044761542438e-05,
+      "loss": 0.5111,
+      "step": 2043
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5569666945204006e-05,
+      "loss": 0.4282,
+      "step": 2044
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556528758082553e-05,
+      "loss": 0.439,
+      "step": 2045
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5560906669624216e-05,
+      "loss": 0.4441,
+      "step": 2046
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5556524212817704e-05,
+      "loss": 0.4825,
+      "step": 2047
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5552140211624052e-05,
+      "loss": 0.4486,
+      "step": 2048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554775466726176e-05,
+      "loss": 0.4145,
+      "step": 2049
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5543367580949745e-05,
+      "loss": 0.4616,
+      "step": 2050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5538978953907368e-05,
+      "loss": 0.5865,
+      "step": 2051
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5534588787354407e-05,
+      "loss": 0.4552,
+      "step": 2052
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530197082511066e-05,
+      "loss": 0.4645,
+      "step": 2053
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5525803840597985e-05,
+      "loss": 0.5249,
+      "step": 2054
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5521409062836222e-05,
+      "loss": 0.4848,
+      "step": 2055
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551701275044727e-05,
+      "loss": 0.4586,
+      "step": 2056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551261490465305e-05,
+      "loss": 0.4485,
+      "step": 2057
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5508215526675897e-05,
+      "loss": 0.524,
+      "step": 2058
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5503814617738588e-05,
+      "loss": 0.4548,
+      "step": 2059
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5499412179064305e-05,
+      "loss": 0.5445,
+      "step": 2060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5495008211876678e-05,
+      "loss": 0.492,
+      "step": 2061
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549060271739975e-05,
+      "loss": 0.4362,
+      "step": 2062
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548619569685798e-05,
+      "loss": 0.4593,
+      "step": 2063
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548178715147627e-05,
+      "loss": 0.4669,
+      "step": 2064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5477377082479934e-05,
+      "loss": 0.4351,
+      "step": 2065
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5472965491094714e-05,
+      "loss": 0.4252,
+      "step": 2066
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5468552378546768e-05,
+      "loss": 0.4465,
+      "step": 2067
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5464137746062684e-05,
+      "loss": 0.4606,
+      "step": 2068
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5459721594869475e-05,
+      "loss": 0.4756,
+      "step": 2069
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5455303926194565e-05,
+      "loss": 0.4815,
+      "step": 2070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545088474126581e-05,
+      "loss": 0.57,
+      "step": 2071
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544646404131148e-05,
+      "loss": 0.4299,
+      "step": 2072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5442041827560274e-05,
+      "loss": 0.5377,
+      "step": 2073
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5437618101241307e-05,
+      "loss": 0.4784,
+      "step": 2074
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543319286358411e-05,
+      "loss": 0.5206,
+      "step": 2075
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5428766115818644e-05,
+      "loss": 0.4641,
+      "step": 2076
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542433785917529e-05,
+      "loss": 0.4484,
+      "step": 2077
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5419908094884827e-05,
+      "loss": 0.4675,
+      "step": 2078
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5415476824178483e-05,
+      "loss": 0.4632,
+      "step": 2079
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5411044048287887e-05,
+      "loss": 0.4668,
+      "step": 2080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5406609768445085e-05,
+      "loss": 0.4873,
+      "step": 2081
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5402173985882555e-05,
+      "loss": 0.4015,
+      "step": 2082
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5397736701833176e-05,
+      "loss": 0.4611,
+      "step": 2083
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5393297917530262e-05,
+      "loss": 0.4408,
+      "step": 2084
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538885763420752e-05,
+      "loss": 0.4783,
+      "step": 2085
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5384415853099098e-05,
+      "loss": 0.4351,
+      "step": 2086
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537997257543955e-05,
+      "loss": 0.4645,
+      "step": 2087
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5375527802463845e-05,
+      "loss": 0.5017,
+      "step": 2088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537108153540736e-05,
+      "loss": 0.5099,
+      "step": 2089
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5366633775505907e-05,
+      "loss": 0.5101,
+      "step": 2090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5362184523995702e-05,
+      "loss": 0.4537,
+      "step": 2091
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5357733782113368e-05,
+      "loss": 0.4782,
+      "step": 2092
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535328155109595e-05,
+      "loss": 0.5283,
+      "step": 2093
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5348827832180914e-05,
+      "loss": 0.5083,
+      "step": 2094
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5344372626606126e-05,
+      "loss": 0.4756,
+      "step": 2095
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5339915935609876e-05,
+      "loss": 0.4575,
+      "step": 2096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5335457760430854e-05,
+      "loss": 0.4464,
+      "step": 2097
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5330998102308177e-05,
+      "loss": 0.4381,
+      "step": 2098
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5326536962481368e-05,
+      "loss": 0.5524,
+      "step": 2099
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.532207434219036e-05,
+      "loss": 0.4822,
+      "step": 2100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53176102426755e-05,
+      "loss": 0.4396,
+      "step": 2101
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5313144665177543e-05,
+      "loss": 0.4998,
+      "step": 2102
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530867761093766e-05,
+      "loss": 0.5071,
+      "step": 2103
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5304209081197425e-05,
+      "loss": 0.4469,
+      "step": 2104
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5299739077198835e-05,
+      "loss": 0.5134,
+      "step": 2105
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529526760018428e-05,
+      "loss": 0.4651,
+      "step": 2106
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290794651396573e-05,
+      "loss": 0.4777,
+      "step": 2107
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5286320232078928e-05,
+      "loss": 0.4438,
+      "step": 2108
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528184434347497e-05,
+      "loss": 0.4567,
+      "step": 2109
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5277366986828736e-05,
+      "loss": 0.4692,
+      "step": 2110
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5272888163384668e-05,
+      "loss": 0.4364,
+      "step": 2111
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526840787438761e-05,
+      "loss": 0.4734,
+      "step": 2112
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5263926121082827e-05,
+      "loss": 0.4491,
+      "step": 2113
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5259442904715976e-05,
+      "loss": 0.3863,
+      "step": 2114
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5254958226533131e-05,
+      "loss": 0.536,
+      "step": 2115
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5250472087780773e-05,
+      "loss": 0.3949,
+      "step": 2116
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245984489705773e-05,
+      "loss": 0.4953,
+      "step": 2117
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5241495433555433e-05,
+      "loss": 0.4466,
+      "step": 2118
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5237004920577438e-05,
+      "loss": 0.4981,
+      "step": 2119
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5232512952019894e-05,
+      "loss": 0.4971,
+      "step": 2120
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5228019529131297e-05,
+      "loss": 0.4681,
+      "step": 2121
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5223524653160555e-05,
+      "loss": 0.5102,
+      "step": 2122
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521902832535698e-05,
+      "loss": 0.4969,
+      "step": 2123
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5214530546970294e-05,
+      "loss": 0.45,
+      "step": 2124
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5210031319250607e-05,
+      "loss": 0.449,
+      "step": 2125
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205530643448444e-05,
+      "loss": 0.4586,
+      "step": 2126
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5201028520814722e-05,
+      "loss": 0.4732,
+      "step": 2127
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5196524952600776e-05,
+      "loss": 0.4999,
+      "step": 2128
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5192019940058322e-05,
+      "loss": 0.4484,
+      "step": 2129
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187513484439502e-05,
+      "loss": 0.4757,
+      "step": 2130
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5183005586996835e-05,
+      "loss": 0.4263,
+      "step": 2131
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5178496248983254e-05,
+      "loss": 0.4788,
+      "step": 2132
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5173985471652095e-05,
+      "loss": 0.4645,
+      "step": 2133
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5169473256257083e-05,
+      "loss": 0.438,
+      "step": 2134
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5164959604052352e-05,
+      "loss": 0.4806,
+      "step": 2135
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.516044451629243e-05,
+      "loss": 0.4124,
+      "step": 2136
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515592799423225e-05,
+      "loss": 0.4524,
+      "step": 2137
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5151410039127133e-05,
+      "loss": 0.4584,
+      "step": 2138
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5146890652232806e-05,
+      "loss": 0.4605,
+      "step": 2139
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5142369834805401e-05,
+      "loss": 0.4471,
+      "step": 2140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5137847588101431e-05,
+      "loss": 0.4934,
+      "step": 2141
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5133323913377817e-05,
+      "loss": 0.4826,
+      "step": 2142
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5128798811891879e-05,
+      "loss": 0.4175,
+      "step": 2143
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512427228490132e-05,
+      "loss": 0.4462,
+      "step": 2144
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511974433366426e-05,
+      "loss": 0.4955,
+      "step": 2145
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5115214959439197e-05,
+      "loss": 0.4942,
+      "step": 2146
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5110684163485028e-05,
+      "loss": 0.5016,
+      "step": 2147
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5106151947061052e-05,
+      "loss": 0.458,
+      "step": 2148
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5101618311426958e-05,
+      "loss": 0.4552,
+      "step": 2149
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097083257842831e-05,
+      "loss": 0.4211,
+      "step": 2150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5092546787569151e-05,
+      "loss": 0.4402,
+      "step": 2151
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5088008901866784e-05,
+      "loss": 0.5002,
+      "step": 2152
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5083469601996999e-05,
+      "loss": 0.4583,
+      "step": 2153
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5078928889221457e-05,
+      "loss": 0.4609,
+      "step": 2154
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5074386764802207e-05,
+      "loss": 0.4541,
+      "step": 2155
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506984323000169e-05,
+      "loss": 0.4615,
+      "step": 2156
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5065298286082746e-05,
+      "loss": 0.5063,
+      "step": 2157
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5060751934308602e-05,
+      "loss": 0.5103,
+      "step": 2158
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5056204175942873e-05,
+      "loss": 0.5183,
+      "step": 2159
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505165501224957e-05,
+      "loss": 0.4396,
+      "step": 2160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5047104444493097e-05,
+      "loss": 0.4522,
+      "step": 2161
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5042552473938235e-05,
+      "loss": 0.4392,
+      "step": 2162
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5037999101850175e-05,
+      "loss": 0.4852,
+      "step": 2163
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5033444329494483e-05,
+      "loss": 0.4934,
+      "step": 2164
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5028888158137114e-05,
+      "loss": 0.525,
+      "step": 2165
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.502433058904442e-05,
+      "loss": 0.4222,
+      "step": 2166
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5019771623483136e-05,
+      "loss": 0.5013,
+      "step": 2167
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501521126272039e-05,
+      "loss": 0.4482,
+      "step": 2168
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501064950802369e-05,
+      "loss": 0.529,
+      "step": 2169
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5006086360660934e-05,
+      "loss": 0.4802,
+      "step": 2170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5001521821900414e-05,
+      "loss": 0.4096,
+      "step": 2171
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4996955893010798e-05,
+      "loss": 0.475,
+      "step": 2172
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992388575261152e-05,
+      "loss": 0.4919,
+      "step": 2173
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4987819869920919e-05,
+      "loss": 0.5232,
+      "step": 2174
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4983249778259928e-05,
+      "loss": 0.4149,
+      "step": 2175
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4978678301548397e-05,
+      "loss": 0.474,
+      "step": 2176
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497410544105693e-05,
+      "loss": 0.4582,
+      "step": 2177
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4969531198056514e-05,
+      "loss": 0.5182,
+      "step": 2178
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4964955573818518e-05,
+      "loss": 0.4667,
+      "step": 2179
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4960378569614695e-05,
+      "loss": 0.4749,
+      "step": 2180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4955800186717187e-05,
+      "loss": 0.5286,
+      "step": 2181
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4951220426398512e-05,
+      "loss": 0.5134,
+      "step": 2182
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4946639289931575e-05,
+      "loss": 0.4707,
+      "step": 2183
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4942056778589665e-05,
+      "loss": 0.4604,
+      "step": 2184
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4937472893646448e-05,
+      "loss": 0.4767,
+      "step": 2185
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4932887636375975e-05,
+      "loss": 0.5018,
+      "step": 2186
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4928301008052675e-05,
+      "loss": 0.4907,
+      "step": 2187
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4923713009951371e-05,
+      "loss": 0.4405,
+      "step": 2188
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4919123643347245e-05,
+      "loss": 0.5111,
+      "step": 2189
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4914532909515878e-05,
+      "loss": 0.4063,
+      "step": 2190
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4909940809733223e-05,
+      "loss": 0.4941,
+      "step": 2191
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905347345275616e-05,
+      "loss": 0.5009,
+      "step": 2192
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4900752517419763e-05,
+      "loss": 0.4139,
+      "step": 2193
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4896156327442764e-05,
+      "loss": 0.4164,
+      "step": 2194
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4891558776622087e-05,
+      "loss": 0.4811,
+      "step": 2195
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4886959866235575e-05,
+      "loss": 0.4254,
+      "step": 2196
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4882359597561466e-05,
+      "loss": 0.4333,
+      "step": 2197
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4877757971878356e-05,
+      "loss": 0.4419,
+      "step": 2198
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4873154990465228e-05,
+      "loss": 0.4135,
+      "step": 2199
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4868550654601443e-05,
+      "loss": 0.3491,
+      "step": 2200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4863944965566735e-05,
+      "loss": 0.4781,
+      "step": 2201
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485933792464121e-05,
+      "loss": 0.4536,
+      "step": 2202
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485472953310536e-05,
+      "loss": 0.5146,
+      "step": 2203
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4850119792240047e-05,
+      "loss": 0.4524,
+      "step": 2204
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4845508703326504e-05,
+      "loss": 0.4683,
+      "step": 2205
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4840896267646346e-05,
+      "loss": 0.4261,
+      "step": 2206
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4836282486481557e-05,
+      "loss": 0.4862,
+      "step": 2207
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.48316673611145e-05,
+      "loss": 0.4453,
+      "step": 2208
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4827050892827903e-05,
+      "loss": 0.5038,
+      "step": 2209
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4822433082904879e-05,
+      "loss": 0.3974,
+      "step": 2210
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4817813932628904e-05,
+      "loss": 0.442,
+      "step": 2211
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4813193443283834e-05,
+      "loss": 0.4054,
+      "step": 2212
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808571616153887e-05,
+      "loss": 0.4418,
+      "step": 2213
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4803948452523664e-05,
+      "loss": 0.4763,
+      "step": 2214
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4799323953678134e-05,
+      "loss": 0.4765,
+      "step": 2215
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4794698120902634e-05,
+      "loss": 0.4642,
+      "step": 2216
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790070955482872e-05,
+      "loss": 0.5138,
+      "step": 2217
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4785442458704928e-05,
+      "loss": 0.4309,
+      "step": 2218
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4780812631855257e-05,
+      "loss": 0.4438,
+      "step": 2219
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4776181476220672e-05,
+      "loss": 0.4097,
+      "step": 2220
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4771548993088367e-05,
+      "loss": 0.443,
+      "step": 2221
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.47669151837459e-05,
+      "loss": 0.4585,
+      "step": 2222
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4762280049481193e-05,
+      "loss": 0.4591,
+      "step": 2223
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4757643591582544e-05,
+      "loss": 0.4078,
+      "step": 2224
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475300581133862e-05,
+      "loss": 0.4484,
+      "step": 2225
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4748366710038439e-05,
+      "loss": 0.4295,
+      "step": 2226
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4743726288971414e-05,
+      "loss": 0.4113,
+      "step": 2227
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4739084549427297e-05,
+      "loss": 0.5556,
+      "step": 2228
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4734441492696224e-05,
+      "loss": 0.4251,
+      "step": 2229
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4729797120068693e-05,
+      "loss": 0.4801,
+      "step": 2230
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4725151432835564e-05,
+      "loss": 0.4692,
+      "step": 2231
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4720504432288067e-05,
+      "loss": 0.4561,
+      "step": 2232
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715856119717792e-05,
+      "loss": 0.4828,
+      "step": 2233
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4711206496416703e-05,
+      "loss": 0.4927,
+      "step": 2234
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4706555563677114e-05,
+      "loss": 0.4648,
+      "step": 2235
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701903322791716e-05,
+      "loss": 0.4581,
+      "step": 2236
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4697249775053557e-05,
+      "loss": 0.5049,
+      "step": 2237
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4692594921756045e-05,
+      "loss": 0.4587,
+      "step": 2238
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4687938764192967e-05,
+      "loss": 0.3852,
+      "step": 2239
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4683281303658454e-05,
+      "loss": 0.4837,
+      "step": 2240
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4678622541447003e-05,
+      "loss": 0.4493,
+      "step": 2241
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4673962478853483e-05,
+      "loss": 0.4288,
+      "step": 2242
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4669301117173112e-05,
+      "loss": 0.4082,
+      "step": 2243
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4664638457701479e-05,
+      "loss": 0.4038,
+      "step": 2244
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465997450173453e-05,
+      "loss": 0.4603,
+      "step": 2245
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4655309250568563e-05,
+      "loss": 0.43,
+      "step": 2246
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4650642705500249e-05,
+      "loss": 0.4502,
+      "step": 2247
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645974867826614e-05,
+      "loss": 0.4331,
+      "step": 2248
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4641305738845043e-05,
+      "loss": 0.4527,
+      "step": 2249
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4636635319853274e-05,
+      "loss": 0.4568,
+      "step": 2250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4631963612149415e-05,
+      "loss": 0.4713,
+      "step": 2251
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4627290617031924e-05,
+      "loss": 0.4492,
+      "step": 2252
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4622616335799617e-05,
+      "loss": 0.4449,
+      "step": 2253
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4617940769751676e-05,
+      "loss": 0.4374,
+      "step": 2254
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4613263920187625e-05,
+      "loss": 0.4187,
+      "step": 2255
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4608585788407354e-05,
+      "loss": 0.4808,
+      "step": 2256
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4603906375711119e-05,
+      "loss": 0.4758,
+      "step": 2257
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459922568339951e-05,
+      "loss": 0.4957,
+      "step": 2258
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4594543712773488e-05,
+      "loss": 0.5038,
+      "step": 2259
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4589860465134368e-05,
+      "loss": 0.4563,
+      "step": 2260
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4585175941783818e-05,
+      "loss": 0.4572,
+      "step": 2261
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4580490144023856e-05,
+      "loss": 0.5265,
+      "step": 2262
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.457580307315686e-05,
+      "loss": 0.4662,
+      "step": 2263
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4571114730485562e-05,
+      "loss": 0.4675,
+      "step": 2264
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4566425117313043e-05,
+      "loss": 0.5395,
+      "step": 2265
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4561734234942741e-05,
+      "loss": 0.5186,
+      "step": 2266
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4557042084678449e-05,
+      "loss": 0.4703,
+      "step": 2267
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4552348667824304e-05,
+      "loss": 0.4947,
+      "step": 2268
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.45476539856848e-05,
+      "loss": 0.4643,
+      "step": 2269
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4542958039564785e-05,
+      "loss": 0.4557,
+      "step": 2270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4538260830769457e-05,
+      "loss": 0.524,
+      "step": 2271
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4533562360604366e-05,
+      "loss": 0.4636,
+      "step": 2272
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528862630375403e-05,
+      "loss": 0.4507,
+      "step": 2273
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524161641388818e-05,
+      "loss": 0.4372,
+      "step": 2274
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4519459394951216e-05,
+      "loss": 0.4846,
+      "step": 2275
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4514755892369543e-05,
+      "loss": 0.3997,
+      "step": 2276
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4510051134951091e-05,
+      "loss": 0.4475,
+      "step": 2277
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4505345124003512e-05,
+      "loss": 0.5009,
+      "step": 2278
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4500637860834798e-05,
+      "loss": 0.4622,
+      "step": 2279
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4495929346753292e-05,
+      "loss": 0.4693,
+      "step": 2280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4491219583067684e-05,
+      "loss": 0.4357,
+      "step": 2281
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4486508571087012e-05,
+      "loss": 0.4767,
+      "step": 2282
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448179631212066e-05,
+      "loss": 0.3987,
+      "step": 2283
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4477082807478358e-05,
+      "loss": 0.4824,
+      "step": 2284
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4472368058470183e-05,
+      "loss": 0.4351,
+      "step": 2285
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446765206640656e-05,
+      "loss": 0.4996,
+      "step": 2286
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4462934832598259e-05,
+      "loss": 0.4236,
+      "step": 2287
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.445821635835639e-05,
+      "loss": 0.4968,
+      "step": 2288
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4453496644992413e-05,
+      "loss": 0.4553,
+      "step": 2289
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4448775693818129e-05,
+      "loss": 0.4499,
+      "step": 2290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444405350614569e-05,
+      "loss": 0.4818,
+      "step": 2291
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4439330083287582e-05,
+      "loss": 0.4682,
+      "step": 2292
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443460542655664e-05,
+      "loss": 0.4273,
+      "step": 2293
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429879537266046e-05,
+      "loss": 0.4145,
+      "step": 2294
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442515241672931e-05,
+      "loss": 0.525,
+      "step": 2295
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4420424066260296e-05,
+      "loss": 0.4637,
+      "step": 2296
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4415694487173216e-05,
+      "loss": 0.4802,
+      "step": 2297
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4410963680782606e-05,
+      "loss": 0.471,
+      "step": 2298
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4406231648403354e-05,
+      "loss": 0.4734,
+      "step": 2299
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440149839135069e-05,
+      "loss": 0.4813,
+      "step": 2300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4396763910940182e-05,
+      "loss": 0.4532,
+      "step": 2301
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439202820848773e-05,
+      "loss": 0.5313,
+      "step": 2302
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4387291285309583e-05,
+      "loss": 0.4351,
+      "step": 2303
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4382553142722336e-05,
+      "loss": 0.4384,
+      "step": 2304
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4377813782042901e-05,
+      "loss": 0.4451,
+      "step": 2305
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373073204588556e-05,
+      "loss": 0.5116,
+      "step": 2306
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4368331411676892e-05,
+      "loss": 0.4386,
+      "step": 2307
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436358840462585e-05,
+      "loss": 0.4831,
+      "step": 2308
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4358844184753713e-05,
+      "loss": 0.4347,
+      "step": 2309
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4354098753379089e-05,
+      "loss": 0.5188,
+      "step": 2310
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4349352111820931e-05,
+      "loss": 0.456,
+      "step": 2311
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4344604261398529e-05,
+      "loss": 0.5104,
+      "step": 2312
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4339855203431505e-05,
+      "loss": 0.4015,
+      "step": 2313
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4335104939239814e-05,
+      "loss": 0.5491,
+      "step": 2314
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4330353470143756e-05,
+      "loss": 0.4509,
+      "step": 2315
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4325600797463958e-05,
+      "loss": 0.4782,
+      "step": 2316
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.432084692252138e-05,
+      "loss": 0.4347,
+      "step": 2317
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4316091846637327e-05,
+      "loss": 0.4359,
+      "step": 2318
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4311335571133427e-05,
+      "loss": 0.4776,
+      "step": 2319
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430657809733164e-05,
+      "loss": 0.4114,
+      "step": 2320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4301819426554274e-05,
+      "loss": 0.4561,
+      "step": 2321
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4297059560123952e-05,
+      "loss": 0.448,
+      "step": 2322
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4292298499363639e-05,
+      "loss": 0.4766,
+      "step": 2323
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4287536245596632e-05,
+      "loss": 0.4499,
+      "step": 2324
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4282772800146556e-05,
+      "loss": 0.4765,
+      "step": 2325
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427800816433737e-05,
+      "loss": 0.499,
+      "step": 2326
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427324233949336e-05,
+      "loss": 0.5181,
+      "step": 2327
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4268475326939148e-05,
+      "loss": 0.4872,
+      "step": 2328
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4263707127999685e-05,
+      "loss": 0.5276,
+      "step": 2329
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4258937744000244e-05,
+      "loss": 0.4955,
+      "step": 2330
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425416717626644e-05,
+      "loss": 0.425,
+      "step": 2331
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4249395426124203e-05,
+      "loss": 0.4388,
+      "step": 2332
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4244622494899807e-05,
+      "loss": 0.471,
+      "step": 2333
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4239848383919843e-05,
+      "loss": 0.4179,
+      "step": 2334
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423507309451123e-05,
+      "loss": 0.4682,
+      "step": 2335
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4230296628001224e-05,
+      "loss": 0.5028,
+      "step": 2336
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4225518985717396e-05,
+      "loss": 0.4632,
+      "step": 2337
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4220740168987654e-05,
+      "loss": 0.4846,
+      "step": 2338
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421596017914023e-05,
+      "loss": 0.4323,
+      "step": 2339
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4211179017503671e-05,
+      "loss": 0.4815,
+      "step": 2340
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4206396685406864e-05,
+      "loss": 0.4373,
+      "step": 2341
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4201613184179022e-05,
+      "loss": 0.4632,
+      "step": 2342
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4196828515149673e-05,
+      "loss": 0.4134,
+      "step": 2343
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4192042679648668e-05,
+      "loss": 0.4427,
+      "step": 2344
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4187255679006194e-05,
+      "loss": 0.4837,
+      "step": 2345
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4182467514552757e-05,
+      "loss": 0.4909,
+      "step": 2346
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4177678187619181e-05,
+      "loss": 0.4759,
+      "step": 2347
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4172887699536622e-05,
+      "loss": 0.4673,
+      "step": 2348
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4168096051636551e-05,
+      "loss": 0.4891,
+      "step": 2349
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4163303245250762e-05,
+      "loss": 0.4576,
+      "step": 2350
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4158509281711379e-05,
+      "loss": 0.4798,
+      "step": 2351
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4153714162350839e-05,
+      "loss": 0.5362,
+      "step": 2352
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41489178885019e-05,
+      "loss": 0.44,
+      "step": 2353
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4144120461497655e-05,
+      "loss": 0.4622,
+      "step": 2354
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4139321882671495e-05,
+      "loss": 0.5207,
+      "step": 2355
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134522153357152e-05,
+      "loss": 0.4406,
+      "step": 2356
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.412972127488866e-05,
+      "loss": 0.4087,
+      "step": 2357
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4124919248600388e-05,
+      "loss": 0.458,
+      "step": 2358
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4120116075827012e-05,
+      "loss": 0.4834,
+      "step": 2359
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4115311757903536e-05,
+      "loss": 0.4493,
+      "step": 2360
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4110506296165278e-05,
+      "loss": 0.4662,
+      "step": 2361
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4105699691947869e-05,
+      "loss": 0.4698,
+      "step": 2362
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4100891946587267e-05,
+      "loss": 0.4294,
+      "step": 2363
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.409608306141974e-05,
+      "loss": 0.4332,
+      "step": 2364
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4091273037781879e-05,
+      "loss": 0.478,
+      "step": 2365
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4086461877010585e-05,
+      "loss": 0.4881,
+      "step": 2366
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4081649580443076e-05,
+      "loss": 0.4686,
+      "step": 2367
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4076836149416889e-05,
+      "loss": 0.4989,
+      "step": 2368
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4072021585269877e-05,
+      "loss": 0.4932,
+      "step": 2369
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4067205889340207e-05,
+      "loss": 0.4906,
+      "step": 2370
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4062389062966357e-05,
+      "loss": 0.4433,
+      "step": 2371
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405757110748712e-05,
+      "loss": 0.4265,
+      "step": 2372
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4052752024241608e-05,
+      "loss": 0.4617,
+      "step": 2373
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4047931814569239e-05,
+      "loss": 0.4161,
+      "step": 2374
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.404311047980975e-05,
+      "loss": 0.4663,
+      "step": 2375
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403828802130319e-05,
+      "loss": 0.4309,
+      "step": 2376
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4033464440389917e-05,
+      "loss": 0.4772,
+      "step": 2377
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4028639738410604e-05,
+      "loss": 0.4733,
+      "step": 2378
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4023813916706232e-05,
+      "loss": 0.4627,
+      "step": 2379
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4018986976618103e-05,
+      "loss": 0.4371,
+      "step": 2380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4014158919487814e-05,
+      "loss": 0.4686,
+      "step": 2381
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4009329746657287e-05,
+      "loss": 0.462,
+      "step": 2382
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4004499459468745e-05,
+      "loss": 0.4765,
+      "step": 2383
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3999668059264723e-05,
+      "loss": 0.4346,
+      "step": 2384
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399483554738807e-05,
+      "loss": 0.4958,
+      "step": 2385
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.399000192518194e-05,
+      "loss": 0.5293,
+      "step": 2386
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398516719398979e-05,
+      "loss": 0.4786,
+      "step": 2387
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3980331355155401e-05,
+      "loss": 0.4985,
+      "step": 2388
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3975494410022843e-05,
+      "loss": 0.4484,
+      "step": 2389
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3970656359936508e-05,
+      "loss": 0.442,
+      "step": 2390
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3965817206241085e-05,
+      "loss": 0.3886,
+      "step": 2391
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396097695028158e-05,
+      "loss": 0.477,
+      "step": 2392
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3956135593403293e-05,
+      "loss": 0.471,
+      "step": 2393
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395129313695184e-05,
+      "loss": 0.4167,
+      "step": 2394
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394644958227314e-05,
+      "loss": 0.4565,
+      "step": 2395
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3941604930713417e-05,
+      "loss": 0.444,
+      "step": 2396
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3936759183619197e-05,
+      "loss": 0.5007,
+      "step": 2397
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3931912342337314e-05,
+      "loss": 0.4904,
+      "step": 2398
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3927064408214905e-05,
+      "loss": 0.4396,
+      "step": 2399
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3922215382599413e-05,
+      "loss": 0.416,
+      "step": 2400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3917365266838581e-05,
+      "loss": 0.4264,
+      "step": 2401
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912514062280448e-05,
+      "loss": 0.4645,
+      "step": 2402
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3907661770273377e-05,
+      "loss": 0.4585,
+      "step": 2403
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3902808392166013e-05,
+      "loss": 0.4761,
+      "step": 2404
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3897953929307307e-05,
+      "loss": 0.5054,
+      "step": 2405
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3893098383046522e-05,
+      "loss": 0.5435,
+      "step": 2406
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.388824175473321e-05,
+      "loss": 0.5147,
+      "step": 2407
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3883384045717226e-05,
+      "loss": 0.4337,
+      "step": 2408
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387852525734873e-05,
+      "loss": 0.4725,
+      "step": 2409
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3873665390978184e-05,
+      "loss": 0.4306,
+      "step": 2410
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3868804447956335e-05,
+      "loss": 0.4185,
+      "step": 2411
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.386394242963425e-05,
+      "loss": 0.4579,
+      "step": 2412
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3859079337363279e-05,
+      "loss": 0.4812,
+      "step": 2413
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3854215172495073e-05,
+      "loss": 0.4161,
+      "step": 2414
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3849349936381593e-05,
+      "loss": 0.4294,
+      "step": 2415
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.384448363037508e-05,
+      "loss": 0.4694,
+      "step": 2416
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3839616255828086e-05,
+      "loss": 0.5043,
+      "step": 2417
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3834747814093452e-05,
+      "loss": 0.4647,
+      "step": 2418
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382987830652432e-05,
+      "loss": 0.4053,
+      "step": 2419
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3825007734474125e-05,
+      "loss": 0.4023,
+      "step": 2420
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3820136099296604e-05,
+      "loss": 0.477,
+      "step": 2421
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3815263402345785e-05,
+      "loss": 0.4279,
+      "step": 2422
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3810389644975983e-05,
+      "loss": 0.4799,
+      "step": 2423
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3805514828541827e-05,
+      "loss": 0.4724,
+      "step": 2424
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3800638954398222e-05,
+      "loss": 0.4819,
+      "step": 2425
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3795762023900377e-05,
+      "loss": 0.4777,
+      "step": 2426
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3790884038403796e-05,
+      "loss": 0.451,
+      "step": 2427
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3786004999264266e-05,
+      "loss": 0.4398,
+      "step": 2428
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3781124907837873e-05,
+      "loss": 0.4876,
+      "step": 2429
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3776243765481001e-05,
+      "loss": 0.5101,
+      "step": 2430
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3771361573550317e-05,
+      "loss": 0.4668,
+      "step": 2431
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3766478333402784e-05,
+      "loss": 0.4638,
+      "step": 2432
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3761594046395656e-05,
+      "loss": 0.4492,
+      "step": 2433
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3756708713886478e-05,
+      "loss": 0.4797,
+      "step": 2434
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3751822337233083e-05,
+      "loss": 0.4452,
+      "step": 2435
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.37469349177936e-05,
+      "loss": 0.5086,
+      "step": 2436
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3742046456926444e-05,
+      "loss": 0.5338,
+      "step": 2437
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3737156955990316e-05,
+      "loss": 0.4751,
+      "step": 2438
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3732266416344213e-05,
+      "loss": 0.4345,
+      "step": 2439
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3727374839347418e-05,
+      "loss": 0.4748,
+      "step": 2440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.37224822263595e-05,
+      "loss": 0.4911,
+      "step": 2441
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3717588578740317e-05,
+      "loss": 0.4655,
+      "step": 2442
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371269389785002e-05,
+      "loss": 0.476,
+      "step": 2443
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3707798185049037e-05,
+      "loss": 0.4881,
+      "step": 2444
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3702901441698093e-05,
+      "loss": 0.4379,
+      "step": 2445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3698003669158193e-05,
+      "loss": 0.4874,
+      "step": 2446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369310486879063e-05,
+      "loss": 0.443,
+      "step": 2447
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3688205041956982e-05,
+      "loss": 0.4442,
+      "step": 2448
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3683304190019113e-05,
+      "loss": 0.4281,
+      "step": 2449
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3678402314339171e-05,
+      "loss": 0.4162,
+      "step": 2450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3673499416279591e-05,
+      "loss": 0.4614,
+      "step": 2451
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3668595497203091e-05,
+      "loss": 0.4176,
+      "step": 2452
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3663690558472671e-05,
+      "loss": 0.4633,
+      "step": 2453
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3658784601451614e-05,
+      "loss": 0.452,
+      "step": 2454
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3653877627503492e-05,
+      "loss": 0.4188,
+      "step": 2455
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364896963799215e-05,
+      "loss": 0.4115,
+      "step": 2456
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3644060634281725e-05,
+      "loss": 0.4987,
+      "step": 2457
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3639150617736627e-05,
+      "loss": 0.4903,
+      "step": 2458
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.363423958972156e-05,
+      "loss": 0.4768,
+      "step": 2459
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3629327551601491e-05,
+      "loss": 0.4595,
+      "step": 2460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3624414504741688e-05,
+      "loss": 0.4285,
+      "step": 2461
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361950045050768e-05,
+      "loss": 0.467,
+      "step": 2462
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3614585390265291e-05,
+      "loss": 0.535,
+      "step": 2463
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3609669325380618e-05,
+      "loss": 0.462,
+      "step": 2464
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3604752257220036e-05,
+      "loss": 0.4304,
+      "step": 2465
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3599834187150203e-05,
+      "loss": 0.4958,
+      "step": 2466
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3594915116538054e-05,
+      "loss": 0.4843,
+      "step": 2467
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35899950467508e-05,
+      "loss": 0.4187,
+      "step": 2468
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3585073979155927e-05,
+      "loss": 0.4482,
+      "step": 2469
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3580151915121213e-05,
+      "loss": 0.4536,
+      "step": 2470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3575228856014693e-05,
+      "loss": 0.4654,
+      "step": 2471
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3570304803204693e-05,
+      "loss": 0.4686,
+      "step": 2472
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3565379758059805e-05,
+      "loss": 0.4334,
+      "step": 2473
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3560453721948912e-05,
+      "loss": 0.4739,
+      "step": 2474
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3555526696241152e-05,
+      "loss": 0.4535,
+      "step": 2475
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3550598682305954e-05,
+      "loss": 0.4564,
+      "step": 2476
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3545669681513016e-05,
+      "loss": 0.4937,
+      "step": 2477
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3540739695232308e-05,
+      "loss": 0.4541,
+      "step": 2478
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3535808724834081e-05,
+      "loss": 0.4526,
+      "step": 2479
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.353087677168885e-05,
+      "loss": 0.4182,
+      "step": 2480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352594383716741e-05,
+      "loss": 0.481,
+      "step": 2481
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3521009922640827e-05,
+      "loss": 0.4636,
+      "step": 2482
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.351607502948044e-05,
+      "loss": 0.3945,
+      "step": 2483
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3511139159057858e-05,
+      "loss": 0.4575,
+      "step": 2484
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3506202312744962e-05,
+      "loss": 0.4487,
+      "step": 2485
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3501264491913909e-05,
+      "loss": 0.4713,
+      "step": 2486
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3496325697937119e-05,
+      "loss": 0.5062,
+      "step": 2487
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3491385932187287e-05,
+      "loss": 0.462,
+      "step": 2488
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3486445196037381e-05,
+      "loss": 0.4362,
+      "step": 2489
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348150349086063e-05,
+      "loss": 0.4372,
+      "step": 2490
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347656081803054e-05,
+      "loss": 0.4249,
+      "step": 2491
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3471617178920888e-05,
+      "loss": 0.4574,
+      "step": 2492
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3466672574905704e-05,
+      "loss": 0.5012,
+      "step": 2493
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.346172700735931e-05,
+      "loss": 0.4258,
+      "step": 2494
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3456780477656275e-05,
+      "loss": 0.4285,
+      "step": 2495
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3451832987171445e-05,
+      "loss": 0.4553,
+      "step": 2496
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3446884537279932e-05,
+      "loss": 0.4757,
+      "step": 2497
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3441935129357115e-05,
+      "loss": 0.4592,
+      "step": 2498
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3436984764778638e-05,
+      "loss": 0.4718,
+      "step": 2499
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3432033444920413e-05,
+      "loss": 0.5401,
+      "step": 2500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3427081171158613e-05,
+      "loss": 0.4028,
+      "step": 2501
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3422127944869681e-05,
+      "loss": 0.5263,
+      "step": 2502
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417173767430321e-05,
+      "loss": 0.3751,
+      "step": 2503
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3412218640217504e-05,
+      "loss": 0.4675,
+      "step": 2504
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3407262564608465e-05,
+      "loss": 0.4591,
+      "step": 2505
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34023055419807e-05,
+      "loss": 0.4101,
+      "step": 2506
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3397347573711975e-05,
+      "loss": 0.4804,
+      "step": 2507
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3392388661180303e-05,
+      "loss": 0.4793,
+      "step": 2508
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3387428805763982e-05,
+      "loss": 0.4373,
+      "step": 2509
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3382468008841554e-05,
+      "loss": 0.4222,
+      "step": 2510
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3377506271791829e-05,
+      "loss": 0.4567,
+      "step": 2511
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3372543595993883e-05,
+      "loss": 0.4544,
+      "step": 2512
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3367579982827041e-05,
+      "loss": 0.4721,
+      "step": 2513
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33626154336709e-05,
+      "loss": 0.4439,
+      "step": 2514
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3357649949905312e-05,
+      "loss": 0.4177,
+      "step": 2515
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3352683532910391e-05,
+      "loss": 0.4503,
+      "step": 2516
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3347716184066509e-05,
+      "loss": 0.4732,
+      "step": 2517
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342747904754292e-05,
+      "loss": 0.4716,
+      "step": 2518
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3337778696354637e-05,
+      "loss": 0.5206,
+      "step": 2519
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3332808560248687e-05,
+      "loss": 0.481,
+      "step": 2520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3327837497817849e-05,
+      "loss": 0.5077,
+      "step": 2521
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3322865510443787e-05,
+      "loss": 0.4173,
+      "step": 2522
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3317892599508417e-05,
+      "loss": 0.4591,
+      "step": 2523
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3312918766393925e-05,
+      "loss": 0.3993,
+      "step": 2524
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3307944012482737e-05,
+      "loss": 0.5089,
+      "step": 2525
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330296833915754e-05,
+      "loss": 0.5028,
+      "step": 2526
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3297991747801288e-05,
+      "loss": 0.4219,
+      "step": 2527
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3293014239797173e-05,
+      "loss": 0.4456,
+      "step": 2528
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3288035816528653e-05,
+      "loss": 0.5068,
+      "step": 2529
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3283056479379434e-05,
+      "loss": 0.5017,
+      "step": 2530
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3278076229733486e-05,
+      "loss": 0.4741,
+      "step": 2531
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3273095068975014e-05,
+      "loss": 0.3911,
+      "step": 2532
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32681129984885e-05,
+      "loss": 0.4508,
+      "step": 2533
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3263130019658661e-05,
+      "loss": 0.4755,
+      "step": 2534
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3258146133870473e-05,
+      "loss": 0.5374,
+      "step": 2535
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3253161342509162e-05,
+      "loss": 0.4497,
+      "step": 2536
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.324817564696021e-05,
+      "loss": 0.4132,
+      "step": 2537
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3243189048609344e-05,
+      "loss": 0.4827,
+      "step": 2538
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3238201548842546e-05,
+      "loss": 0.4275,
+      "step": 2539
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.323321314904605e-05,
+      "loss": 0.4345,
+      "step": 2540
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3228223850606335e-05,
+      "loss": 0.4758,
+      "step": 2541
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3223233654910136e-05,
+      "loss": 0.4248,
+      "step": 2542
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3218242563344432e-05,
+      "loss": 0.4401,
+      "step": 2543
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3213250577296453e-05,
+      "loss": 0.4617,
+      "step": 2544
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3208257698153677e-05,
+      "loss": 0.4153,
+      "step": 2545
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3203263927303835e-05,
+      "loss": 0.4535,
+      "step": 2546
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3198269266134895e-05,
+      "loss": 0.5016,
+      "step": 2547
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3193273716035084e-05,
+      "loss": 0.4187,
+      "step": 2548
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3188277278392873e-05,
+      "loss": 0.5192,
+      "step": 2549
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3183279954596974e-05,
+      "loss": 0.4195,
+      "step": 2550
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317828174603635e-05,
+      "loss": 0.4754,
+      "step": 2551
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3173282654100213e-05,
+      "loss": 0.4149,
+      "step": 2552
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316828268017801e-05,
+      "loss": 0.4586,
+      "step": 2553
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3163281825659443e-05,
+      "loss": 0.4355,
+      "step": 2554
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3158280091934457e-05,
+      "loss": 0.4465,
+      "step": 2555
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.315327748039324e-05,
+      "loss": 0.4571,
+      "step": 2556
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3148273992426219e-05,
+      "loss": 0.4886,
+      "step": 2557
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314326962942408e-05,
+      "loss": 0.4755,
+      "step": 2558
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3138264392777732e-05,
+      "loss": 0.5226,
+      "step": 2559
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3133258283878339e-05,
+      "loss": 0.4237,
+      "step": 2560
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312825130411731e-05,
+      "loss": 0.4325,
+      "step": 2561
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3123243454886284e-05,
+      "loss": 0.4841,
+      "step": 2562
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311823473757715e-05,
+      "loss": 0.4089,
+      "step": 2563
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3113225153582044e-05,
+      "loss": 0.4284,
+      "step": 2564
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108214704293324e-05,
+      "loss": 0.418,
+      "step": 2565
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3103203391103611e-05,
+      "loss": 0.5275,
+      "step": 2566
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3098191215405752e-05,
+      "loss": 0.4451,
+      "step": 2567
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3093178178592838e-05,
+      "loss": 0.4558,
+      "step": 2568
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3088164282058196e-05,
+      "loss": 0.5288,
+      "step": 2569
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3083149527195393e-05,
+      "loss": 0.3893,
+      "step": 2570
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3078133915398245e-05,
+      "loss": 0.5091,
+      "step": 2571
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3073117448060786e-05,
+      "loss": 0.4875,
+      "step": 2572
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3068100126577308e-05,
+      "loss": 0.4435,
+      "step": 2573
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306308195234233e-05,
+      "loss": 0.45,
+      "step": 2574
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3058062926750598e-05,
+      "loss": 0.4724,
+      "step": 2575
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3053043051197126e-05,
+      "loss": 0.4786,
+      "step": 2576
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3048022327077127e-05,
+      "loss": 0.4564,
+      "step": 2577
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3043000755786074e-05,
+      "loss": 0.5046,
+      "step": 2578
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3037978338719674e-05,
+      "loss": 0.4471,
+      "step": 2579
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3032955077273856e-05,
+      "loss": 0.4395,
+      "step": 2580
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3027930972844794e-05,
+      "loss": 0.5071,
+      "step": 2581
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3022906026828893e-05,
+      "loss": 0.4248,
+      "step": 2582
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3017880240622795e-05,
+      "loss": 0.4175,
+      "step": 2583
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3012853615623368e-05,
+      "loss": 0.5128,
+      "step": 2584
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3007826153227725e-05,
+      "loss": 0.3959,
+      "step": 2585
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3002797854833203e-05,
+      "loss": 0.4277,
+      "step": 2586
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2997768721837365e-05,
+      "loss": 0.4554,
+      "step": 2587
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2992738755638029e-05,
+      "loss": 0.4465,
+      "step": 2588
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2987707957633222e-05,
+      "loss": 0.5304,
+      "step": 2589
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2982676329221206e-05,
+      "loss": 0.4707,
+      "step": 2590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2977643871800487e-05,
+      "loss": 0.495,
+      "step": 2591
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2972610586769785e-05,
+      "loss": 0.3808,
+      "step": 2592
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2967576475528061e-05,
+      "loss": 0.4611,
+      "step": 2593
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2962541539474505e-05,
+      "loss": 0.4349,
+      "step": 2594
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2957505780008526e-05,
+      "loss": 0.3953,
+      "step": 2595
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2952469198529773e-05,
+      "loss": 0.4721,
+      "step": 2596
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.294743179643812e-05,
+      "loss": 0.4364,
+      "step": 2597
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2942393575133671e-05,
+      "loss": 0.4111,
+      "step": 2598
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2937354536016749e-05,
+      "loss": 0.4448,
+      "step": 2599
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2932314680487914e-05,
+      "loss": 0.4375,
+      "step": 2600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2927274009947954e-05,
+      "loss": 0.4654,
+      "step": 2601
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292223252579787e-05,
+      "loss": 0.4667,
+      "step": 2602
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2917190229438908e-05,
+      "loss": 0.4613,
+      "step": 2603
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2912147122272523e-05,
+      "loss": 0.4754,
+      "step": 2604
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2907103205700402e-05,
+      "loss": 0.456,
+      "step": 2605
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2902058481124465e-05,
+      "loss": 0.4474,
+      "step": 2606
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289701294994684e-05,
+      "loss": 0.5034,
+      "step": 2607
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2891966613569896e-05,
+      "loss": 0.4868,
+      "step": 2608
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2886919473396212e-05,
+      "loss": 0.3748,
+      "step": 2609
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2881871530828597e-05,
+      "loss": 0.4609,
+      "step": 2610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2876822787270084e-05,
+      "loss": 0.4601,
+      "step": 2611
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2871773244123929e-05,
+      "loss": 0.4438,
+      "step": 2612
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2866722902793604e-05,
+      "loss": 0.5009,
+      "step": 2613
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2861671764682808e-05,
+      "loss": 0.3691,
+      "step": 2614
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2856619831195462e-05,
+      "loss": 0.4856,
+      "step": 2615
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2851567103735709e-05,
+      "loss": 0.4402,
+      "step": 2616
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2846513583707902e-05,
+      "loss": 0.4452,
+      "step": 2617
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2841459272516632e-05,
+      "loss": 0.4531,
+      "step": 2618
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2836404171566693e-05,
+      "loss": 0.4676,
+      "step": 2619
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283134828226311e-05,
+      "loss": 0.4557,
+      "step": 2620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2826291606011123e-05,
+      "loss": 0.4899,
+      "step": 2621
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282123414421619e-05,
+      "loss": 0.4883,
+      "step": 2622
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2816175898283986e-05,
+      "loss": 0.4862,
+      "step": 2623
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2811116869620408e-05,
+      "loss": 0.4643,
+      "step": 2624
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2806057059631573e-05,
+      "loss": 0.4242,
+      "step": 2625
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2800996469723804e-05,
+      "loss": 0.4943,
+      "step": 2626
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279593510130365e-05,
+      "loss": 0.4528,
+      "step": 2627
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2790872955777873e-05,
+      "loss": 0.4899,
+      "step": 2628
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2785810034553455e-05,
+      "loss": 0.4336,
+      "step": 2629
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2780746339037584e-05,
+      "loss": 0.4575,
+      "step": 2630
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277568187063768e-05,
+      "loss": 0.4032,
+      "step": 2631
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2770616630761356e-05,
+      "loss": 0.3948,
+      "step": 2632
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276555062081646e-05,
+      "loss": 0.4042,
+      "step": 2633
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2760483842211042e-05,
+      "loss": 0.4143,
+      "step": 2634
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2755416296353369e-05,
+      "loss": 0.469,
+      "step": 2635
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2750347984651916e-05,
+      "loss": 0.4738,
+      "step": 2636
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2745278908515382e-05,
+      "loss": 0.46,
+      "step": 2637
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274020906935267e-05,
+      "loss": 0.4132,
+      "step": 2638
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2735138468572899e-05,
+      "loss": 0.3772,
+      "step": 2639
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2730067107585395e-05,
+      "loss": 0.4695,
+      "step": 2640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27249949877997e-05,
+      "loss": 0.4486,
+      "step": 2641
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2719922110625564e-05,
+      "loss": 0.4483,
+      "step": 2642
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2714848477472952e-05,
+      "loss": 0.5316,
+      "step": 2643
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2709774089752032e-05,
+      "loss": 0.5511,
+      "step": 2644
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2704698948873186e-05,
+      "loss": 0.5159,
+      "step": 2645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2699623056247008e-05,
+      "loss": 0.4455,
+      "step": 2646
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2694546413284295e-05,
+      "loss": 0.4636,
+      "step": 2647
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2689469021396055e-05,
+      "loss": 0.4513,
+      "step": 2648
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2684390881993505e-05,
+      "loss": 0.48,
+      "step": 2649
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2679311996488071e-05,
+      "loss": 0.4132,
+      "step": 2650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.267423236629138e-05,
+      "loss": 0.4497,
+      "step": 2651
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2669151992815278e-05,
+      "loss": 0.3862,
+      "step": 2652
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2664070877471806e-05,
+      "loss": 0.5032,
+      "step": 2653
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2658989021673212e-05,
+      "loss": 0.4594,
+      "step": 2654
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2653906426831961e-05,
+      "loss": 0.4752,
+      "step": 2655
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264882309436071e-05,
+      "loss": 0.399,
+      "step": 2656
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2643739025672326e-05,
+      "loss": 0.4511,
+      "step": 2657
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2638654222179887e-05,
+      "loss": 0.4019,
+      "step": 2658
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2633568685296664e-05,
+      "loss": 0.4321,
+      "step": 2659
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2628482416436139e-05,
+      "loss": 0.4561,
+      "step": 2660
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2623395417011997e-05,
+      "loss": 0.4787,
+      "step": 2661
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2618307688438125e-05,
+      "loss": 0.4721,
+      "step": 2662
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2613219232128608e-05,
+      "loss": 0.4227,
+      "step": 2663
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2608130049497742e-05,
+      "loss": 0.419,
+      "step": 2664
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2603040141960024e-05,
+      "loss": 0.454,
+      "step": 2665
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2597949510930136e-05,
+      "loss": 0.4608,
+      "step": 2666
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2592858157822992e-05,
+      "loss": 0.4438,
+      "step": 2667
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2587766084053678e-05,
+      "loss": 0.465,
+      "step": 2668
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2582673291037488e-05,
+      "loss": 0.4147,
+      "step": 2669
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2577579780189932e-05,
+      "loss": 0.4283,
+      "step": 2670
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2572485552926693e-05,
+      "loss": 0.4214,
+      "step": 2671
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2567390610663677e-05,
+      "loss": 0.5383,
+      "step": 2672
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2562294954816974e-05,
+      "loss": 0.4912,
+      "step": 2673
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2557198586802874e-05,
+      "loss": 0.4318,
+      "step": 2674
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.255210150803787e-05,
+      "loss": 0.5119,
+      "step": 2675
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2547003719938652e-05,
+      "loss": 0.5163,
+      "step": 2676
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2541905223922109e-05,
+      "loss": 0.4586,
+      "step": 2677
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2536806021405315e-05,
+      "loss": 0.4534,
+      "step": 2678
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2531706113805553e-05,
+      "loss": 0.4123,
+      "step": 2679
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2526605502540299e-05,
+      "loss": 0.4691,
+      "step": 2680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2521504189027216e-05,
+      "loss": 0.4289,
+      "step": 2681
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2516402174684179e-05,
+      "loss": 0.4558,
+      "step": 2682
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2511299460929245e-05,
+      "loss": 0.4612,
+      "step": 2683
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250619604918066e-05,
+      "loss": 0.435,
+      "step": 2684
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2501091940856886e-05,
+      "loss": 0.4015,
+      "step": 2685
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2495987137376555e-05,
+      "loss": 0.4748,
+      "step": 2686
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2490881640158508e-05,
+      "loss": 0.497,
+      "step": 2687
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2485775450621772e-05,
+      "loss": 0.4638,
+      "step": 2688
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2480668570185563e-05,
+      "loss": 0.4425,
+      "step": 2689
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2475561000269298e-05,
+      "loss": 0.4381,
+      "step": 2690
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2470452742292582e-05,
+      "loss": 0.4152,
+      "step": 2691
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2465343797675211e-05,
+      "loss": 0.4101,
+      "step": 2692
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2460234167837166e-05,
+      "loss": 0.5,
+      "step": 2693
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2455123854198627e-05,
+      "loss": 0.4336,
+      "step": 2694
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2450012858179962e-05,
+      "loss": 0.5326,
+      "step": 2695
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2444901181201724e-05,
+      "loss": 0.4625,
+      "step": 2696
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2439788824684667e-05,
+      "loss": 0.4335,
+      "step": 2697
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2434675790049717e-05,
+      "loss": 0.4616,
+      "step": 2698
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2429562078717995e-05,
+      "loss": 0.4386,
+      "step": 2699
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2424447692110824e-05,
+      "loss": 0.3967,
+      "step": 2700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2419332631649691e-05,
+      "loss": 0.4652,
+      "step": 2701
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2414216898756288e-05,
+      "loss": 0.492,
+      "step": 2702
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2409100494852488e-05,
+      "loss": 0.48,
+      "step": 2703
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240398342136035e-05,
+      "loss": 0.4605,
+      "step": 2704
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2398865679702119e-05,
+      "loss": 0.5235,
+      "step": 2705
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2393747271300228e-05,
+      "loss": 0.4591,
+      "step": 2706
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2388628197577291e-05,
+      "loss": 0.4343,
+      "step": 2707
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238350845995611e-05,
+      "loss": 0.4543,
+      "step": 2708
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2378388059859673e-05,
+      "loss": 0.485,
+      "step": 2709
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2373266998711152e-05,
+      "loss": 0.4841,
+      "step": 2710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2368145277933895e-05,
+      "loss": 0.4625,
+      "step": 2711
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2363022898951442e-05,
+      "loss": 0.4789,
+      "step": 2712
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2357899863187515e-05,
+      "loss": 0.41,
+      "step": 2713
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2352776172066016e-05,
+      "loss": 0.4041,
+      "step": 2714
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2347651827011031e-05,
+      "loss": 0.4175,
+      "step": 2715
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2342526829446826e-05,
+      "loss": 0.4466,
+      "step": 2716
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2337401180797845e-05,
+      "loss": 0.4573,
+      "step": 2717
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2332274882488724e-05,
+      "loss": 0.4058,
+      "step": 2718
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232714793594427e-05,
+      "loss": 0.442,
+      "step": 2719
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232202034258947e-05,
+      "loss": 0.4421,
+      "step": 2720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2316892103849496e-05,
+      "loss": 0.4927,
+      "step": 2721
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.23117632211497e-05,
+      "loss": 0.4217,
+      "step": 2722
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2306633695915602e-05,
+      "loss": 0.4397,
+      "step": 2723
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2301503529572914e-05,
+      "loss": 0.4158,
+      "step": 2724
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2296372723547522e-05,
+      "loss": 0.4351,
+      "step": 2725
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2291241279265484e-05,
+      "loss": 0.4542,
+      "step": 2726
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2286109198153042e-05,
+      "loss": 0.4701,
+      "step": 2727
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280976481636614e-05,
+      "loss": 0.4106,
+      "step": 2728
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2275843131142795e-05,
+      "loss": 0.4668,
+      "step": 2729
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.227070914809835e-05,
+      "loss": 0.4114,
+      "step": 2730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2265574533930224e-05,
+      "loss": 0.4438,
+      "step": 2731
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2260439290065544e-05,
+      "loss": 0.4276,
+      "step": 2732
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2255303417931602e-05,
+      "loss": 0.4351,
+      "step": 2733
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225016691895587e-05,
+      "loss": 0.4596,
+      "step": 2734
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2245029794565992e-05,
+      "loss": 0.4127,
+      "step": 2735
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2239892046189787e-05,
+      "loss": 0.4889,
+      "step": 2736
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2234753675255248e-05,
+      "loss": 0.4147,
+      "step": 2737
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2229614683190536e-05,
+      "loss": 0.4718,
+      "step": 2738
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2224475071423993e-05,
+      "loss": 0.469,
+      "step": 2739
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221933484138413e-05,
+      "loss": 0.4749,
+      "step": 2740
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2214193994499623e-05,
+      "loss": 0.4688,
+      "step": 2741
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.220905253219933e-05,
+      "loss": 0.4729,
+      "step": 2742
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2203910455912273e-05,
+      "loss": 0.3789,
+      "step": 2743
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2198767767067652e-05,
+      "loss": 0.5743,
+      "step": 2744
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2193624467094825e-05,
+      "loss": 0.5372,
+      "step": 2745
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218848055742333e-05,
+      "loss": 0.4606,
+      "step": 2746
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2183336039482873e-05,
+      "loss": 0.4427,
+      "step": 2747
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2178190914703325e-05,
+      "loss": 0.5225,
+      "step": 2748
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2173045184514733e-05,
+      "loss": 0.472,
+      "step": 2749
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2167898850347301e-05,
+      "loss": 0.4234,
+      "step": 2750
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2162751913631412e-05,
+      "loss": 0.494,
+      "step": 2751
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2157604375797613e-05,
+      "loss": 0.4812,
+      "step": 2752
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2152456238276613e-05,
+      "loss": 0.4439,
+      "step": 2753
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2147307502499293e-05,
+      "loss": 0.423,
+      "step": 2754
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2142158169896697e-05,
+      "loss": 0.4828,
+      "step": 2755
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2137008241900044e-05,
+      "loss": 0.475,
+      "step": 2756
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.21318577199407e-05,
+      "loss": 0.429,
+      "step": 2757
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2126706605450218e-05,
+      "loss": 0.4009,
+      "step": 2758
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2121554899860302e-05,
+      "loss": 0.4124,
+      "step": 2759
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2116402604602814e-05,
+      "loss": 0.4516,
+      "step": 2760
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2111249721109806e-05,
+      "loss": 0.4103,
+      "step": 2761
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2106096250813466e-05,
+      "loss": 0.4805,
+      "step": 2762
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2100942195146155e-05,
+      "loss": 0.4477,
+      "step": 2763
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2095787555540404e-05,
+      "loss": 0.3921,
+      "step": 2764
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2090632333428894e-05,
+      "loss": 0.4563,
+      "step": 2765
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2085476530244476e-05,
+      "loss": 0.4665,
+      "step": 2766
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2080320147420164e-05,
+      "loss": 0.4087,
+      "step": 2767
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207516318638912e-05,
+      "loss": 0.4914,
+      "step": 2768
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2070005648584685e-05,
+      "loss": 0.4751,
+      "step": 2769
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2064847535440345e-05,
+      "loss": 0.518,
+      "step": 2770
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205968884838976e-05,
+      "loss": 0.4884,
+      "step": 2771
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2054529588866733e-05,
+      "loss": 0.4629,
+      "step": 2772
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204936975830524e-05,
+      "loss": 0.4484,
+      "step": 2773
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2044209358139412e-05,
+      "loss": 0.4407,
+      "step": 2774
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.203904838980353e-05,
+      "loss": 0.4826,
+      "step": 2775
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2033886854732047e-05,
+      "loss": 0.4098,
+      "step": 2776
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2028724754359567e-05,
+      "loss": 0.4522,
+      "step": 2777
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202356209012084e-05,
+      "loss": 0.4661,
+      "step": 2778
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018398863450795e-05,
+      "loss": 0.3718,
+      "step": 2779
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.20132350757845e-05,
+      "loss": 0.4408,
+      "step": 2780
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2008070728557186e-05,
+      "loss": 0.4528,
+      "step": 2781
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2002905823204239e-05,
+      "loss": 0.4736,
+      "step": 2782
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1997740361161194e-05,
+      "loss": 0.4858,
+      "step": 2783
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1992574343863752e-05,
+      "loss": 0.5125,
+      "step": 2784
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1987407772747757e-05,
+      "loss": 0.4645,
+      "step": 2785
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1982240649249217e-05,
+      "loss": 0.4724,
+      "step": 2786
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1977072974804287e-05,
+      "loss": 0.4564,
+      "step": 2787
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1971904750849272e-05,
+      "loss": 0.4666,
+      "step": 2788
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1966735978820643e-05,
+      "loss": 0.441,
+      "step": 2789
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1961566660155008e-05,
+      "loss": 0.4568,
+      "step": 2790
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1956396796289137e-05,
+      "loss": 0.5125,
+      "step": 2791
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.195122638865995e-05,
+      "loss": 0.451,
+      "step": 2792
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1946055438704507e-05,
+      "loss": 0.4427,
+      "step": 2793
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194088394786004e-05,
+      "loss": 0.4254,
+      "step": 2794
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1935711917563912e-05,
+      "loss": 0.5438,
+      "step": 2795
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1930539349253646e-05,
+      "loss": 0.4258,
+      "step": 2796
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1925366244366914e-05,
+      "loss": 0.4524,
+      "step": 2797
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1920192604341531e-05,
+      "loss": 0.4959,
+      "step": 2798
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1915018430615468e-05,
+      "loss": 0.3501,
+      "step": 2799
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1909843724626839e-05,
+      "loss": 0.4724,
+      "step": 2800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190466848781391e-05,
+      "loss": 0.4785,
+      "step": 2801
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1899492721615091e-05,
+      "loss": 0.386,
+      "step": 2802
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1894316427468942e-05,
+      "loss": 0.4733,
+      "step": 2803
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188913960681417e-05,
+      "loss": 0.4146,
+      "step": 2804
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1883962261089621e-05,
+      "loss": 0.4684,
+      "step": 2805
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.18787843917343e-05,
+      "loss": 0.4955,
+      "step": 2806
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1873606000187342e-05,
+      "loss": 0.4111,
+      "step": 2807
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1868427087888047e-05,
+      "loss": 0.4786,
+      "step": 2808
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1863247656275837e-05,
+      "loss": 0.4559,
+      "step": 2809
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1858067706790297e-05,
+      "loss": 0.4249,
+      "step": 2810
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852887240871145e-05,
+      "loss": 0.4646,
+      "step": 2811
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1847706259958245e-05,
+      "loss": 0.4438,
+      "step": 2812
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1842524765491609e-05,
+      "loss": 0.4788,
+      "step": 2813
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1837342758911386e-05,
+      "loss": 0.4166,
+      "step": 2814
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1832160241657869e-05,
+      "loss": 0.4311,
+      "step": 2815
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1826977215171494e-05,
+      "loss": 0.4724,
+      "step": 2816
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1821793680892835e-05,
+      "loss": 0.4285,
+      "step": 2817
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1816609640262612e-05,
+      "loss": 0.4733,
+      "step": 2818
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1811425094721687e-05,
+      "loss": 0.4314,
+      "step": 2819
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1806240045711053e-05,
+      "loss": 0.465,
+      "step": 2820
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1801054494671851e-05,
+      "loss": 0.4852,
+      "step": 2821
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1795868443045358e-05,
+      "loss": 0.5266,
+      "step": 2822
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1790681892272997e-05,
+      "loss": 0.5096,
+      "step": 2823
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1785494843796319e-05,
+      "loss": 0.4646,
+      "step": 2824
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1780307299057019e-05,
+      "loss": 0.4051,
+      "step": 2825
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1775119259496933e-05,
+      "loss": 0.4283,
+      "step": 2826
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1769930726558022e-05,
+      "loss": 0.4368,
+      "step": 2827
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176474170168241e-05,
+      "loss": 0.4368,
+      "step": 2828
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1759552186312324e-05,
+      "loss": 0.4519,
+      "step": 2829
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1754362181890152e-05,
+      "loss": 0.4378,
+      "step": 2830
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1749171689858411e-05,
+      "loss": 0.4215,
+      "step": 2831
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.174398071165975e-05,
+      "loss": 0.4342,
+      "step": 2832
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1738789248736958e-05,
+      "loss": 0.4533,
+      "step": 2833
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1733597302532956e-05,
+      "loss": 0.4485,
+      "step": 2834
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1728404874490795e-05,
+      "loss": 0.4331,
+      "step": 2835
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1723211966053672e-05,
+      "loss": 0.4139,
+      "step": 2836
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1718018578664904e-05,
+      "loss": 0.4214,
+      "step": 2837
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1712824713767956e-05,
+      "loss": 0.4884,
+      "step": 2838
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1707630372806408e-05,
+      "loss": 0.4475,
+      "step": 2839
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1702435557223988e-05,
+      "loss": 0.4488,
+      "step": 2840
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1697240268464546e-05,
+      "loss": 0.4573,
+      "step": 2841
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1692044507972065e-05,
+      "loss": 0.4839,
+      "step": 2842
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1686848277190667e-05,
+      "loss": 0.4506,
+      "step": 2843
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681651577564597e-05,
+      "loss": 0.5016,
+      "step": 2844
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1676454410538225e-05,
+      "loss": 0.4133,
+      "step": 2845
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1671256777556068e-05,
+      "loss": 0.4539,
+      "step": 2846
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1666058680062757e-05,
+      "loss": 0.4617,
+      "step": 2847
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1660860119503057e-05,
+      "loss": 0.4205,
+      "step": 2848
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1655661097321864e-05,
+      "loss": 0.4451,
+      "step": 2849
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1650461614964203e-05,
+      "loss": 0.4765,
+      "step": 2850
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1645261673875222e-05,
+      "loss": 0.4764,
+      "step": 2851
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1640061275500194e-05,
+      "loss": 0.4635,
+      "step": 2852
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1634860421284535e-05,
+      "loss": 0.455,
+      "step": 2853
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629659112673764e-05,
+      "loss": 0.4534,
+      "step": 2854
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1624457351113552e-05,
+      "loss": 0.477,
+      "step": 2855
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1619255138049676e-05,
+      "loss": 0.4222,
+      "step": 2856
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1614052474928041e-05,
+      "loss": 0.4444,
+      "step": 2857
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1608849363194694e-05,
+      "loss": 0.4804,
+      "step": 2858
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1603645804295783e-05,
+      "loss": 0.4673,
+      "step": 2859
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1598441799677596e-05,
+      "loss": 0.4231,
+      "step": 2860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159323735078654e-05,
+      "loss": 0.4152,
+      "step": 2861
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1588032459069146e-05,
+      "loss": 0.4744,
+      "step": 2862
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1582827125972067e-05,
+      "loss": 0.4395,
+      "step": 2863
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1577621352942081e-05,
+      "loss": 0.4139,
+      "step": 2864
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1572415141426088e-05,
+      "loss": 0.4694,
+      "step": 2865
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1567208492871105e-05,
+      "loss": 0.4537,
+      "step": 2866
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1562001408724276e-05,
+      "loss": 0.4801,
+      "step": 2867
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1556793890432869e-05,
+      "loss": 0.4453,
+      "step": 2868
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1551585939444259e-05,
+      "loss": 0.4751,
+      "step": 2869
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154637755720596e-05,
+      "loss": 0.4935,
+      "step": 2870
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154116874516559e-05,
+      "loss": 0.395,
+      "step": 2871
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1535959504770888e-05,
+      "loss": 0.4753,
+      "step": 2872
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.153074983746973e-05,
+      "loss": 0.4464,
+      "step": 2873
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1525539744710087e-05,
+      "loss": 0.4071,
+      "step": 2874
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1520329227940059e-05,
+      "loss": 0.4184,
+      "step": 2875
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.151511828860787e-05,
+      "loss": 0.4849,
+      "step": 2876
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1509906928161847e-05,
+      "loss": 0.4437,
+      "step": 2877
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1504695148050446e-05,
+      "loss": 0.4413,
+      "step": 2878
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1499482949722234e-05,
+      "loss": 0.4778,
+      "step": 2879
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1494270334625901e-05,
+      "loss": 0.4605,
+      "step": 2880
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148905730421024e-05,
+      "loss": 0.4388,
+      "step": 2881
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1483843859924171e-05,
+      "loss": 0.4412,
+      "step": 2882
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1478630003216727e-05,
+      "loss": 0.4171,
+      "step": 2883
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1473415735537049e-05,
+      "loss": 0.4812,
+      "step": 2884
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1468201058334398e-05,
+      "loss": 0.4819,
+      "step": 2885
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1462985973058152e-05,
+      "loss": 0.4595,
+      "step": 2886
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1457770481157788e-05,
+      "loss": 0.5154,
+      "step": 2887
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1452554584082923e-05,
+      "loss": 0.4322,
+      "step": 2888
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1447338283283254e-05,
+      "loss": 0.4813,
+      "step": 2889
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1442121580208613e-05,
+      "loss": 0.4357,
+      "step": 2890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1436904476308939e-05,
+      "loss": 0.4176,
+      "step": 2891
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1431686973034275e-05,
+      "loss": 0.4166,
+      "step": 2892
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1426469071834782e-05,
+      "loss": 0.445,
+      "step": 2893
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1421250774160731e-05,
+      "loss": 0.4835,
+      "step": 2894
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1416032081462502e-05,
+      "loss": 0.4664,
+      "step": 2895
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1410812995190584e-05,
+      "loss": 0.4704,
+      "step": 2896
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1405593516795574e-05,
+      "loss": 0.4553,
+      "step": 2897
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1400373647728187e-05,
+      "loss": 0.4814,
+      "step": 2898
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1395153389439232e-05,
+      "loss": 0.3846,
+      "step": 2899
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1389932743379637e-05,
+      "loss": 0.435,
+      "step": 2900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1384711711000438e-05,
+      "loss": 0.4259,
+      "step": 2901
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137949029375277e-05,
+      "loss": 0.4145,
+      "step": 2902
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1374268493087884e-05,
+      "loss": 0.414,
+      "step": 2903
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136904631045713e-05,
+      "loss": 0.4319,
+      "step": 2904
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1363823747311971e-05,
+      "loss": 0.3725,
+      "step": 2905
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1358600805103973e-05,
+      "loss": 0.437,
+      "step": 2906
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1353377485284804e-05,
+      "loss": 0.4204,
+      "step": 2907
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.134815378930624e-05,
+      "loss": 0.4787,
+      "step": 2908
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1342929718620164e-05,
+      "loss": 0.4375,
+      "step": 2909
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1337705274678562e-05,
+      "loss": 0.4757,
+      "step": 2910
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1332480458933515e-05,
+      "loss": 0.4755,
+      "step": 2911
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1327255272837221e-05,
+      "loss": 0.4992,
+      "step": 2912
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1322029717841976e-05,
+      "loss": 0.4538,
+      "step": 2913
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1316803795400168e-05,
+      "loss": 0.4131,
+      "step": 2914
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1311577506964307e-05,
+      "loss": 0.4954,
+      "step": 2915
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1306350853986987e-05,
+      "loss": 0.5144,
+      "step": 2916
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1301123837920915e-05,
+      "loss": 0.4922,
+      "step": 2917
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1295896460218891e-05,
+      "loss": 0.41,
+      "step": 2918
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.129066872233382e-05,
+      "loss": 0.4163,
+      "step": 2919
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1285440625718706e-05,
+      "loss": 0.4308,
+      "step": 2920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1280212171826648e-05,
+      "loss": 0.3979,
+      "step": 2921
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1274983362110859e-05,
+      "loss": 0.4221,
+      "step": 2922
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1269754198024634e-05,
+      "loss": 0.4328,
+      "step": 2923
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1264524681021367e-05,
+      "loss": 0.4331,
+      "step": 2924
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125929481255457e-05,
+      "loss": 0.4486,
+      "step": 2925
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1254064594077831e-05,
+      "loss": 0.4053,
+      "step": 2926
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1248834027044847e-05,
+      "loss": 0.4528,
+      "step": 2927
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1243603112909402e-05,
+      "loss": 0.5176,
+      "step": 2928
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1238371853125389e-05,
+      "loss": 0.5479,
+      "step": 2929
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1233140249146788e-05,
+      "loss": 0.4261,
+      "step": 2930
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1227908302427677e-05,
+      "loss": 0.5077,
+      "step": 2931
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1222676014422231e-05,
+      "loss": 0.4398,
+      "step": 2932
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1217443386584717e-05,
+      "loss": 0.4267,
+      "step": 2933
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.12122104203695e-05,
+      "loss": 0.4291,
+      "step": 2934
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1206977117231037e-05,
+      "loss": 0.4411,
+      "step": 2935
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1201743478623874e-05,
+      "loss": 0.4421,
+      "step": 2936
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1196509506002661e-05,
+      "loss": 0.501,
+      "step": 2937
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1191275200822136e-05,
+      "loss": 0.4776,
+      "step": 2938
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1186040564537117e-05,
+      "loss": 0.4904,
+      "step": 2939
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.118080559860254e-05,
+      "loss": 0.4428,
+      "step": 2940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1175570304473406e-05,
+      "loss": 0.4212,
+      "step": 2941
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1170334683604825e-05,
+      "loss": 0.4547,
+      "step": 2942
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1165098737451992e-05,
+      "loss": 0.4894,
+      "step": 2943
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1159862467470188e-05,
+      "loss": 0.4576,
+      "step": 2944
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1154625875114791e-05,
+      "loss": 0.4626,
+      "step": 2945
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1149388961841266e-05,
+      "loss": 0.4393,
+      "step": 2946
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.114415172910517e-05,
+      "loss": 0.4442,
+      "step": 2947
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.113891417836214e-05,
+      "loss": 0.4146,
+      "step": 2948
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.113367631106791e-05,
+      "loss": 0.4693,
+      "step": 2949
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1128438128678303e-05,
+      "loss": 0.393,
+      "step": 2950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1123199632649218e-05,
+      "loss": 0.5471,
+      "step": 2951
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1117960824436656e-05,
+      "loss": 0.393,
+      "step": 2952
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1112721705496691e-05,
+      "loss": 0.4289,
+      "step": 2953
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1107482277285497e-05,
+      "loss": 0.3855,
+      "step": 2954
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1102242541259327e-05,
+      "loss": 0.4549,
+      "step": 2955
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1097002498874514e-05,
+      "loss": 0.4387,
+      "step": 2956
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1091762151587483e-05,
+      "loss": 0.4805,
+      "step": 2957
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1086521500854746e-05,
+      "loss": 0.4125,
+      "step": 2958
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1081280548132896e-05,
+      "loss": 0.4637,
+      "step": 2959
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1076039294878605e-05,
+      "loss": 0.4687,
+      "step": 2960
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1070797742548636e-05,
+      "loss": 0.3992,
+      "step": 2961
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1065555892599835e-05,
+      "loss": 0.5153,
+      "step": 2962
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1060313746489124e-05,
+      "loss": 0.426,
+      "step": 2963
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1055071305673512e-05,
+      "loss": 0.4315,
+      "step": 2964
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1049828571610095e-05,
+      "loss": 0.3987,
+      "step": 2965
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1044585545756035e-05,
+      "loss": 0.4534,
+      "step": 2966
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1039342229568595e-05,
+      "loss": 0.4375,
+      "step": 2967
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1034098624505103e-05,
+      "loss": 0.5234,
+      "step": 2968
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1028854732022972e-05,
+      "loss": 0.4027,
+      "step": 2969
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1023610553579702e-05,
+      "loss": 0.4598,
+      "step": 2970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1018366090632862e-05,
+      "loss": 0.464,
+      "step": 2971
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1013121344640104e-05,
+      "loss": 0.4589,
+      "step": 2972
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1007876317059161e-05,
+      "loss": 0.3883,
+      "step": 2973
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1002631009347842e-05,
+      "loss": 0.4181,
+      "step": 2974
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0997385422964033e-05,
+      "loss": 0.5356,
+      "step": 2975
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0992139559365697e-05,
+      "loss": 0.3925,
+      "step": 2976
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0986893420010884e-05,
+      "loss": 0.4188,
+      "step": 2977
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0981647006357703e-05,
+      "loss": 0.4114,
+      "step": 2978
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0976400319864353e-05,
+      "loss": 0.4,
+      "step": 2979
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0971153361989106e-05,
+      "loss": 0.4014,
+      "step": 2980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0965906134190303e-05,
+      "loss": 0.4512,
+      "step": 2981
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0960658637926367e-05,
+      "loss": 0.4903,
+      "step": 2982
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0955410874655794e-05,
+      "loss": 0.4755,
+      "step": 2983
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0950162845837154e-05,
+      "loss": 0.4049,
+      "step": 2984
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0944914552929094e-05,
+      "loss": 0.4846,
+      "step": 2985
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0939665997390324e-05,
+      "loss": 0.4124,
+      "step": 2986
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0934417180679637e-05,
+      "loss": 0.4101,
+      "step": 2987
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0929168104255896e-05,
+      "loss": 0.4685,
+      "step": 2988
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0923918769578037e-05,
+      "loss": 0.4333,
+      "step": 2989
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0918669178105063e-05,
+      "loss": 0.4965,
+      "step": 2990
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0913419331296056e-05,
+      "loss": 0.4003,
+      "step": 2991
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0908169230610163e-05,
+      "loss": 0.4572,
+      "step": 2992
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0902918877506604e-05,
+      "loss": 0.4592,
+      "step": 2993
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0897668273444663e-05,
+      "loss": 0.4659,
+      "step": 2994
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0892417419883708e-05,
+      "loss": 0.4761,
+      "step": 2995
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0887166318283164e-05,
+      "loss": 0.3955,
+      "step": 2996
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0881914970102528e-05,
+      "loss": 0.4255,
+      "step": 2997
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0876663376801364e-05,
+      "loss": 0.3864,
+      "step": 2998
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.087141153983931e-05,
+      "loss": 0.471,
+      "step": 2999
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0866159460676064e-05,
+      "loss": 0.4564,
+      "step": 3000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0860907140771404e-05,
+      "loss": 0.418,
+      "step": 3001
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0855654581585155e-05,
+      "loss": 0.482,
+      "step": 3002
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0850401784577227e-05,
+      "loss": 0.4123,
+      "step": 3003
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.084514875120759e-05,
+      "loss": 0.4339,
+      "step": 3004
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0839895482936271e-05,
+      "loss": 0.4794,
+      "step": 3005
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0834641981223375e-05,
+      "loss": 0.4855,
+      "step": 3006
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0829388247529067e-05,
+      "loss": 0.4093,
+      "step": 3007
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0824134283313574e-05,
+      "loss": 0.438,
+      "step": 3008
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.081888009003719e-05,
+      "loss": 0.5209,
+      "step": 3009
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0813625669160273e-05,
+      "loss": 0.4081,
+      "step": 3010
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0808371022143242e-05,
+      "loss": 0.4681,
+      "step": 3011
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.080311615044658e-05,
+      "loss": 0.4916,
+      "step": 3012
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0797861055530832e-05,
+      "loss": 0.5024,
+      "step": 3013
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079260573885661e-05,
+      "loss": 0.4934,
+      "step": 3014
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0787350201884573e-05,
+      "loss": 0.4712,
+      "step": 3015
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0782094446075461e-05,
+      "loss": 0.4434,
+      "step": 3016
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0776838472890065e-05,
+      "loss": 0.4562,
+      "step": 3017
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0771582283789229e-05,
+      "loss": 0.4958,
+      "step": 3018
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0766325880233871e-05,
+      "loss": 0.5227,
+      "step": 3019
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0761069263684963e-05,
+      "loss": 0.4244,
+      "step": 3020
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0755812435603528e-05,
+      "loss": 0.4031,
+      "step": 3021
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0750555397450666e-05,
+      "loss": 0.4275,
+      "step": 3022
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0745298150687513e-05,
+      "loss": 0.4597,
+      "step": 3023
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0740040696775282e-05,
+      "loss": 0.4523,
+      "step": 3024
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0734783037175235e-05,
+      "loss": 0.4358,
+      "step": 3025
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0729525173348696e-05,
+      "loss": 0.4117,
+      "step": 3026
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0724267106757034e-05,
+      "loss": 0.4167,
+      "step": 3027
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0719008838861684e-05,
+      "loss": 0.4536,
+      "step": 3028
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0713750371124142e-05,
+      "loss": 0.4344,
+      "step": 3029
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0708491705005945e-05,
+      "loss": 0.4245,
+      "step": 3030
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.07032328419687e-05,
+      "loss": 0.479,
+      "step": 3031
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069797378347406e-05,
+      "loss": 0.4511,
+      "step": 3032
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0692714530983728e-05,
+      "loss": 0.416,
+      "step": 3033
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0687455085959478e-05,
+      "loss": 0.4675,
+      "step": 3034
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0682195449863115e-05,
+      "loss": 0.4166,
+      "step": 3035
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0676935624156517e-05,
+      "loss": 0.4287,
+      "step": 3036
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0671675610301604e-05,
+      "loss": 0.5429,
+      "step": 3037
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066641540976035e-05,
+      "loss": 0.4343,
+      "step": 3038
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0661155023994779e-05,
+      "loss": 0.4761,
+      "step": 3039
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0655894454466972e-05,
+      "loss": 0.395,
+      "step": 3040
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0650633702639064e-05,
+      "loss": 0.4703,
+      "step": 3041
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0645372769973223e-05,
+      "loss": 0.4498,
+      "step": 3042
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0640111657931687e-05,
+      "loss": 0.4343,
+      "step": 3043
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0634850367976735e-05,
+      "loss": 0.4677,
+      "step": 3044
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0629588901570691e-05,
+      "loss": 0.547,
+      "step": 3045
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0624327260175943e-05,
+      "loss": 0.3557,
+      "step": 3046
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0619065445254907e-05,
+      "loss": 0.456,
+      "step": 3047
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0613803458270066e-05,
+      "loss": 0.5055,
+      "step": 3048
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0608541300683946e-05,
+      "loss": 0.454,
+      "step": 3049
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0603278973959105e-05,
+      "loss": 0.4517,
+      "step": 3050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598016479558174e-05,
+      "loss": 0.4896,
+      "step": 3051
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0592753818943806e-05,
+      "loss": 0.4398,
+      "step": 3052
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0587490993578721e-05,
+      "loss": 0.4281,
+      "step": 3053
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0582228004925668e-05,
+      "loss": 0.392,
+      "step": 3054
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057696485444745e-05,
+      "loss": 0.4268,
+      "step": 3055
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057170154360692e-05,
+      "loss": 0.4423,
+      "step": 3056
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.056643807386696e-05,
+      "loss": 0.4352,
+      "step": 3057
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0561174446690508e-05,
+      "loss": 0.4595,
+      "step": 3058
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0555910663540548e-05,
+      "loss": 0.5063,
+      "step": 3059
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0550646725880091e-05,
+      "loss": 0.4823,
+      "step": 3060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545382635172214e-05,
+      "loss": 0.3597,
+      "step": 3061
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.054011839288002e-05,
+      "loss": 0.4376,
+      "step": 3062
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0534854000466658e-05,
+      "loss": 0.4675,
+      "step": 3063
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0529589459395324e-05,
+      "loss": 0.467,
+      "step": 3064
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0524324771129247e-05,
+      "loss": 0.4226,
+      "step": 3065
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.05190599371317e-05,
+      "loss": 0.3863,
+      "step": 3066
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0513794958865997e-05,
+      "loss": 0.4809,
+      "step": 3067
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0508529837795501e-05,
+      "loss": 0.48,
+      "step": 3068
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0503264575383594e-05,
+      "loss": 0.393,
+      "step": 3069
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0497999173093718e-05,
+      "loss": 0.4753,
+      "step": 3070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0492733632389344e-05,
+      "loss": 0.4543,
+      "step": 3071
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0487467954733978e-05,
+      "loss": 0.4489,
+      "step": 3072
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0482202141591174e-05,
+      "loss": 0.3819,
+      "step": 3073
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0476936194424519e-05,
+      "loss": 0.3998,
+      "step": 3074
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0471670114697632e-05,
+      "loss": 0.4835,
+      "step": 3075
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0466403903874176e-05,
+      "loss": 0.5241,
+      "step": 3076
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046113756341785e-05,
+      "loss": 0.4573,
+      "step": 3077
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0455871094792385e-05,
+      "loss": 0.4035,
+      "step": 3078
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0450604499461551e-05,
+      "loss": 0.4733,
+      "step": 3079
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0445337778889151e-05,
+      "loss": 0.5353,
+      "step": 3080
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0440070934539025e-05,
+      "loss": 0.456,
+      "step": 3081
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0434803967875042e-05,
+      "loss": 0.4994,
+      "step": 3082
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0429536880361114e-05,
+      "loss": 0.4198,
+      "step": 3083
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042426967346118e-05,
+      "loss": 0.428,
+      "step": 3084
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0419002348639215e-05,
+      "loss": 0.4223,
+      "step": 3085
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0413734907359225e-05,
+      "loss": 0.4775,
+      "step": 3086
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0408467351085248e-05,
+      "loss": 0.4208,
+      "step": 3087
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0403199681281355e-05,
+      "loss": 0.4575,
+      "step": 3088
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0397931899411651e-05,
+      "loss": 0.4603,
+      "step": 3089
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0392664006940267e-05,
+      "loss": 0.4201,
+      "step": 3090
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0387396005331367e-05,
+      "loss": 0.4418,
+      "step": 3091
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0382127896049147e-05,
+      "loss": 0.4382,
+      "step": 3092
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0376859680557839e-05,
+      "loss": 0.4204,
+      "step": 3093
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0371591360321683e-05,
+      "loss": 0.4158,
+      "step": 3094
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0366322936804973e-05,
+      "loss": 0.4181,
+      "step": 3095
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0361054411472018e-05,
+      "loss": 0.4045,
+      "step": 3096
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0355785785787155e-05,
+      "loss": 0.4941,
+      "step": 3097
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035051706121476e-05,
+      "loss": 0.4259,
+      "step": 3098
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0345248239219221e-05,
+      "loss": 0.3991,
+      "step": 3099
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0339979321264965e-05,
+      "loss": 0.4809,
+      "step": 3100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0334710308816442e-05,
+      "loss": 0.4548,
+      "step": 3101
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0329441203338126e-05,
+      "loss": 0.464,
+      "step": 3102
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0324172006294517e-05,
+      "loss": 0.4407,
+      "step": 3103
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0318902719150145e-05,
+      "loss": 0.4252,
+      "step": 3104
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0313633343369563e-05,
+      "loss": 0.4686,
+      "step": 3105
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0308363880417345e-05,
+      "loss": 0.4712,
+      "step": 3106
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030309433175809e-05,
+      "loss": 0.4325,
+      "step": 3107
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0297824698856431e-05,
+      "loss": 0.4424,
+      "step": 3108
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0292554983177005e-05,
+      "loss": 0.5109,
+      "step": 3109
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0287285186184492e-05,
+      "loss": 0.4258,
+      "step": 3110
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0282015309343585e-05,
+      "loss": 0.4309,
+      "step": 3111
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0276745354118994e-05,
+      "loss": 0.4261,
+      "step": 3112
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0271475321975463e-05,
+      "loss": 0.4057,
+      "step": 3113
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026620521437775e-05,
+      "loss": 0.5096,
+      "step": 3114
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026093503279063e-05,
+      "loss": 0.4619,
+      "step": 3115
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0255664778678914e-05,
+      "loss": 0.4454,
+      "step": 3116
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0250394453507413e-05,
+      "loss": 0.4049,
+      "step": 3117
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024512405874097e-05,
+      "loss": 0.423,
+      "step": 3118
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0239853595844447e-05,
+      "loss": 0.4644,
+      "step": 3119
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0234583066282724e-05,
+      "loss": 0.5043,
+      "step": 3120
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0229312471520695e-05,
+      "loss": 0.4297,
+      "step": 3121
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0224041813023278e-05,
+      "loss": 0.4752,
+      "step": 3122
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0218771092255407e-05,
+      "loss": 0.4652,
+      "step": 3123
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0213500310682025e-05,
+      "loss": 0.3906,
+      "step": 3124
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0208229469768112e-05,
+      "loss": 0.4248,
+      "step": 3125
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0202958570978646e-05,
+      "loss": 0.4237,
+      "step": 3126
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019768761577862e-05,
+      "loss": 0.419,
+      "step": 3127
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0192416605633061e-05,
+      "loss": 0.4152,
+      "step": 3128
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0187145542006994e-05,
+      "loss": 0.4916,
+      "step": 3129
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0181874426365466e-05,
+      "loss": 0.4209,
+      "step": 3130
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0176603260173539e-05,
+      "loss": 0.4191,
+      "step": 3131
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0171332044896285e-05,
+      "loss": 0.4286,
+      "step": 3132
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0166060781998792e-05,
+      "loss": 0.4549,
+      "step": 3133
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0160789472946161e-05,
+      "loss": 0.4159,
+      "step": 3134
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0155518119203511e-05,
+      "loss": 0.4447,
+      "step": 3135
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0150246722235962e-05,
+      "loss": 0.4277,
+      "step": 3136
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0144975283508656e-05,
+      "loss": 0.4779,
+      "step": 3137
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0139703804486742e-05,
+      "loss": 0.4306,
+      "step": 3138
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013443228663538e-05,
+      "loss": 0.4577,
+      "step": 3139
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0129160731419747e-05,
+      "loss": 0.4786,
+      "step": 3140
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.012388914030502e-05,
+      "loss": 0.5023,
+      "step": 3141
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0118617514756394e-05,
+      "loss": 0.4383,
+      "step": 3142
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0113345856239074e-05,
+      "loss": 0.432,
+      "step": 3143
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0108074166218264e-05,
+      "loss": 0.4867,
+      "step": 3144
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.010280244615919e-05,
+      "loss": 0.4412,
+      "step": 3145
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0097530697527078e-05,
+      "loss": 0.4463,
+      "step": 3146
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0092258921787167e-05,
+      "loss": 0.4714,
+      "step": 3147
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0086987120404696e-05,
+      "loss": 0.4678,
+      "step": 3148
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0081715294844918e-05,
+      "loss": 0.4052,
+      "step": 3149
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0076443446573092e-05,
+      "loss": 0.447,
+      "step": 3150
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0071171577054482e-05,
+      "loss": 0.4605,
+      "step": 3151
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0065899687754355e-05,
+      "loss": 0.4162,
+      "step": 3152
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0060627780137989e-05,
+      "loss": 0.4572,
+      "step": 3153
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0055355855670663e-05,
+      "loss": 0.4984,
+      "step": 3154
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050083915817661e-05,
+      "loss": 0.4452,
+      "step": 3155
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0044811962044276e-05,
+      "loss": 0.4818,
+      "step": 3156
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039539995815798e-05,
+      "loss": 0.4439,
+      "step": 3157
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003426801859753e-05,
+      "loss": 0.4412,
+      "step": 3158
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0028996031854764e-05,
+      "loss": 0.4151,
+      "step": 3159
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0023724037052805e-05,
+      "loss": 0.4538,
+      "step": 3160
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0018452035656956e-05,
+      "loss": 0.3845,
+      "step": 3161
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001318002913253e-05,
+      "loss": 0.4471,
+      "step": 3162
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0007908018944827e-05,
+      "loss": 0.479,
+      "step": 3163
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.000263600655916e-05,
+      "loss": 0.4728,
+      "step": 3164
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.997363993440841e-06,
+      "loss": 0.466,
+      "step": 3165
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.992091981055175e-06,
+      "loss": 0.432,
+      "step": 3166
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986819970867471e-06,
+      "loss": 0.3741,
+      "step": 3167
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.981547964343044e-06,
+      "loss": 0.4706,
+      "step": 3168
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9762759629472e-06,
+      "loss": 0.5019,
+      "step": 3169
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97100396814524e-06,
+      "loss": 0.4165,
+      "step": 3170
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965731981402476e-06,
+      "loss": 0.4057,
+      "step": 3171
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960460004184202e-06,
+      "loss": 0.4439,
+      "step": 3172
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.955188037955724e-06,
+      "loss": 0.3547,
+      "step": 3173
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.949916084182342e-06,
+      "loss": 0.4301,
+      "step": 3174
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94464414432934e-06,
+      "loss": 0.4256,
+      "step": 3175
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.939372219862016e-06,
+      "loss": 0.5149,
+      "step": 3176
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.934100312245647e-06,
+      "loss": 0.4923,
+      "step": 3177
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.928828422945523e-06,
+      "loss": 0.4629,
+      "step": 3178
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.92355655342691e-06,
+      "loss": 0.4044,
+      "step": 3179
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.918284705155085e-06,
+      "loss": 0.4175,
+      "step": 3180
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.913012879595307e-06,
+      "loss": 0.4114,
+      "step": 3181
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.907741078212834e-06,
+      "loss": 0.4398,
+      "step": 3182
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.902469302472922e-06,
+      "loss": 0.5111,
+      "step": 3183
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.897197553840811e-06,
+      "loss": 0.4322,
+      "step": 3184
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.891925833781738e-06,
+      "loss": 0.4292,
+      "step": 3185
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.886654143760931e-06,
+      "loss": 0.4653,
+      "step": 3186
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.881382485243606e-06,
+      "loss": 0.4525,
+      "step": 3187
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87611085969498e-06,
+      "loss": 0.4725,
+      "step": 3188
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.870839268580255e-06,
+      "loss": 0.3847,
+      "step": 3189
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.865567713364622e-06,
+      "loss": 0.5117,
+      "step": 3190
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.860296195513263e-06,
+      "loss": 0.4678,
+      "step": 3191
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.855024716491345e-06,
+      "loss": 0.4503,
+      "step": 3192
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84975327776404e-06,
+      "loss": 0.4286,
+      "step": 3193
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.844481880796492e-06,
+      "loss": 0.4543,
+      "step": 3194
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.839210527053842e-06,
+      "loss": 0.4702,
+      "step": 3195
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.833939218001211e-06,
+      "loss": 0.3917,
+      "step": 3196
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.828667955103717e-06,
+      "loss": 0.3924,
+      "step": 3197
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.823396739826464e-06,
+      "loss": 0.4806,
+      "step": 3198
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.818125573634537e-06,
+      "loss": 0.4568,
+      "step": 3199
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.812854457993008e-06,
+      "loss": 0.4419,
+      "step": 3200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.807583394366944e-06,
+      "loss": 0.4586,
+      "step": 3201
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.80231238422138e-06,
+      "loss": 0.4726,
+      "step": 3202
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.797041429021358e-06,
+      "loss": 0.49,
+      "step": 3203
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.791770530231891e-06,
+      "loss": 0.5102,
+      "step": 3204
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.786499689317976e-06,
+      "loss": 0.4405,
+      "step": 3205
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.781228907744598e-06,
+      "loss": 0.4489,
+      "step": 3206
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.775958186976724e-06,
+      "loss": 0.4572,
+      "step": 3207
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.770687528479308e-06,
+      "loss": 0.4549,
+      "step": 3208
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76541693371728e-06,
+      "loss": 0.4632,
+      "step": 3209
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.760146404155556e-06,
+      "loss": 0.5368,
+      "step": 3210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.754875941259036e-06,
+      "loss": 0.4086,
+      "step": 3211
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74960554649259e-06,
+      "loss": 0.4786,
+      "step": 3212
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74433522132109e-06,
+      "loss": 0.4343,
+      "step": 3213
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.739064967209371e-06,
+      "loss": 0.4917,
+      "step": 3214
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.733794785622254e-06,
+      "loss": 0.4267,
+      "step": 3215
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.728524678024542e-06,
+      "loss": 0.4545,
+      "step": 3216
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.723254645881008e-06,
+      "loss": 0.4115,
+      "step": 3217
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.717984690656417e-06,
+      "loss": 0.4341,
+      "step": 3218
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.712714813815511e-06,
+      "loss": 0.4101,
+      "step": 3219
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.707445016822999e-06,
+      "loss": 0.4693,
+      "step": 3220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.702175301143576e-06,
+      "loss": 0.4508,
+      "step": 3221
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69690566824191e-06,
+      "loss": 0.4089,
+      "step": 3222
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.691636119582659e-06,
+      "loss": 0.506,
+      "step": 3223
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686366656630438e-06,
+      "loss": 0.5269,
+      "step": 3224
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.681097280849858e-06,
+      "loss": 0.4348,
+      "step": 3225
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.675827993705488e-06,
+      "loss": 0.4891,
+      "step": 3226
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.670558796661877e-06,
+      "loss": 0.4816,
+      "step": 3227
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.665289691183561e-06,
+      "loss": 0.3973,
+      "step": 3228
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.660020678735036e-06,
+      "loss": 0.4101,
+      "step": 3229
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654751760780782e-06,
+      "loss": 0.4524,
+      "step": 3230
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.649482938785246e-06,
+      "loss": 0.471,
+      "step": 3231
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.644214214212847e-06,
+      "loss": 0.4273,
+      "step": 3232
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.638945588527984e-06,
+      "loss": 0.4714,
+      "step": 3233
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.63367706319503e-06,
+      "loss": 0.419,
+      "step": 3234
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.628408639678322e-06,
+      "loss": 0.4907,
+      "step": 3235
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.623140319442166e-06,
+      "loss": 0.4534,
+      "step": 3236
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.617872103950851e-06,
+      "loss": 0.3445,
+      "step": 3237
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.612603994668635e-06,
+      "loss": 0.4238,
+      "step": 3238
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.607335993059736e-06,
+      "loss": 0.4489,
+      "step": 3239
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.602068100588354e-06,
+      "loss": 0.4658,
+      "step": 3240
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.596800318718651e-06,
+      "loss": 0.4041,
+      "step": 3241
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.591532648914755e-06,
+      "loss": 0.4989,
+      "step": 3242
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.586265092640777e-06,
+      "loss": 0.4155,
+      "step": 3243
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.580997651360788e-06,
+      "loss": 0.466,
+      "step": 3244
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.575730326538822e-06,
+      "loss": 0.4474,
+      "step": 3245
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57046311963889e-06,
+      "loss": 0.4843,
+      "step": 3246
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56519603212496e-06,
+      "loss": 0.5057,
+      "step": 3247
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.55992906546098e-06,
+      "loss": 0.4667,
+      "step": 3248
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.554662221110852e-06,
+      "loss": 0.4233,
+      "step": 3249
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.549395500538454e-06,
+      "loss": 0.4172,
+      "step": 3250
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.544128905207618e-06,
+      "loss": 0.441,
+      "step": 3251
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.53886243658215e-06,
+      "loss": 0.4537,
+      "step": 3252
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.533596096125826e-06,
+      "loss": 0.4686,
+      "step": 3253
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.528329885302371e-06,
+      "loss": 0.4744,
+      "step": 3254
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.523063805575486e-06,
+      "loss": 0.401,
+      "step": 3255
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.517797858408831e-06,
+      "loss": 0.4902,
+      "step": 3256
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.512532045266023e-06,
+      "loss": 0.45,
+      "step": 3257
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50726636761066e-06,
+      "loss": 0.4914,
+      "step": 3258
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.502000826906286e-06,
+      "loss": 0.5373,
+      "step": 3259
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.496735424616408e-06,
+      "loss": 0.4327,
+      "step": 3260
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.491470162204504e-06,
+      "loss": 0.4458,
+      "step": 3261
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.486205041134003e-06,
+      "loss": 0.509,
+      "step": 3262
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.480940062868304e-06,
+      "loss": 0.469,
+      "step": 3263
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.475675228870757e-06,
+      "loss": 0.4362,
+      "step": 3264
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.47041054060468e-06,
+      "loss": 0.4736,
+      "step": 3265
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.465145999533342e-06,
+      "loss": 0.4407,
+      "step": 3266
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45988160711998e-06,
+      "loss": 0.4737,
+      "step": 3267
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.454617364827787e-06,
+      "loss": 0.444,
+      "step": 3268
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44935327411991e-06,
+      "loss": 0.4063,
+      "step": 3269
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.444089336459457e-06,
+      "loss": 0.4538,
+      "step": 3270
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.438825553309492e-06,
+      "loss": 0.4781,
+      "step": 3271
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.433561926133043e-06,
+      "loss": 0.4295,
+      "step": 3272
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.428298456393082e-06,
+      "loss": 0.4771,
+      "step": 3273
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.423035145552551e-06,
+      "loss": 0.4385,
+      "step": 3274
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.417771995074334e-06,
+      "loss": 0.4343,
+      "step": 3275
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.412509006421279e-06,
+      "loss": 0.4357,
+      "step": 3276
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.407246181056194e-06,
+      "loss": 0.4254,
+      "step": 3277
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40198352044183e-06,
+      "loss": 0.4632,
+      "step": 3278
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.396721026040896e-06,
+      "loss": 0.4577,
+      "step": 3279
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.391458699316061e-06,
+      "loss": 0.4935,
+      "step": 3280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.386196541729932e-06,
+      "loss": 0.4394,
+      "step": 3281
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.380934554745093e-06,
+      "loss": 0.4579,
+      "step": 3282
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.37567273982406e-06,
+      "loss": 0.4771,
+      "step": 3283
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.370411098429312e-06,
+      "loss": 0.4149,
+      "step": 3284
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.36514963202327e-06,
+      "loss": 0.4336,
+      "step": 3285
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.359888342068313e-06,
+      "loss": 0.4639,
+      "step": 3286
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.35462723002678e-06,
+      "loss": 0.4937,
+      "step": 3287
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.34936629736094e-06,
+      "loss": 0.4703,
+      "step": 3288
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.34410554553303e-06,
+      "loss": 0.4164,
+      "step": 3289
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.338844976005223e-06,
+      "loss": 0.426,
+      "step": 3290
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.333584590239652e-06,
+      "loss": 0.4479,
+      "step": 3291
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.3283243896984e-06,
+      "loss": 0.4039,
+      "step": 3292
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.323064375843486e-06,
+      "loss": 0.5328,
+      "step": 3293
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.317804550136888e-06,
+      "loss": 0.4136,
+      "step": 3294
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.312544914040529e-06,
+      "loss": 0.4194,
+      "step": 3295
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.307285469016272e-06,
+      "loss": 0.4308,
+      "step": 3296
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.302026216525943e-06,
+      "loss": 0.3757,
+      "step": 3297
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.296767158031303e-06,
+      "loss": 0.4114,
+      "step": 3298
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.291508294994058e-06,
+      "loss": 0.4403,
+      "step": 3299
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.286249628875863e-06,
+      "loss": 0.5162,
+      "step": 3300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280991161138316e-06,
+      "loss": 0.3832,
+      "step": 3301
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.27573289324297e-06,
+      "loss": 0.5199,
+      "step": 3302
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.270474826651308e-06,
+      "loss": 0.451,
+      "step": 3303
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.265216962824766e-06,
+      "loss": 0.4386,
+      "step": 3304
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.259959303224721e-06,
+      "loss": 0.4872,
+      "step": 3305
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.254701849312488e-06,
+      "loss": 0.4586,
+      "step": 3306
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.249444602549339e-06,
+      "loss": 0.3728,
+      "step": 3307
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.244187564396474e-06,
+      "loss": 0.4068,
+      "step": 3308
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.238930736315042e-06,
+      "loss": 0.3905,
+      "step": 3309
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.233674119766132e-06,
+      "loss": 0.5054,
+      "step": 3310
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.228417716210773e-06,
+      "loss": 0.4309,
+      "step": 3311
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.223161527109938e-06,
+      "loss": 0.4885,
+      "step": 3312
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21790555392454e-06,
+      "loss": 0.4397,
+      "step": 3313
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21264979811543e-06,
+      "loss": 0.4214,
+      "step": 3314
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.207394261143396e-06,
+      "loss": 0.4351,
+      "step": 3315
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.202138944469168e-06,
+      "loss": 0.4651,
+      "step": 3316
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196883849553423e-06,
+      "loss": 0.4654,
+      "step": 3317
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.191628977856761e-06,
+      "loss": 0.4749,
+      "step": 3318
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.186374330839732e-06,
+      "loss": 0.412,
+      "step": 3319
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.181119909962814e-06,
+      "loss": 0.5026,
+      "step": 3320
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.175865716686429e-06,
+      "loss": 0.4571,
+      "step": 3321
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.170611752470934e-06,
+      "loss": 0.5235,
+      "step": 3322
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.165358018776627e-06,
+      "loss": 0.4064,
+      "step": 3323
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16010451706373e-06,
+      "loss": 0.3819,
+      "step": 3324
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.154851248792415e-06,
+      "loss": 0.4602,
+      "step": 3325
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.149598215422773e-06,
+      "loss": 0.4245,
+      "step": 3326
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.144345418414846e-06,
+      "loss": 0.485,
+      "step": 3327
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.139092859228599e-06,
+      "loss": 0.4212,
+      "step": 3328
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.133840539323938e-06,
+      "loss": 0.4132,
+      "step": 3329
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.128588460160694e-06,
+      "loss": 0.4196,
+      "step": 3330
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.123336623198638e-06,
+      "loss": 0.4877,
+      "step": 3331
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.118085029897477e-06,
+      "loss": 0.4245,
+      "step": 3332
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.11283368171684e-06,
+      "loss": 0.4671,
+      "step": 3333
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.107582580116296e-06,
+      "loss": 0.4188,
+      "step": 3334
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.102331726555342e-06,
+      "loss": 0.4795,
+      "step": 3335
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.097081122493401e-06,
+      "loss": 0.5159,
+      "step": 3336
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.091830769389838e-06,
+      "loss": 0.3809,
+      "step": 3337
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.086580668703946e-06,
+      "loss": 0.4101,
+      "step": 3338
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.081330821894938e-06,
+      "loss": 0.4409,
+      "step": 3339
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.076081230421967e-06,
+      "loss": 0.4956,
+      "step": 3340
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.070831895744104e-06,
+      "loss": 0.4853,
+      "step": 3341
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.065582819320366e-06,
+      "loss": 0.4377,
+      "step": 3342
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.06033400260968e-06,
+      "loss": 0.4414,
+      "step": 3343
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.055085447070911e-06,
+      "loss": 0.5203,
+      "step": 3344
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.049837154162848e-06,
+      "loss": 0.5031,
+      "step": 3345
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.044589125344206e-06,
+      "loss": 0.3643,
+      "step": 3346
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.039341362073638e-06,
+      "loss": 0.469,
+      "step": 3347
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.0340938658097e-06,
+      "loss": 0.5087,
+      "step": 3348
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.0288466380109e-06,
+      "loss": 0.4369,
+      "step": 3349
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.023599680135652e-06,
+      "loss": 0.48,
+      "step": 3350
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.018352993642299e-06,
+      "loss": 0.5101,
+      "step": 3351
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.013106579989119e-06,
+      "loss": 0.4532,
+      "step": 3352
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.007860440634304e-06,
+      "loss": 0.442,
+      "step": 3353
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.00261457703597e-06,
+      "loss": 0.4603,
+      "step": 3354
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.997368990652163e-06,
+      "loss": 0.4211,
+      "step": 3355
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.99212368294084e-06,
+      "loss": 0.4655,
+      "step": 3356
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.986878655359898e-06,
+      "loss": 0.4637,
+      "step": 3357
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.981633909367141e-06,
+      "loss": 0.42,
+      "step": 3358
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.976389446420301e-06,
+      "loss": 0.4404,
+      "step": 3359
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.97114526797703e-06,
+      "loss": 0.4631,
+      "step": 3360
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.965901375494899e-06,
+      "loss": 0.4378,
+      "step": 3361
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.960657770431409e-06,
+      "loss": 0.3952,
+      "step": 3362
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.955414454243968e-06,
+      "loss": 0.4623,
+      "step": 3363
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.950171428389912e-06,
+      "loss": 0.4413,
+      "step": 3364
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94492869432649e-06,
+      "loss": 0.462,
+      "step": 3365
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93968625351088e-06,
+      "loss": 0.4591,
+      "step": 3366
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.934444107400168e-06,
+      "loss": 0.4534,
+      "step": 3367
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.929202257451368e-06,
+      "loss": 0.4361,
+      "step": 3368
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.923960705121398e-06,
+      "loss": 0.4443,
+      "step": 3369
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.918719451867105e-06,
+      "loss": 0.4646,
+      "step": 3370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.913478499145255e-06,
+      "loss": 0.4427,
+      "step": 3371
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.90823784841252e-06,
+      "loss": 0.4531,
+      "step": 3372
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.902997501125489e-06,
+      "loss": 0.4399,
+      "step": 3373
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897757458740678e-06,
+      "loss": 0.4245,
+      "step": 3374
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.892517722714502e-06,
+      "loss": 0.4112,
+      "step": 3375
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.887278294503307e-06,
+      "loss": 0.4455,
+      "step": 3376
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.882039175563347e-06,
+      "loss": 0.4621,
+      "step": 3377
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.876800367350786e-06,
+      "loss": 0.4402,
+      "step": 3378
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.871561871321702e-06,
+      "loss": 0.4355,
+      "step": 3379
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.866323688932089e-06,
+      "loss": 0.4405,
+      "step": 3380
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.861085821637862e-06,
+      "loss": 0.4923,
+      "step": 3381
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.855848270894832e-06,
+      "loss": 0.536,
+      "step": 3382
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.850611038158736e-06,
+      "loss": 0.435,
+      "step": 3383
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.845374124885214e-06,
+      "loss": 0.4946,
+      "step": 3384
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.840137532529815e-06,
+      "loss": 0.4665,
+      "step": 3385
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.834901262548013e-06,
+      "loss": 0.4824,
+      "step": 3386
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.82966531639518e-06,
+      "loss": 0.4275,
+      "step": 3387
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.824429695526597e-06,
+      "loss": 0.471,
+      "step": 3388
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.819194401397466e-06,
+      "loss": 0.4345,
+      "step": 3389
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.813959435462882e-06,
+      "loss": 0.4087,
+      "step": 3390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.808724799177868e-06,
+      "loss": 0.4204,
+      "step": 3391
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.80349049399734e-06,
+      "loss": 0.3797,
+      "step": 3392
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.798256521376128e-06,
+      "loss": 0.4224,
+      "step": 3393
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.793022882768968e-06,
+      "loss": 0.4447,
+      "step": 3394
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.787789579630501e-06,
+      "loss": 0.4101,
+      "step": 3395
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.782556613415285e-06,
+      "loss": 0.5108,
+      "step": 3396
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.777323985577772e-06,
+      "loss": 0.4628,
+      "step": 3397
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.772091697572328e-06,
+      "loss": 0.468,
+      "step": 3398
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.766859750853219e-06,
+      "loss": 0.4996,
+      "step": 3399
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.761628146874614e-06,
+      "loss": 0.3734,
+      "step": 3400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.7563968870906e-06,
+      "loss": 0.4927,
+      "step": 3401
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.751165972955158e-06,
+      "loss": 0.3818,
+      "step": 3402
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.74593540592217e-06,
+      "loss": 0.4302,
+      "step": 3403
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.740705187445433e-06,
+      "loss": 0.4553,
+      "step": 3404
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.735475318978633e-06,
+      "loss": 0.3919,
+      "step": 3405
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.73024580197537e-06,
+      "loss": 0.4375,
+      "step": 3406
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.725016637889144e-06,
+      "loss": 0.444,
+      "step": 3407
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.719787828173354e-06,
+      "loss": 0.3896,
+      "step": 3408
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.7145593742813e-06,
+      "loss": 0.4958,
+      "step": 3409
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.709331277666182e-06,
+      "loss": 0.4379,
+      "step": 3410
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.704103539781112e-06,
+      "loss": 0.4678,
+      "step": 3411
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.698876162079087e-06,
+      "loss": 0.4555,
+      "step": 3412
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.693649146013015e-06,
+      "loss": 0.3755,
+      "step": 3413
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.688422493035698e-06,
+      "loss": 0.4662,
+      "step": 3414
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.683196204599834e-06,
+      "loss": 0.4723,
+      "step": 3415
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.677970282158029e-06,
+      "loss": 0.4605,
+      "step": 3416
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.672744727162782e-06,
+      "loss": 0.4612,
+      "step": 3417
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.667519541066488e-06,
+      "loss": 0.4033,
+      "step": 3418
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.662294725321445e-06,
+      "loss": 0.4021,
+      "step": 3419
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.657070281379837e-06,
+      "loss": 0.4359,
+      "step": 3420
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.651846210693764e-06,
+      "loss": 0.4299,
+      "step": 3421
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6466225147152e-06,
+      "loss": 0.4051,
+      "step": 3422
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.641399194896032e-06,
+      "loss": 0.4498,
+      "step": 3423
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63617625268803e-06,
+      "loss": 0.3767,
+      "step": 3424
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63095368954287e-06,
+      "loss": 0.3929,
+      "step": 3425
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.625731506912118e-06,
+      "loss": 0.4205,
+      "step": 3426
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.620509706247231e-06,
+      "loss": 0.4291,
+      "step": 3427
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.615288288999565e-06,
+      "loss": 0.4195,
+      "step": 3428
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.610067256620367e-06,
+      "loss": 0.4792,
+      "step": 3429
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.604846610560771e-06,
+      "loss": 0.4488,
+      "step": 3430
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.599626352271817e-06,
+      "loss": 0.5018,
+      "step": 3431
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.594406483204429e-06,
+      "loss": 0.4483,
+      "step": 3432
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.58918700480942e-06,
+      "loss": 0.3997,
+      "step": 3433
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.583967918537503e-06,
+      "loss": 0.4496,
+      "step": 3434
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.57874922583927e-06,
+      "loss": 0.419,
+      "step": 3435
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.573530928165221e-06,
+      "loss": 0.4928,
+      "step": 3436
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.568313026965729e-06,
+      "loss": 0.405,
+      "step": 3437
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.563095523691066e-06,
+      "loss": 0.4696,
+      "step": 3438
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55787841979139e-06,
+      "loss": 0.4378,
+      "step": 3439
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.552661716716747e-06,
+      "loss": 0.454,
+      "step": 3440
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.547445415917082e-06,
+      "loss": 0.432,
+      "step": 3441
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.542229518842214e-06,
+      "loss": 0.4739,
+      "step": 3442
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.537014026941854e-06,
+      "loss": 0.4226,
+      "step": 3443
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.531798941665607e-06,
+      "loss": 0.4879,
+      "step": 3444
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.526584264462956e-06,
+      "loss": 0.4633,
+      "step": 3445
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.521369996783278e-06,
+      "loss": 0.4111,
+      "step": 3446
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.516156140075834e-06,
+      "loss": 0.4416,
+      "step": 3447
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.510942695789763e-06,
+      "loss": 0.4213,
+      "step": 3448
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.505729665374106e-06,
+      "loss": 0.4583,
+      "step": 3449
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.500517050277766e-06,
+      "loss": 0.5342,
+      "step": 3450
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.495304851949558e-06,
+      "loss": 0.4129,
+      "step": 3451
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.490093071838156e-06,
+      "loss": 0.4411,
+      "step": 3452
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.484881711392137e-06,
+      "loss": 0.5167,
+      "step": 3453
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.479670772059944e-06,
+      "loss": 0.3877,
+      "step": 3454
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.474460255289915e-06,
+      "loss": 0.3716,
+      "step": 3455
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.469250162530273e-06,
+      "loss": 0.4447,
+      "step": 3456
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.464040495229114e-06,
+      "loss": 0.4107,
+      "step": 3457
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.458831254834415e-06,
+      "loss": 0.4271,
+      "step": 3458
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.453622442794042e-06,
+      "loss": 0.409,
+      "step": 3459
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.448414060555741e-06,
+      "loss": 0.3989,
+      "step": 3460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.443206109567133e-06,
+      "loss": 0.424,
+      "step": 3461
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.437998591275725e-06,
+      "loss": 0.4567,
+      "step": 3462
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.432791507128897e-06,
+      "loss": 0.4294,
+      "step": 3463
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.427584858573912e-06,
+      "loss": 0.4738,
+      "step": 3464
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.422378647057919e-06,
+      "loss": 0.4439,
+      "step": 3465
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.417172874027936e-06,
+      "loss": 0.4199,
+      "step": 3466
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.411967540930856e-06,
+      "loss": 0.478,
+      "step": 3467
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.406762649213464e-06,
+      "loss": 0.4237,
+      "step": 3468
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.401558200322405e-06,
+      "loss": 0.4728,
+      "step": 3469
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.396354195704219e-06,
+      "loss": 0.4799,
+      "step": 3470
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.39115063680531e-06,
+      "loss": 0.4076,
+      "step": 3471
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.38594752507196e-06,
+      "loss": 0.4661,
+      "step": 3472
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.38074486195033e-06,
+      "loss": 0.4739,
+      "step": 3473
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.375542648886448e-06,
+      "loss": 0.4729,
+      "step": 3474
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.370340887326236e-06,
+      "loss": 0.4563,
+      "step": 3475
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.365139578715468e-06,
+      "loss": 0.5126,
+      "step": 3476
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.359938724499808e-06,
+      "loss": 0.4535,
+      "step": 3477
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.354738326124785e-06,
+      "loss": 0.4172,
+      "step": 3478
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3495383850358e-06,
+      "loss": 0.4515,
+      "step": 3479
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.344338902678136e-06,
+      "loss": 0.4672,
+      "step": 3480
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.339139880496945e-06,
+      "loss": 0.4267,
+      "step": 3481
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.333941319937247e-06,
+      "loss": 0.4083,
+      "step": 3482
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.328743222443937e-06,
+      "loss": 0.4732,
+      "step": 3483
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.323545589461775e-06,
+      "loss": 0.4327,
+      "step": 3484
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.318348422435406e-06,
+      "loss": 0.4578,
+      "step": 3485
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.313151722809334e-06,
+      "loss": 0.4177,
+      "step": 3486
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.307955492027938e-06,
+      "loss": 0.4392,
+      "step": 3487
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.302759731535459e-06,
+      "loss": 0.4605,
+      "step": 3488
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.297564442776014e-06,
+      "loss": 0.4627,
+      "step": 3489
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.292369627193593e-06,
+      "loss": 0.3946,
+      "step": 3490
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.287175286232046e-06,
+      "loss": 0.4534,
+      "step": 3491
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.281981421335097e-06,
+      "loss": 0.5246,
+      "step": 3492
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.276788033946335e-06,
+      "loss": 0.438,
+      "step": 3493
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.271595125509207e-06,
+      "loss": 0.4135,
+      "step": 3494
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.266402697467049e-06,
+      "loss": 0.4415,
+      "step": 3495
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.261210751263046e-06,
+      "loss": 0.4609,
+      "step": 3496
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.256019288340253e-06,
+      "loss": 0.4038,
+      "step": 3497
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.250828310141594e-06,
+      "loss": 0.4332,
+      "step": 3498
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.24563781810985e-06,
+      "loss": 0.5207,
+      "step": 3499
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.24044781368768e-06,
+      "loss": 0.4508,
+      "step": 3500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.235258298317595e-06,
+      "loss": 0.4132,
+      "step": 3501
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23006927344198e-06,
+      "loss": 0.3868,
+      "step": 3502
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.224880740503072e-06,
+      "loss": 0.5106,
+      "step": 3503
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.219692700942981e-06,
+      "loss": 0.3944,
+      "step": 3504
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.214505156203685e-06,
+      "loss": 0.4508,
+      "step": 3505
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.209318107727006e-06,
+      "loss": 0.4163,
+      "step": 3506
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.204131556954644e-06,
+      "loss": 0.4993,
+      "step": 3507
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.198945505328156e-06,
+      "loss": 0.4703,
+      "step": 3508
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.19375995428895e-06,
+      "loss": 0.4776,
+      "step": 3509
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.188574905278316e-06,
+      "loss": 0.4876,
+      "step": 3510
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18339035973739e-06,
+      "loss": 0.4708,
+      "step": 3511
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.178206319107166e-06,
+      "loss": 0.4604,
+      "step": 3512
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.173022784828511e-06,
+      "loss": 0.4291,
+      "step": 3513
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.167839758342131e-06,
+      "loss": 0.4356,
+      "step": 3514
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.162657241088616e-06,
+      "loss": 0.502,
+      "step": 3515
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.157475234508393e-06,
+      "loss": 0.4527,
+      "step": 3516
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.152293740041758e-06,
+      "loss": 0.5209,
+      "step": 3517
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147112759128859e-06,
+      "loss": 0.4416,
+      "step": 3518
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.141932293209704e-06,
+      "loss": 0.462,
+      "step": 3519
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.136752343724164e-06,
+      "loss": 0.4423,
+      "step": 3520
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.131572912111956e-06,
+      "loss": 0.4251,
+      "step": 3521
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12639399981266e-06,
+      "loss": 0.447,
+      "step": 3522
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.121215608265707e-06,
+      "loss": 0.397,
+      "step": 3523
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11603773891038e-06,
+      "loss": 0.487,
+      "step": 3524
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.110860393185833e-06,
+      "loss": 0.4346,
+      "step": 3525
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10568357253106e-06,
+      "loss": 0.4515,
+      "step": 3526
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10050727838491e-06,
+      "loss": 0.4043,
+      "step": 3527
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.095331512186093e-06,
+      "loss": 0.4462,
+      "step": 3528
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.090156275373163e-06,
+      "loss": 0.5042,
+      "step": 3529
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.084981569384535e-06,
+      "loss": 0.3614,
+      "step": 3530
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07980739565847e-06,
+      "loss": 0.4308,
+      "step": 3531
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07463375563309e-06,
+      "loss": 0.3961,
+      "step": 3532
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.069460650746357e-06,
+      "loss": 0.4389,
+      "step": 3533
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06428808243609e-06,
+      "loss": 0.3951,
+      "step": 3534
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.059116052139964e-06,
+      "loss": 0.4421,
+      "step": 3535
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.053944561295496e-06,
+      "loss": 0.4291,
+      "step": 3536
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.048773611340057e-06,
+      "loss": 0.4416,
+      "step": 3537
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.043603203710868e-06,
+      "loss": 0.4876,
+      "step": 3538
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.038433339844995e-06,
+      "loss": 0.4162,
+      "step": 3539
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.033264021179359e-06,
+      "loss": 0.5177,
+      "step": 3540
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.02809524915073e-06,
+      "loss": 0.4443,
+      "step": 3541
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.022927025195716e-06,
+      "loss": 0.3837,
+      "step": 3542
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.017759350750786e-06,
+      "loss": 0.4159,
+      "step": 3543
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.012592227252243e-06,
+      "loss": 0.4902,
+      "step": 3544
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.007425656136252e-06,
+      "loss": 0.4381,
+      "step": 3545
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.002259638838809e-06,
+      "loss": 0.3839,
+      "step": 3546
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.997094176795766e-06,
+      "loss": 0.4944,
+      "step": 3547
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.991929271442817e-06,
+      "loss": 0.4434,
+      "step": 3548
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.9867649242155e-06,
+      "loss": 0.488,
+      "step": 3549
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.981601136549207e-06,
+      "loss": 0.4206,
+      "step": 3550
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.976437909879163e-06,
+      "loss": 0.4279,
+      "step": 3551
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.97127524564044e-06,
+      "loss": 0.4164,
+      "step": 3552
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.966113145267957e-06,
+      "loss": 0.4593,
+      "step": 3553
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.960951610196472e-06,
+      "loss": 0.3803,
+      "step": 3554
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.955790641860592e-06,
+      "loss": 0.4077,
+      "step": 3555
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.950630241694763e-06,
+      "loss": 0.4442,
+      "step": 3556
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.945470411133272e-06,
+      "loss": 0.5085,
+      "step": 3557
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940311151610244e-06,
+      "loss": 0.4861,
+      "step": 3558
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.935152464559655e-06,
+      "loss": 0.4621,
+      "step": 3559
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.929994351415319e-06,
+      "loss": 0.4178,
+      "step": 3560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.924836813610881e-06,
+      "loss": 0.4154,
+      "step": 3561
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.919679852579843e-06,
+      "loss": 0.4502,
+      "step": 3562
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.914523469755524e-06,
+      "loss": 0.3997,
+      "step": 3563
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.909367666571106e-06,
+      "loss": 0.4657,
+      "step": 3564
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.904212444459598e-06,
+      "loss": 0.4451,
+      "step": 3565
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899057804853847e-06,
+      "loss": 0.3996,
+      "step": 3566
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.893903749186538e-06,
+      "loss": 0.4777,
+      "step": 3567
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.888750278890194e-06,
+      "loss": 0.4483,
+      "step": 3568
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.883597395397184e-06,
+      "loss": 0.512,
+      "step": 3569
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.878445100139702e-06,
+      "loss": 0.4034,
+      "step": 3570
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.873293394549785e-06,
+      "loss": 0.4617,
+      "step": 3571
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.868142280059303e-06,
+      "loss": 0.4236,
+      "step": 3572
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86299175809996e-06,
+      "loss": 0.4045,
+      "step": 3573
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.857841830103303e-06,
+      "loss": 0.4218,
+      "step": 3574
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.852692497500712e-06,
+      "loss": 0.467,
+      "step": 3575
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84754376172339e-06,
+      "loss": 0.4063,
+      "step": 3576
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.842395624202392e-06,
+      "loss": 0.4649,
+      "step": 3577
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.837248086368587e-06,
+      "loss": 0.4764,
+      "step": 3578
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.832101149652699e-06,
+      "loss": 0.4808,
+      "step": 3579
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82695481548527e-06,
+      "loss": 0.3938,
+      "step": 3580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.821809085296678e-06,
+      "loss": 0.4583,
+      "step": 3581
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81666396051713e-06,
+      "loss": 0.4401,
+      "step": 3582
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81151944257667e-06,
+      "loss": 0.3832,
+      "step": 3583
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.806375532905178e-06,
+      "loss": 0.4511,
+      "step": 3584
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.801232232932351e-06,
+      "loss": 0.4194,
+      "step": 3585
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.796089544087729e-06,
+      "loss": 0.3798,
+      "step": 3586
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.790947467800675e-06,
+      "loss": 0.4409,
+      "step": 3587
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.785806005500379e-06,
+      "loss": 0.453,
+      "step": 3588
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.780665158615873e-06,
+      "loss": 0.4098,
+      "step": 3589
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.775524928576009e-06,
+      "loss": 0.4292,
+      "step": 3590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.770385316809465e-06,
+      "loss": 0.4315,
+      "step": 3591
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.765246324744757e-06,
+      "loss": 0.5043,
+      "step": 3592
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.760107953810215e-06,
+      "loss": 0.4115,
+      "step": 3593
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.75497020543401e-06,
+      "loss": 0.4151,
+      "step": 3594
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.749833081044131e-06,
+      "loss": 0.3766,
+      "step": 3595
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.7446965820684e-06,
+      "loss": 0.4475,
+      "step": 3596
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73956070993446e-06,
+      "loss": 0.4488,
+      "step": 3597
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.734425466069776e-06,
+      "loss": 0.4423,
+      "step": 3598
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.729290851901654e-06,
+      "loss": 0.4104,
+      "step": 3599
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72415686885721e-06,
+      "loss": 0.4145,
+      "step": 3600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719023518363389e-06,
+      "loss": 0.4098,
+      "step": 3601
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.713890801846963e-06,
+      "loss": 0.4453,
+      "step": 3602
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.708758720734519e-06,
+      "loss": 0.4914,
+      "step": 3603
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.703627276452481e-06,
+      "loss": 0.4663,
+      "step": 3604
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69849647042709e-06,
+      "loss": 0.4188,
+      "step": 3605
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.693366304084401e-06,
+      "loss": 0.4395,
+      "step": 3606
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.688236778850307e-06,
+      "loss": 0.4222,
+      "step": 3607
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.683107896150504e-06,
+      "loss": 0.4055,
+      "step": 3608
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.677979657410532e-06,
+      "loss": 0.4469,
+      "step": 3609
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.672852064055733e-06,
+      "loss": 0.4015,
+      "step": 3610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.667725117511279e-06,
+      "loss": 0.4412,
+      "step": 3611
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.662598819202156e-06,
+      "loss": 0.5011,
+      "step": 3612
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.657473170553176e-06,
+      "loss": 0.3988,
+      "step": 3613
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65234817298897e-06,
+      "loss": 0.4401,
+      "step": 3614
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.647223827933987e-06,
+      "loss": 0.4407,
+      "step": 3615
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.642100136812486e-06,
+      "loss": 0.4454,
+      "step": 3616
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.636977101048563e-06,
+      "loss": 0.4528,
+      "step": 3617
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.631854722066109e-06,
+      "loss": 0.4973,
+      "step": 3618
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.626733001288852e-06,
+      "loss": 0.4053,
+      "step": 3619
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.62161194014033e-06,
+      "loss": 0.3931,
+      "step": 3620
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6164915400438925e-06,
+      "loss": 0.4762,
+      "step": 3621
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.611371802422713e-06,
+      "loss": 0.3826,
+      "step": 3622
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6062527286997745e-06,
+      "loss": 0.4126,
+      "step": 3623
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.601134320297884e-06,
+      "loss": 0.4271,
+      "step": 3624
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.596016578639651e-06,
+      "loss": 0.3797,
+      "step": 3625
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.590899505147516e-06,
+      "loss": 0.421,
+      "step": 3626
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.585783101243714e-06,
+      "loss": 0.3637,
+      "step": 3627
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5806673683503094e-06,
+      "loss": 0.4316,
+      "step": 3628
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5755523078891794e-06,
+      "loss": 0.4246,
+      "step": 3629
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.570437921282007e-06,
+      "loss": 0.4138,
+      "step": 3630
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.565324209950289e-06,
+      "loss": 0.4572,
+      "step": 3631
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.560211175315338e-06,
+      "loss": 0.4155,
+      "step": 3632
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.555098818798276e-06,
+      "loss": 0.4518,
+      "step": 3633
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.549987141820038e-06,
+      "loss": 0.4497,
+      "step": 3634
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.544876145801375e-06,
+      "loss": 0.4168,
+      "step": 3635
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.539765832162838e-06,
+      "loss": 0.4054,
+      "step": 3636
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.534656202324795e-06,
+      "loss": 0.3578,
+      "step": 3637
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.529547257707418e-06,
+      "loss": 0.493,
+      "step": 3638
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.524438999730703e-06,
+      "loss": 0.4383,
+      "step": 3639
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.51933142981444e-06,
+      "loss": 0.4591,
+      "step": 3640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.514224549378234e-06,
+      "loss": 0.4051,
+      "step": 3641
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.509118359841496e-06,
+      "loss": 0.5123,
+      "step": 3642
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.504012862623446e-06,
+      "loss": 0.414,
+      "step": 3643
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.498908059143117e-06,
+      "loss": 0.4477,
+      "step": 3644
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.493803950819343e-06,
+      "loss": 0.483,
+      "step": 3645
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4887005390707615e-06,
+      "loss": 0.4048,
+      "step": 3646
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4835978253158246e-06,
+      "loss": 0.4434,
+      "step": 3647
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.478495810972785e-06,
+      "loss": 0.3921,
+      "step": 3648
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4733944974597036e-06,
+      "loss": 0.4331,
+      "step": 3649
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.46829388619445e-06,
+      "loss": 0.4007,
+      "step": 3650
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.463193978594689e-06,
+      "loss": 0.3948,
+      "step": 3651
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.458094776077895e-06,
+      "loss": 0.4196,
+      "step": 3652
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.452996280061346e-06,
+      "loss": 0.4043,
+      "step": 3653
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.447898491962131e-06,
+      "loss": 0.4355,
+      "step": 3654
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.442801413197129e-06,
+      "loss": 0.4837,
+      "step": 3655
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.437705045183032e-06,
+      "loss": 0.4082,
+      "step": 3656
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.432609389336325e-06,
+      "loss": 0.4548,
+      "step": 3657
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.427514447073307e-06,
+      "loss": 0.4253,
+      "step": 3658
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.422420219810071e-06,
+      "loss": 0.4392,
+      "step": 3659
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.417326708962513e-06,
+      "loss": 0.4319,
+      "step": 3660
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4122339159463275e-06,
+      "loss": 0.3552,
+      "step": 3661
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.407141842177009e-06,
+      "loss": 0.4669,
+      "step": 3662
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.402050489069863e-06,
+      "loss": 0.4678,
+      "step": 3663
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.39695985803998e-06,
+      "loss": 0.4886,
+      "step": 3664
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.391869950502259e-06,
+      "loss": 0.4592,
+      "step": 3665
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3867807678713965e-06,
+      "loss": 0.4443,
+      "step": 3666
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.381692311561879e-06,
+      "loss": 0.497,
+      "step": 3667
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.376604582988005e-06,
+      "loss": 0.4621,
+      "step": 3668
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.371517583563863e-06,
+      "loss": 0.4719,
+      "step": 3669
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.366431314703338e-06,
+      "loss": 0.5161,
+      "step": 3670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.361345777820117e-06,
+      "loss": 0.3922,
+      "step": 3671
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.356260974327673e-06,
+      "loss": 0.4376,
+      "step": 3672
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3511769056392935e-06,
+      "loss": 0.4375,
+      "step": 3673
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.346093573168041e-06,
+      "loss": 0.4707,
+      "step": 3674
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.34101097832679e-06,
+      "loss": 0.4293,
+      "step": 3675
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.335929122528198e-06,
+      "loss": 0.4081,
+      "step": 3676
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.330848007184723e-06,
+      "loss": 0.4462,
+      "step": 3677
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.32576763370862e-06,
+      "loss": 0.4284,
+      "step": 3678
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.320688003511932e-06,
+      "loss": 0.5008,
+      "step": 3679
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3156091180064994e-06,
+      "loss": 0.4749,
+      "step": 3680
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3105309786039515e-06,
+      "loss": 0.4314,
+      "step": 3681
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.305453586715709e-06,
+      "loss": 0.4415,
+      "step": 3682
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.300376943752994e-06,
+      "loss": 0.3942,
+      "step": 3683
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.295301051126816e-06,
+      "loss": 0.4456,
+      "step": 3684
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.29022591024797e-06,
+      "loss": 0.4684,
+      "step": 3685
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.285151522527052e-06,
+      "loss": 0.4638,
+      "step": 3686
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2800778893744365e-06,
+      "loss": 0.4174,
+      "step": 3687
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.275005012200303e-06,
+      "loss": 0.4175,
+      "step": 3688
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2699328924146065e-06,
+      "loss": 0.4428,
+      "step": 3689
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.264861531427105e-06,
+      "loss": 0.4288,
+      "step": 3690
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2597909306473325e-06,
+      "loss": 0.3935,
+      "step": 3691
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.254721091484618e-06,
+      "loss": 0.4227,
+      "step": 3692
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.249652015348085e-06,
+      "loss": 0.4881,
+      "step": 3693
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.244583703646635e-06,
+      "loss": 0.4929,
+      "step": 3694
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.239516157788962e-06,
+      "loss": 0.4119,
+      "step": 3695
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.234449379183545e-06,
+      "loss": 0.4718,
+      "step": 3696
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.229383369238645e-06,
+      "loss": 0.4812,
+      "step": 3697
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.224318129362323e-06,
+      "loss": 0.487,
+      "step": 3698
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.219253660962417e-06,
+      "loss": 0.461,
+      "step": 3699
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.21418996544655e-06,
+      "loss": 0.4156,
+      "step": 3700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.209127044222131e-06,
+      "loss": 0.4165,
+      "step": 3701
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.204064898696352e-06,
+      "loss": 0.3958,
+      "step": 3702
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1990035302761996e-06,
+      "loss": 0.4528,
+      "step": 3703
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.19394294036843e-06,
+      "loss": 0.4267,
+      "step": 3704
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.188883130379594e-06,
+      "loss": 0.4669,
+      "step": 3705
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183824101716017e-06,
+      "loss": 0.4671,
+      "step": 3706
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.178765855783811e-06,
+      "loss": 0.4139,
+      "step": 3707
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.173708393988879e-06,
+      "loss": 0.394,
+      "step": 3708
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.168651717736893e-06,
+      "loss": 0.4524,
+      "step": 3709
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.16359582843331e-06,
+      "loss": 0.4913,
+      "step": 3710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.158540727483374e-06,
+      "loss": 0.4574,
+      "step": 3711
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153486416292099e-06,
+      "loss": 0.4636,
+      "step": 3712
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1484328962642945e-06,
+      "loss": 0.4397,
+      "step": 3713
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.143380168804539e-06,
+      "loss": 0.3753,
+      "step": 3714
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1383282353171935e-06,
+      "loss": 0.5775,
+      "step": 3715
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.133277097206399e-06,
+      "loss": 0.3952,
+      "step": 3716
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.128226755876073e-06,
+      "loss": 0.4875,
+      "step": 3717
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.123177212729917e-06,
+      "loss": 0.4549,
+      "step": 3718
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.118128469171405e-06,
+      "loss": 0.4353,
+      "step": 3719
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.113080526603793e-06,
+      "loss": 0.4668,
+      "step": 3720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.108033386430107e-06,
+      "loss": 0.4518,
+      "step": 3721
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.102987050053159e-06,
+      "loss": 0.4491,
+      "step": 3722
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.097941518875538e-06,
+      "loss": 0.4292,
+      "step": 3723
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.092896794299599e-06,
+      "loss": 0.4512,
+      "step": 3724
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0878528777274814e-06,
+      "loss": 0.446,
+      "step": 3725
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.082809770561097e-06,
+      "loss": 0.4539,
+      "step": 3726
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.077767474202131e-06,
+      "loss": 0.4403,
+      "step": 3727
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.07272599005205e-06,
+      "loss": 0.4082,
+      "step": 3728
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.067685319512087e-06,
+      "loss": 0.451,
+      "step": 3729
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0626454639832555e-06,
+      "loss": 0.4763,
+      "step": 3730
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.057606424866335e-06,
+      "loss": 0.4549,
+      "step": 3731
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.05256820356188e-06,
+      "loss": 0.3747,
+      "step": 3732
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.04753080147023e-06,
+      "loss": 0.4066,
+      "step": 3733
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.042494219991477e-06,
+      "loss": 0.4258,
+      "step": 3734
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0374584605255006e-06,
+      "loss": 0.4059,
+      "step": 3735
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0324235244719405e-06,
+      "loss": 0.4193,
+      "step": 3736
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.027389413230215e-06,
+      "loss": 0.4412,
+      "step": 3737
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.022356128199517e-06,
+      "loss": 0.4246,
+      "step": 3738
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.017323670778797e-06,
+      "loss": 0.4111,
+      "step": 3739
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.012292042366784e-06,
+      "loss": 0.4182,
+      "step": 3740
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0072612443619755e-06,
+      "loss": 0.4605,
+      "step": 3741
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.002231278162634e-06,
+      "loss": 0.415,
+      "step": 3742
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.997202145166801e-06,
+      "loss": 0.4392,
+      "step": 3743
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.992173846772279e-06,
+      "loss": 0.4034,
+      "step": 3744
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.987146384376635e-06,
+      "loss": 0.4709,
+      "step": 3745
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.982119759377211e-06,
+      "loss": 0.5125,
+      "step": 3746
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.977093973171109e-06,
+      "loss": 0.436,
+      "step": 3747
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.97206902715521e-06,
+      "loss": 0.4641,
+      "step": 3748
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9670449227261474e-06,
+      "loss": 0.3944,
+      "step": 3749
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.962021661280329e-06,
+      "loss": 0.4135,
+      "step": 3750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.956999244213925e-06,
+      "loss": 0.4833,
+      "step": 3751
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.951977672922874e-06,
+      "loss": 0.4894,
+      "step": 3752
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.946956948802879e-06,
+      "loss": 0.4683,
+      "step": 3753
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.941937073249403e-06,
+      "loss": 0.4683,
+      "step": 3754
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9369180476576766e-06,
+      "loss": 0.4137,
+      "step": 3755
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9318998734226935e-06,
+      "loss": 0.375,
+      "step": 3756
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.926882551939216e-06,
+      "loss": 0.3895,
+      "step": 3757
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.921866084601758e-06,
+      "loss": 0.4472,
+      "step": 3758
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.916850472804608e-06,
+      "loss": 0.364,
+      "step": 3759
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.91183571794181e-06,
+      "loss": 0.4096,
+      "step": 3760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.906821821407166e-06,
+      "loss": 0.4096,
+      "step": 3761
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.901808784594248e-06,
+      "loss": 0.3869,
+      "step": 3762
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.89679660889639e-06,
+      "loss": 0.4204,
+      "step": 3763
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8917852957066766e-06,
+      "loss": 0.3744,
+      "step": 3764
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.886774846417963e-06,
+      "loss": 0.4277,
+      "step": 3765
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.88176526242285e-06,
+      "loss": 0.4406,
+      "step": 3766
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.876756545113721e-06,
+      "loss": 0.461,
+      "step": 3767
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.871748695882694e-06,
+      "loss": 0.4801,
+      "step": 3768
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8667417161216635e-06,
+      "loss": 0.4537,
+      "step": 3769
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.861735607222271e-06,
+      "loss": 0.4504,
+      "step": 3770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.856730370575921e-06,
+      "loss": 0.418,
+      "step": 3771
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.85172600757378e-06,
+      "loss": 0.4381,
+      "step": 3772
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.846722519606762e-06,
+      "loss": 0.4081,
+      "step": 3773
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.841719908065546e-06,
+      "loss": 0.4344,
+      "step": 3774
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.836718174340562e-06,
+      "loss": 0.4428,
+      "step": 3775
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.831717319821993e-06,
+      "loss": 0.446,
+      "step": 3776
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.82671734589979e-06,
+      "loss": 0.4466,
+      "step": 3777
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.821718253963652e-06,
+      "loss": 0.4276,
+      "step": 3778
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.816720045403028e-06,
+      "loss": 0.4819,
+      "step": 3779
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8117227216071304e-06,
+      "loss": 0.442,
+      "step": 3780
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806726283964915e-06,
+      "loss": 0.4735,
+      "step": 3781
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.801730733865106e-06,
+      "loss": 0.3833,
+      "step": 3782
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.796736072696169e-06,
+      "loss": 0.4267,
+      "step": 3783
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.791742301846325e-06,
+      "loss": 0.4215,
+      "step": 3784
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.78674942270355e-06,
+      "loss": 0.3856,
+      "step": 3785
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.78175743665557e-06,
+      "loss": 0.4624,
+      "step": 3786
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.776766345089865e-06,
+      "loss": 0.4358,
+      "step": 3787
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.771776149393667e-06,
+      "loss": 0.4562,
+      "step": 3788
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.766786850953953e-06,
+      "loss": 0.4271,
+      "step": 3789
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.761798451157459e-06,
+      "loss": 0.489,
+      "step": 3790
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.756810951390659e-06,
+      "loss": 0.4645,
+      "step": 3791
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.751824353039793e-06,
+      "loss": 0.442,
+      "step": 3792
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.746838657490841e-06,
+      "loss": 0.4335,
+      "step": 3793
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.74185386612953e-06,
+      "loss": 0.3987,
+      "step": 3794
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.736869980341343e-06,
+      "loss": 0.4518,
+      "step": 3795
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.731887001511501e-06,
+      "loss": 0.3966,
+      "step": 3796
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.726904931024987e-06,
+      "loss": 0.4823,
+      "step": 3797
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7219237702665185e-06,
+      "loss": 0.4464,
+      "step": 3798
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.716943520620569e-06,
+      "loss": 0.426,
+      "step": 3799
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7119641834713514e-06,
+      "loss": 0.4613,
+      "step": 3800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7069857602028275e-06,
+      "loss": 0.431,
+      "step": 3801
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.702008252198715e-06,
+      "loss": 0.484,
+      "step": 3802
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.697031660842461e-06,
+      "loss": 0.4729,
+      "step": 3803
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.692055987517267e-06,
+      "loss": 0.4904,
+      "step": 3804
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.68708123360608e-06,
+      "loss": 0.4386,
+      "step": 3805
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.682107400491582e-06,
+      "loss": 0.4476,
+      "step": 3806
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.677134489556215e-06,
+      "loss": 0.4516,
+      "step": 3807
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6721625021821536e-06,
+      "loss": 0.4231,
+      "step": 3808
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.667191439751316e-06,
+      "loss": 0.4299,
+      "step": 3809
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.662221303645367e-06,
+      "loss": 0.3991,
+      "step": 3810
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657252095245708e-06,
+      "loss": 0.4475,
+      "step": 3811
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.652283815933496e-06,
+      "loss": 0.4022,
+      "step": 3812
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6473164670896106e-06,
+      "loss": 0.4201,
+      "step": 3813
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6423500500946915e-06,
+      "loss": 0.4367,
+      "step": 3814
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.637384566329104e-06,
+      "loss": 0.4584,
+      "step": 3815
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.63242001717296e-06,
+      "loss": 0.4581,
+      "step": 3816
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.627456404006121e-06,
+      "loss": 0.4578,
+      "step": 3817
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.622493728208175e-06,
+      "loss": 0.433,
+      "step": 3818
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6175319911584505e-06,
+      "loss": 0.366,
+      "step": 3819
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.612571194236023e-06,
+      "loss": 0.4394,
+      "step": 3820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.607611338819697e-06,
+      "loss": 0.4136,
+      "step": 3821
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.602652426288029e-06,
+      "loss": 0.4644,
+      "step": 3822
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.597694458019302e-06,
+      "loss": 0.4259,
+      "step": 3823
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.592737435391539e-06,
+      "loss": 0.4276,
+      "step": 3824
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.587781359782499e-06,
+      "loss": 0.5032,
+      "step": 3825
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5828262325696815e-06,
+      "loss": 0.4031,
+      "step": 3826
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.577872055130323e-06,
+      "loss": 0.4228,
+      "step": 3827
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5729188288413895e-06,
+      "loss": 0.4511,
+      "step": 3828
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.56796655507959e-06,
+      "loss": 0.4624,
+      "step": 3829
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.563015235221364e-06,
+      "loss": 0.409,
+      "step": 3830
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.558064870642885e-06,
+      "loss": 0.4856,
+      "step": 3831
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.55311546272007e-06,
+      "loss": 0.4554,
+      "step": 3832
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.548167012828559e-06,
+      "loss": 0.4062,
+      "step": 3833
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.543219522343729e-06,
+      "loss": 0.4269,
+      "step": 3834
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.538272992640695e-06,
+      "loss": 0.4541,
+      "step": 3835
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.533327425094297e-06,
+      "loss": 0.4176,
+      "step": 3836
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.528382821079116e-06,
+      "loss": 0.4099,
+      "step": 3837
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.523439181969462e-06,
+      "loss": 0.4273,
+      "step": 3838
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5184965091393745e-06,
+      "loss": 0.3853,
+      "step": 3839
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5135548039626236e-06,
+      "loss": 0.4764,
+      "step": 3840
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.508614067812713e-06,
+      "loss": 0.3629,
+      "step": 3841
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.503674302062883e-06,
+      "loss": 0.3755,
+      "step": 3842
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.498735508086094e-06,
+      "loss": 0.4594,
+      "step": 3843
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.493797687255041e-06,
+      "loss": 0.3956,
+      "step": 3844
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.488860840942148e-06,
+      "loss": 0.3511,
+      "step": 3845
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.483924970519564e-06,
+      "loss": 0.3524,
+      "step": 3846
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4789900773591755e-06,
+      "loss": 0.4019,
+      "step": 3847
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.474056162832594e-06,
+      "loss": 0.3768,
+      "step": 3848
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.469123228311154e-06,
+      "loss": 0.3841,
+      "step": 3849
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464191275165921e-06,
+      "loss": 0.459,
+      "step": 3850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.459260304767694e-06,
+      "loss": 0.4136,
+      "step": 3851
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.454330318486986e-06,
+      "loss": 0.4218,
+      "step": 3852
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.449401317694049e-06,
+      "loss": 0.4622,
+      "step": 3853
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.444473303758853e-06,
+      "loss": 0.5492,
+      "step": 3854
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.439546278051092e-06,
+      "loss": 0.4602,
+      "step": 3855
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.434620241940195e-06,
+      "loss": 0.4666,
+      "step": 3856
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.429695196795312e-06,
+      "loss": 0.4337,
+      "step": 3857
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.42477114398531e-06,
+      "loss": 0.4079,
+      "step": 3858
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.419848084878793e-06,
+      "loss": 0.3802,
+      "step": 3859
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4149260208440736e-06,
+      "loss": 0.457,
+      "step": 3860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.410004953249205e-06,
+      "loss": 0.3837,
+      "step": 3861
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.405084883461949e-06,
+      "loss": 0.4226,
+      "step": 3862
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.400165812849801e-06,
+      "loss": 0.4801,
+      "step": 3863
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3952477427799665e-06,
+      "loss": 0.418,
+      "step": 3864
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.390330674619382e-06,
+      "loss": 0.4302,
+      "step": 3865
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.38541460973471e-06,
+      "loss": 0.3941,
+      "step": 3866
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.38049954949232e-06,
+      "loss": 0.4346,
+      "step": 3867
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.375585495258316e-06,
+      "loss": 0.4821,
+      "step": 3868
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.370672448398512e-06,
+      "loss": 0.4196,
+      "step": 3869
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.365760410278443e-06,
+      "loss": 0.4902,
+      "step": 3870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.360849382263372e-06,
+      "loss": 0.4596,
+      "step": 3871
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.355939365718279e-06,
+      "loss": 0.42,
+      "step": 3872
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3510303620078525e-06,
+      "loss": 0.4714,
+      "step": 3873
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.346122372496514e-06,
+      "loss": 0.4548,
+      "step": 3874
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3412153985483875e-06,
+      "loss": 0.4094,
+      "step": 3875
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.336309441527333e-06,
+      "loss": 0.4234,
+      "step": 3876
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.331404502796912e-06,
+      "loss": 0.4322,
+      "step": 3877
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.326500583720412e-06,
+      "loss": 0.511,
+      "step": 3878
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.321597685660832e-06,
+      "loss": 0.4479,
+      "step": 3879
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.316695809980888e-06,
+      "loss": 0.4909,
+      "step": 3880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.311794958043021e-06,
+      "loss": 0.3987,
+      "step": 3881
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.306895131209374e-06,
+      "loss": 0.4142,
+      "step": 3882
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3019963308418084e-06,
+      "loss": 0.3873,
+      "step": 3883
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.297098558301911e-06,
+      "loss": 0.535,
+      "step": 3884
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2922018149509636e-06,
+      "loss": 0.4292,
+      "step": 3885
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.287306102149982e-06,
+      "loss": 0.4665,
+      "step": 3886
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.282411421259684e-06,
+      "loss": 0.4173,
+      "step": 3887
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.277517773640502e-06,
+      "loss": 0.4488,
+      "step": 3888
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.272625160652587e-06,
+      "loss": 0.422,
+      "step": 3889
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.267733583655789e-06,
+      "loss": 0.4673,
+      "step": 3890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.262843044009688e-06,
+      "loss": 0.4921,
+      "step": 3891
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2579535430735595e-06,
+      "loss": 0.4232,
+      "step": 3892
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.253065082206403e-06,
+      "loss": 0.4119,
+      "step": 3893
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.248177662766919e-06,
+      "loss": 0.4786,
+      "step": 3894
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.243291286113523e-06,
+      "loss": 0.4405,
+      "step": 3895
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.238405953604346e-06,
+      "loss": 0.4795,
+      "step": 3896
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.23352166659722e-06,
+      "loss": 0.4471,
+      "step": 3897
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.228638426449687e-06,
+      "loss": 0.3595,
+      "step": 3898
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.223756234519003e-06,
+      "loss": 0.3926,
+      "step": 3899
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.218875092162129e-06,
+      "loss": 0.4024,
+      "step": 3900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.213995000735738e-06,
+      "loss": 0.4667,
+      "step": 3901
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.209115961596208e-06,
+      "loss": 0.4317,
+      "step": 3902
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.204237976099625e-06,
+      "loss": 0.4116,
+      "step": 3903
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.199361045601782e-06,
+      "loss": 0.3842,
+      "step": 3904
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.194485171458176e-06,
+      "loss": 0.4473,
+      "step": 3905
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.189610355024018e-06,
+      "loss": 0.463,
+      "step": 3906
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.184736597654219e-06,
+      "loss": 0.4007,
+      "step": 3907
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.179863900703399e-06,
+      "loss": 0.4787,
+      "step": 3908
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1749922655258766e-06,
+      "loss": 0.4778,
+      "step": 3909
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.170121693475681e-06,
+      "loss": 0.4021,
+      "step": 3910
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.16525218590655e-06,
+      "loss": 0.4235,
+      "step": 3911
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.160383744171919e-06,
+      "loss": 0.4635,
+      "step": 3912
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.155516369624923e-06,
+      "loss": 0.4219,
+      "step": 3913
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.150650063618412e-06,
+      "loss": 0.4315,
+      "step": 3914
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.145784827504926e-06,
+      "loss": 0.4739,
+      "step": 3915
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.140920662636722e-06,
+      "loss": 0.4288,
+      "step": 3916
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.136057570365752e-06,
+      "loss": 0.4956,
+      "step": 3917
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.131195552043667e-06,
+      "loss": 0.3996,
+      "step": 3918
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.126334609021823e-06,
+      "loss": 0.3947,
+      "step": 3919
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.12147474265127e-06,
+      "loss": 0.4461,
+      "step": 3920
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.116615954282778e-06,
+      "loss": 0.4151,
+      "step": 3921
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.111758245266795e-06,
+      "loss": 0.4828,
+      "step": 3922
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.106901616953482e-06,
+      "loss": 0.427,
+      "step": 3923
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.102046070692696e-06,
+      "loss": 0.4858,
+      "step": 3924
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.09719160783399e-06,
+      "loss": 0.4413,
+      "step": 3925
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.092338229726625e-06,
+      "loss": 0.4562,
+      "step": 3926
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0874859377195535e-06,
+      "loss": 0.4126,
+      "step": 3927
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0826347331614255e-06,
+      "loss": 0.4022,
+      "step": 3928
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.077784617400592e-06,
+      "loss": 0.4912,
+      "step": 3929
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.072935591785096e-06,
+      "loss": 0.4359,
+      "step": 3930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068087657662687e-06,
+      "loss": 0.4486,
+      "step": 3931
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.063240816380805e-06,
+      "loss": 0.4492,
+      "step": 3932
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.058395069286588e-06,
+      "loss": 0.472,
+      "step": 3933
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0535504177268635e-06,
+      "loss": 0.4627,
+      "step": 3934
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.048706863048162e-06,
+      "loss": 0.3996,
+      "step": 3935
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0438644065967115e-06,
+      "loss": 0.3713,
+      "step": 3936
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.039023049718425e-06,
+      "loss": 0.4721,
+      "step": 3937
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.034182793758918e-06,
+      "loss": 0.4606,
+      "step": 3938
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.029343640063499e-06,
+      "loss": 0.4926,
+      "step": 3939
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.024505589977159e-06,
+      "loss": 0.3882,
+      "step": 3940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.019668644844601e-06,
+      "loss": 0.5206,
+      "step": 3941
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.014832806010211e-06,
+      "loss": 0.5035,
+      "step": 3942
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.009998074818064e-06,
+      "loss": 0.3811,
+      "step": 3943
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.005164452611934e-06,
+      "loss": 0.4456,
+      "step": 3944
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.000331940735278e-06,
+      "loss": 0.4718,
+      "step": 3945
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.995500540531257e-06,
+      "loss": 0.4454,
+      "step": 3946
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.990670253342716e-06,
+      "loss": 0.4717,
+      "step": 3947
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.985841080512189e-06,
+      "loss": 0.437,
+      "step": 3948
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.981013023381899e-06,
+      "loss": 0.4401,
+      "step": 3949
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.976186083293767e-06,
+      "loss": 0.4447,
+      "step": 3950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.971360261589398e-06,
+      "loss": 0.3908,
+      "step": 3951
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.966535559610085e-06,
+      "loss": 0.4361,
+      "step": 3952
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.961711978696813e-06,
+      "loss": 0.4687,
+      "step": 3953
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.956889520190251e-06,
+      "loss": 0.402,
+      "step": 3954
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.952068185430764e-06,
+      "loss": 0.3902,
+      "step": 3955
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.947247975758396e-06,
+      "loss": 0.425,
+      "step": 3956
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.942428892512884e-06,
+      "loss": 0.3612,
+      "step": 3957
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9376109370336466e-06,
+      "loss": 0.4245,
+      "step": 3958
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.932794110659793e-06,
+      "loss": 0.4196,
+      "step": 3959
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9279784147301234e-06,
+      "loss": 0.4531,
+      "step": 3960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.923163850583114e-06,
+      "loss": 0.432,
+      "step": 3961
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.918350419556928e-06,
+      "loss": 0.4921,
+      "step": 3962
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.913538122989422e-06,
+      "loss": 0.4056,
+      "step": 3963
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.908726962218125e-06,
+      "loss": 0.4611,
+      "step": 3964
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9039169385802605e-06,
+      "loss": 0.4643,
+      "step": 3965
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8991080534127366e-06,
+      "loss": 0.4298,
+      "step": 3966
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8943003080521335e-06,
+      "loss": 0.4832,
+      "step": 3967
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.889493703834726e-06,
+      "loss": 0.4659,
+      "step": 3968
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.884688242096464e-06,
+      "loss": 0.4191,
+      "step": 3969
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.879883924172989e-06,
+      "loss": 0.4115,
+      "step": 3970
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.875080751399614e-06,
+      "loss": 0.4315,
+      "step": 3971
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8702787251113435e-06,
+      "loss": 0.434,
+      "step": 3972
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8654778466428534e-06,
+      "loss": 0.3992,
+      "step": 3973
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8606781173285046e-06,
+      "loss": 0.4324,
+      "step": 3974
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.855879538502347e-06,
+      "loss": 0.4341,
+      "step": 3975
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.851082111498101e-06,
+      "loss": 0.4185,
+      "step": 3976
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.846285837649166e-06,
+      "loss": 0.4176,
+      "step": 3977
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.841490718288627e-06,
+      "loss": 0.4294,
+      "step": 3978
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.836696754749241e-06,
+      "loss": 0.4179,
+      "step": 3979
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.831903948363455e-06,
+      "loss": 0.4084,
+      "step": 3980
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.82711230046338e-06,
+      "loss": 0.3713,
+      "step": 3981
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.82232181238082e-06,
+      "loss": 0.4715,
+      "step": 3982
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.817532485447247e-06,
+      "loss": 0.3624,
+      "step": 3983
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.812744320993806e-06,
+      "loss": 0.3738,
+      "step": 3984
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.807957320351335e-06,
+      "loss": 0.4397,
+      "step": 3985
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.803171484850334e-06,
+      "loss": 0.4308,
+      "step": 3986
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.79838681582098e-06,
+      "loss": 0.413,
+      "step": 3987
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.793603314593137e-06,
+      "loss": 0.4254,
+      "step": 3988
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7888209824963304e-06,
+      "loss": 0.4333,
+      "step": 3989
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.784039820859775e-06,
+      "loss": 0.4366,
+      "step": 3990
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.779259831012349e-06,
+      "loss": 0.5139,
+      "step": 3991
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.774481014282608e-06,
+      "loss": 0.3957,
+      "step": 3992
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7697033719987785e-06,
+      "loss": 0.3942,
+      "step": 3993
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7649269054887695e-06,
+      "loss": 0.405,
+      "step": 3994
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.760151616080158e-06,
+      "loss": 0.4182,
+      "step": 3995
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.755377505100195e-06,
+      "loss": 0.3848,
+      "step": 3996
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7506045738758e-06,
+      "loss": 0.4394,
+      "step": 3997
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.745832823733567e-06,
+      "loss": 0.4809,
+      "step": 3998
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.741062255999759e-06,
+      "loss": 0.4537,
+      "step": 3999
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.736292872000318e-06,
+      "loss": 0.4662,
+      "step": 4000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7315246730608535e-06,
+      "loss": 0.397,
+      "step": 4001
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7267576605066435e-06,
+      "loss": 0.4754,
+      "step": 4002
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.721991835662636e-06,
+      "loss": 0.4523,
+      "step": 4003
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.717227199853447e-06,
+      "loss": 0.3873,
+      "step": 4004
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.712463754403373e-06,
+      "loss": 0.4281,
+      "step": 4005
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.707701500636363e-06,
+      "loss": 0.4314,
+      "step": 4006
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.702940439876052e-06,
+      "loss": 0.4504,
+      "step": 4007
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.698180573445732e-06,
+      "loss": 0.3813,
+      "step": 4008
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.693421902668361e-06,
+      "loss": 0.4831,
+      "step": 4009
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.688664428866579e-06,
+      "loss": 0.4345,
+      "step": 4010
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.683908153362674e-06,
+      "loss": 0.385,
+      "step": 4011
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.679153077478621e-06,
+      "loss": 0.352,
+      "step": 4012
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.674399202536046e-06,
+      "loss": 0.4411,
+      "step": 4013
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.669646529856246e-06,
+      "loss": 0.4826,
+      "step": 4014
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6648950607601875e-06,
+      "loss": 0.388,
+      "step": 4015
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.660144796568499e-06,
+      "loss": 0.4564,
+      "step": 4016
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.655395738601473e-06,
+      "loss": 0.5216,
+      "step": 4017
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.65064788817907e-06,
+      "loss": 0.4394,
+      "step": 4018
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.645901246620911e-06,
+      "loss": 0.4236,
+      "step": 4019
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.64115581524629e-06,
+      "loss": 0.4376,
+      "step": 4020
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.636411595374152e-06,
+      "loss": 0.5362,
+      "step": 4021
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.631668588323114e-06,
+      "loss": 0.4789,
+      "step": 4022
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.626926795411447e-06,
+      "loss": 0.51,
+      "step": 4023
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.622186217957095e-06,
+      "loss": 0.4497,
+      "step": 4024
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6174468572776665e-06,
+      "loss": 0.4805,
+      "step": 4025
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.612708714690418e-06,
+      "loss": 0.4842,
+      "step": 4026
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.607971791512276e-06,
+      "loss": 0.5035,
+      "step": 4027
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.603236089059827e-06,
+      "loss": 0.4171,
+      "step": 4028
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598501608649312e-06,
+      "loss": 0.3987,
+      "step": 4029
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5937683515966455e-06,
+      "loss": 0.4511,
+      "step": 4030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.589036319217397e-06,
+      "loss": 0.3694,
+      "step": 4031
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5843055128267886e-06,
+      "loss": 0.467,
+      "step": 4032
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579575933739707e-06,
+      "loss": 0.4863,
+      "step": 4033
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.574847583270695e-06,
+      "loss": 0.4427,
+      "step": 4034
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570120462733961e-06,
+      "loss": 0.4408,
+      "step": 4035
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.565394573443361e-06,
+      "loss": 0.406,
+      "step": 4036
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.560669916712421e-06,
+      "loss": 0.3793,
+      "step": 4037
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.555946493854315e-06,
+      "loss": 0.427,
+      "step": 4038
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5512243061818726e-06,
+      "loss": 0.4492,
+      "step": 4039
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.546503355007591e-06,
+      "loss": 0.437,
+      "step": 4040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.541783641643617e-06,
+      "loss": 0.4674,
+      "step": 4041
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5370651674017455e-06,
+      "loss": 0.4349,
+      "step": 4042
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.532347933593443e-06,
+      "loss": 0.4681,
+      "step": 4043
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.527631941529818e-06,
+      "loss": 0.4559,
+      "step": 4044
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.522917192521646e-06,
+      "loss": 0.4045,
+      "step": 4045
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.518203687879344e-06,
+      "loss": 0.4491,
+      "step": 4046
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.513491428912994e-06,
+      "loss": 0.408,
+      "step": 4047
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.508780416932318e-06,
+      "loss": 0.4309,
+      "step": 4048
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.504070653246708e-06,
+      "loss": 0.4771,
+      "step": 4049
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.499362139165203e-06,
+      "loss": 0.4526,
+      "step": 4050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49465487599649e-06,
+      "loss": 0.4745,
+      "step": 4051
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.489948865048913e-06,
+      "loss": 0.4501,
+      "step": 4052
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.485244107630461e-06,
+      "loss": 0.393,
+      "step": 4053
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.480540605048783e-06,
+      "loss": 0.5253,
+      "step": 4054
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.475838358611183e-06,
+      "loss": 0.4627,
+      "step": 4055
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.471137369624602e-06,
+      "loss": 0.4501,
+      "step": 4056
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.466437639395641e-06,
+      "loss": 0.4018,
+      "step": 4057
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4617391692305436e-06,
+      "loss": 0.4086,
+      "step": 4058
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.457041960435217e-06,
+      "loss": 0.4027,
+      "step": 4059
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.452346014315199e-06,
+      "loss": 0.4159,
+      "step": 4060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.447651332175699e-06,
+      "loss": 0.4664,
+      "step": 4061
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.442957915321556e-06,
+      "loss": 0.4229,
+      "step": 4062
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.438265765057259e-06,
+      "loss": 0.397,
+      "step": 4063
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.433574882686959e-06,
+      "loss": 0.4707,
+      "step": 4064
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.428885269514443e-06,
+      "loss": 0.3986,
+      "step": 4065
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.424196926843141e-06,
+      "loss": 0.4638,
+      "step": 4066
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.419509855976148e-06,
+      "loss": 0.4451,
+      "step": 4067
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.414824058216184e-06,
+      "loss": 0.4045,
+      "step": 4068
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.410139534865633e-06,
+      "loss": 0.4306,
+      "step": 4069
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.405456287226514e-06,
+      "loss": 0.416,
+      "step": 4070
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.400774316600495e-06,
+      "loss": 0.4588,
+      "step": 4071
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.396093624288885e-06,
+      "loss": 0.4148,
+      "step": 4072
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.391414211592644e-06,
+      "loss": 0.4551,
+      "step": 4073
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.386736079812378e-06,
+      "loss": 0.3911,
+      "step": 4074
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.382059230248329e-06,
+      "loss": 0.4359,
+      "step": 4075
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.377383664200386e-06,
+      "loss": 0.4631,
+      "step": 4076
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.372709382968082e-06,
+      "loss": 0.4207,
+      "step": 4077
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.368036387850588e-06,
+      "loss": 0.4515,
+      "step": 4078
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3633646801467255e-06,
+      "loss": 0.4437,
+      "step": 4079
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.35869426115496e-06,
+      "loss": 0.388,
+      "step": 4080
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.354025132173389e-06,
+      "loss": 0.3991,
+      "step": 4081
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.349357294499755e-06,
+      "loss": 0.3945,
+      "step": 4082
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3446907494314405e-06,
+      "loss": 0.4054,
+      "step": 4083
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3400254982654776e-06,
+      "loss": 0.4402,
+      "step": 4084
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.335361542298523e-06,
+      "loss": 0.4419,
+      "step": 4085
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.33069888282689e-06,
+      "loss": 0.4645,
+      "step": 4086
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.326037521146522e-06,
+      "loss": 0.4105,
+      "step": 4087
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.321377458552999e-06,
+      "loss": 0.4116,
+      "step": 4088
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.316718696341552e-06,
+      "loss": 0.4097,
+      "step": 4089
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.312061235807033e-06,
+      "loss": 0.3813,
+      "step": 4090
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.307405078243954e-06,
+      "loss": 0.4588,
+      "step": 4091
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.302750224946449e-06,
+      "loss": 0.3897,
+      "step": 4092
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298096677208286e-06,
+      "loss": 0.3861,
+      "step": 4093
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.29344443632289e-06,
+      "loss": 0.4,
+      "step": 4094
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.288793503583302e-06,
+      "loss": 0.4371,
+      "step": 4095
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.284143880282208e-06,
+      "loss": 0.4024,
+      "step": 4096
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.279495567711936e-06,
+      "loss": 0.3729,
+      "step": 4097
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.274848567164436e-06,
+      "loss": 0.4117,
+      "step": 4098
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2702028799313076e-06,
+      "loss": 0.5322,
+      "step": 4099
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.265558507303779e-06,
+      "loss": 0.4066,
+      "step": 4100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.260915450572708e-06,
+      "loss": 0.4087,
+      "step": 4101
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.25627371102859e-06,
+      "loss": 0.4646,
+      "step": 4102
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.25163328996156e-06,
+      "loss": 0.4425,
+      "step": 4103
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246994188661384e-06,
+      "loss": 0.4137,
+      "step": 4104
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.242356408417457e-06,
+      "loss": 0.4132,
+      "step": 4105
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2377199505188115e-06,
+      "loss": 0.4698,
+      "step": 4106
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2330848162541055e-06,
+      "loss": 0.3923,
+      "step": 4107
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.228451006911635e-06,
+      "loss": 0.4882,
+      "step": 4108
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.223818523779327e-06,
+      "loss": 0.4412,
+      "step": 4109
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.219187368144745e-06,
+      "loss": 0.397,
+      "step": 4110
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.214557541295073e-06,
+      "loss": 0.4184,
+      "step": 4111
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.209929044517132e-06,
+      "loss": 0.4201,
+      "step": 4112
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.205301879097369e-06,
+      "loss": 0.3862,
+      "step": 4113
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.200676046321869e-06,
+      "loss": 0.3722,
+      "step": 4114
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.196051547476335e-06,
+      "loss": 0.3988,
+      "step": 4115
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.191428383846116e-06,
+      "loss": 0.4181,
+      "step": 4116
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.186806556716172e-06,
+      "loss": 0.4175,
+      "step": 4117
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1821860673710975e-06,
+      "loss": 0.4248,
+      "step": 4118
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.177566917095125e-06,
+      "loss": 0.5263,
+      "step": 4119
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1729491071721014e-06,
+      "loss": 0.377,
+      "step": 4120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1683326388855045e-06,
+      "loss": 0.4822,
+      "step": 4121
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.163717513518447e-06,
+      "loss": 0.4279,
+      "step": 4122
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1591037323536564e-06,
+      "loss": 0.4688,
+      "step": 4123
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1544912966735e-06,
+      "loss": 0.418,
+      "step": 4124
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.149880207759958e-06,
+      "loss": 0.396,
+      "step": 4125
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.14527046689464e-06,
+      "loss": 0.4205,
+      "step": 4126
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.140662075358793e-06,
+      "loss": 0.4088,
+      "step": 4127
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1360550344332675e-06,
+      "loss": 0.4271,
+      "step": 4128
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.131449345398558e-06,
+      "loss": 0.4354,
+      "step": 4129
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.126845009534773e-06,
+      "loss": 0.3618,
+      "step": 4130
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.122242028121649e-06,
+      "loss": 0.413,
+      "step": 4131
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.117640402438536e-06,
+      "loss": 0.3933,
+      "step": 4132
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.113040133764422e-06,
+      "loss": 0.431,
+      "step": 4133
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.108441223377916e-06,
+      "loss": 0.3985,
+      "step": 4134
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.103843672557239e-06,
+      "loss": 0.4152,
+      "step": 4135
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.099247482580241e-06,
+      "loss": 0.4486,
+      "step": 4136
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.094652654724391e-06,
+      "loss": 0.4577,
+      "step": 4137
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.090059190266779e-06,
+      "loss": 0.4572,
+      "step": 4138
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.085467090484122e-06,
+      "loss": 0.406,
+      "step": 4139
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.080876356652758e-06,
+      "loss": 0.4263,
+      "step": 4140
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0762869900486345e-06,
+      "loss": 0.4674,
+      "step": 4141
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0716989919473245e-06,
+      "loss": 0.4568,
+      "step": 4142
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.06711236362403e-06,
+      "loss": 0.4218,
+      "step": 4143
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0625271063535576e-06,
+      "loss": 0.4075,
+      "step": 4144
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.057943221410338e-06,
+      "loss": 0.4019,
+      "step": 4145
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0533607100684265e-06,
+      "loss": 0.3935,
+      "step": 4146
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.048779573601488e-06,
+      "loss": 0.4933,
+      "step": 4147
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.044199813282815e-06,
+      "loss": 0.4272,
+      "step": 4148
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.039621430385306e-06,
+      "loss": 0.3927,
+      "step": 4149
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.035044426181485e-06,
+      "loss": 0.4565,
+      "step": 4150
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0304688019434865e-06,
+      "loss": 0.4563,
+      "step": 4151
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.025894558943067e-06,
+      "loss": 0.4072,
+      "step": 4152
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0213216984516035e-06,
+      "loss": 0.4576,
+      "step": 4153
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.016750221740075e-06,
+      "loss": 0.414,
+      "step": 4154
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.012180130079087e-06,
+      "loss": 0.4281,
+      "step": 4155
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.007611424738854e-06,
+      "loss": 0.4008,
+      "step": 4156
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.003044106989203e-06,
+      "loss": 0.4097,
+      "step": 4157
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.998478178099587e-06,
+      "loss": 0.4212,
+      "step": 4158
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.993913639339067e-06,
+      "loss": 0.3914,
+      "step": 4159
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.989350491976315e-06,
+      "loss": 0.4313,
+      "step": 4160
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.984788737279614e-06,
+      "loss": 0.4084,
+      "step": 4161
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9802283765168635e-06,
+      "loss": 0.3437,
+      "step": 4162
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.975669410955582e-06,
+      "loss": 0.477,
+      "step": 4163
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.971111841862887e-06,
+      "loss": 0.3573,
+      "step": 4164
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.966555670505521e-06,
+      "loss": 0.4524,
+      "step": 4165
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.962000898149828e-06,
+      "loss": 0.403,
+      "step": 4166
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.957447526061765e-06,
+      "loss": 0.452,
+      "step": 4167
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.952895555506908e-06,
+      "loss": 0.4524,
+      "step": 4168
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.948344987750431e-06,
+      "loss": 0.4204,
+      "step": 4169
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943795824057131e-06,
+      "loss": 0.4626,
+      "step": 4170
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.939248065691404e-06,
+      "loss": 0.4499,
+      "step": 4171
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.934701713917255e-06,
+      "loss": 0.4131,
+      "step": 4172
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.930156769998312e-06,
+      "loss": 0.4067,
+      "step": 4173
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.925613235197798e-06,
+      "loss": 0.4116,
+      "step": 4174
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.921071110778545e-06,
+      "loss": 0.4231,
+      "step": 4175
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.916530398003003e-06,
+      "loss": 0.4447,
+      "step": 4176
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.911991098133217e-06,
+      "loss": 0.4032,
+      "step": 4177
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.907453212430853e-06,
+      "loss": 0.4241,
+      "step": 4178
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.902916742157171e-06,
+      "loss": 0.3912,
+      "step": 4179
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.898381688573046e-06,
+      "loss": 0.4394,
+      "step": 4180
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8938480529389494e-06,
+      "loss": 0.4181,
+      "step": 4181
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.889315836514972e-06,
+      "loss": 0.3897,
+      "step": 4182
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8847850405608054e-06,
+      "loss": 0.4355,
+      "step": 4183
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.880255666335742e-06,
+      "loss": 0.3699,
+      "step": 4184
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8757277150986815e-06,
+      "loss": 0.4057,
+      "step": 4185
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.871201188108128e-06,
+      "loss": 0.4114,
+      "step": 4186
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.866676086622185e-06,
+      "loss": 0.4182,
+      "step": 4187
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.86215241189857e-06,
+      "loss": 0.403,
+      "step": 4188
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.857630165194602e-06,
+      "loss": 0.4389,
+      "step": 4189
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.853109347767196e-06,
+      "loss": 0.4299,
+      "step": 4190
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.848589960872874e-06,
+      "loss": 0.3741,
+      "step": 4191
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.844072005767755e-06,
+      "loss": 0.3945,
+      "step": 4192
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.839555483707573e-06,
+      "loss": 0.4503,
+      "step": 4193
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.83504039594765e-06,
+      "loss": 0.4403,
+      "step": 4194
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.830526743742919e-06,
+      "loss": 0.4522,
+      "step": 4195
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.826014528347909e-06,
+      "loss": 0.415,
+      "step": 4196
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.821503751016746e-06,
+      "loss": 0.4092,
+      "step": 4197
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8169944130031685e-06,
+      "loss": 0.3781,
+      "step": 4198
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.812486515560504e-06,
+      "loss": 0.3981,
+      "step": 4199
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.807980059941678e-06,
+      "loss": 0.4499,
+      "step": 4200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.80347504739923e-06,
+      "loss": 0.4109,
+      "step": 4201
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.798971479185278e-06,
+      "loss": 0.425,
+      "step": 4202
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794469356551561e-06,
+      "loss": 0.4134,
+      "step": 4203
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.789968680749397e-06,
+      "loss": 0.4106,
+      "step": 4204
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.785469453029707e-06,
+      "loss": 0.4569,
+      "step": 4205
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.78097167464302e-06,
+      "loss": 0.4701,
+      "step": 4206
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.776475346839446e-06,
+      "loss": 0.4721,
+      "step": 4207
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.771980470868707e-06,
+      "loss": 0.3996,
+      "step": 4208
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.767487047980111e-06,
+      "loss": 0.409,
+      "step": 4209
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.762995079422566e-06,
+      "loss": 0.3937,
+      "step": 4210
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.758504566444569e-06,
+      "loss": 0.4421,
+      "step": 4211
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.754015510294226e-06,
+      "loss": 0.448,
+      "step": 4212
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.74952791221923e-06,
+      "loss": 0.4725,
+      "step": 4213
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.745041773466871e-06,
+      "loss": 0.3791,
+      "step": 4214
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.740557095284029e-06,
+      "loss": 0.446,
+      "step": 4215
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.73607387891718e-06,
+      "loss": 0.3961,
+      "step": 4216
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.731592125612393e-06,
+      "loss": 0.4897,
+      "step": 4217
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.727111836615335e-06,
+      "loss": 0.4305,
+      "step": 4218
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.722633013171266e-06,
+      "loss": 0.3695,
+      "step": 4219
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7181556565250334e-06,
+      "loss": 0.4284,
+      "step": 4220
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7136797679210774e-06,
+      "loss": 0.4439,
+      "step": 4221
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.709205348603431e-06,
+      "loss": 0.4688,
+      "step": 4222
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7047323998157236e-06,
+      "loss": 0.3702,
+      "step": 4223
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.700260922801168e-06,
+      "loss": 0.4253,
+      "step": 4224
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.695790918802577e-06,
+      "loss": 0.4194,
+      "step": 4225
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.691322389062345e-06,
+      "loss": 0.4697,
+      "step": 4226
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.68685533482246e-06,
+      "loss": 0.4122,
+      "step": 4227
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6823897573245045e-06,
+      "loss": 0.433,
+      "step": 4228
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.677925657809646e-06,
+      "loss": 0.4844,
+      "step": 4229
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.673463037518634e-06,
+      "loss": 0.4592,
+      "step": 4230
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.669001897691826e-06,
+      "loss": 0.4255,
+      "step": 4231
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.664542239569147e-06,
+      "loss": 0.4123,
+      "step": 4232
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.660084064390129e-06,
+      "loss": 0.434,
+      "step": 4233
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655627373393877e-06,
+      "loss": 0.4598,
+      "step": 4234
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.651172167819091e-06,
+      "loss": 0.4268,
+      "step": 4235
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.646718448904051e-06,
+      "loss": 0.4035,
+      "step": 4236
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.642266217886634e-06,
+      "loss": 0.4653,
+      "step": 4237
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6378154760043e-06,
+      "loss": 0.3646,
+      "step": 4238
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.633366224494094e-06,
+      "loss": 0.4485,
+      "step": 4239
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.628918464592642e-06,
+      "loss": 0.4365,
+      "step": 4240
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.624472197536159e-06,
+      "loss": 0.3688,
+      "step": 4241
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.62002742456045e-06,
+      "loss": 0.3795,
+      "step": 4242
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.615584146900901e-06,
+      "loss": 0.432,
+      "step": 4243
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6111423657924815e-06,
+      "loss": 0.3753,
+      "step": 4244
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.606702082469744e-06,
+      "loss": 0.3924,
+      "step": 4245
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.602263298166824e-06,
+      "loss": 0.4,
+      "step": 4246
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.597826014117448e-06,
+      "loss": 0.367,
+      "step": 4247
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.593390231554914e-06,
+      "loss": 0.3095,
+      "step": 4248
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.588955951712116e-06,
+      "loss": 0.4355,
+      "step": 4249
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.58452317582152e-06,
+      "loss": 0.4129,
+      "step": 4250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.580091905115173e-06,
+      "loss": 0.4671,
+      "step": 4251
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.575662140824715e-06,
+      "loss": 0.4055,
+      "step": 4252
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.571233884181357e-06,
+      "loss": 0.4193,
+      "step": 4253
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5668071364158895e-06,
+      "loss": 0.3854,
+      "step": 4254
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.562381898758697e-06,
+      "loss": 0.4291,
+      "step": 4255
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.557958172439726e-06,
+      "loss": 0.3881,
+      "step": 4256
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.553535958688521e-06,
+      "loss": 0.4519,
+      "step": 4257
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.549115258734195e-06,
+      "loss": 0.3639,
+      "step": 4258
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.544696073805441e-06,
+      "loss": 0.4086,
+      "step": 4259
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5402784051305285e-06,
+      "loss": 0.3597,
+      "step": 4260
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.535862253937314e-06,
+      "loss": 0.401,
+      "step": 4261
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531447621453233e-06,
+      "loss": 0.4337,
+      "step": 4262
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.527034508905289e-06,
+      "loss": 0.4321,
+      "step": 4263
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.522622917520069e-06,
+      "loss": 0.4209,
+      "step": 4264
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.518212848523734e-06,
+      "loss": 0.4636,
+      "step": 4265
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.513804303142022e-06,
+      "loss": 0.3816,
+      "step": 4266
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.509397282600253e-06,
+      "loss": 0.3998,
+      "step": 4267
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.504991788123323e-06,
+      "loss": 0.3714,
+      "step": 4268
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.500587820935697e-06,
+      "loss": 0.4045,
+      "step": 4269
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.496185382261419e-06,
+      "loss": 0.4106,
+      "step": 4270
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4917844733241034e-06,
+      "loss": 0.4142,
+      "step": 4271
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4873850953469534e-06,
+      "loss": 0.3718,
+      "step": 4272
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4829872495527296e-06,
+      "loss": 0.4092,
+      "step": 4273
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.47859093716378e-06,
+      "loss": 0.389,
+      "step": 4274
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.474196159402021e-06,
+      "loss": 0.3711,
+      "step": 4275
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4698029174889365e-06,
+      "loss": 0.4996,
+      "step": 4276
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4654112126455976e-06,
+      "loss": 0.3732,
+      "step": 4277
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.461021046092635e-06,
+      "loss": 0.4333,
+      "step": 4278
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.456632419050255e-06,
+      "loss": 0.4285,
+      "step": 4279
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.452245332738245e-06,
+      "loss": 0.404,
+      "step": 4280
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.44785978837595e-06,
+      "loss": 0.4348,
+      "step": 4281
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4434757871822995e-06,
+      "loss": 0.4322,
+      "step": 4282
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.439093330375786e-06,
+      "loss": 0.4162,
+      "step": 4283
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4347124191744705e-06,
+      "loss": 0.4219,
+      "step": 4284
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.430333054795996e-06,
+      "loss": 0.4621,
+      "step": 4285
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.425955238457562e-06,
+      "loss": 0.402,
+      "step": 4286
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.421578971375951e-06,
+      "loss": 0.3435,
+      "step": 4287
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4172042547675035e-06,
+      "loss": 0.4396,
+      "step": 4288
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.412831089848134e-06,
+      "loss": 0.4086,
+      "step": 4289
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4084594778333225e-06,
+      "loss": 0.3832,
+      "step": 4290
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4040894199381235e-06,
+      "loss": 0.3494,
+      "step": 4291
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.399720917377159e-06,
+      "loss": 0.3642,
+      "step": 4292
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.395353971364614e-06,
+      "loss": 0.4166,
+      "step": 4293
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.390988583114241e-06,
+      "loss": 0.3914,
+      "step": 4294
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.386624753839362e-06,
+      "loss": 0.4073,
+      "step": 4295
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.382262484752863e-06,
+      "loss": 0.3895,
+      "step": 4296
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3779017770671985e-06,
+      "loss": 0.4067,
+      "step": 4297
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.373542631994396e-06,
+      "loss": 0.4158,
+      "step": 4298
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.369185050746036e-06,
+      "loss": 0.4279,
+      "step": 4299
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.364829034533271e-06,
+      "loss": 0.4103,
+      "step": 4300
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.360474584566814e-06,
+      "loss": 0.4004,
+      "step": 4301
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.356121702056951e-06,
+      "loss": 0.3929,
+      "step": 4302
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3517703882135244e-06,
+      "loss": 0.3748,
+      "step": 4303
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.347420644245946e-06,
+      "loss": 0.4397,
+      "step": 4304
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.34307247136319e-06,
+      "loss": 0.4295,
+      "step": 4305
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.338725870773786e-06,
+      "loss": 0.4349,
+      "step": 4306
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.334380843685844e-06,
+      "loss": 0.4585,
+      "step": 4307
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.330037391307018e-06,
+      "loss": 0.4035,
+      "step": 4308
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.325695514844534e-06,
+      "loss": 0.4086,
+      "step": 4309
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.321355215505181e-06,
+      "loss": 0.4775,
+      "step": 4310
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.317016494495302e-06,
+      "loss": 0.4229,
+      "step": 4311
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3126793530208135e-06,
+      "loss": 0.4113,
+      "step": 4312
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.308343792287183e-06,
+      "loss": 0.4792,
+      "step": 4313
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.30400981349944e-06,
+      "loss": 0.4708,
+      "step": 4314
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.299677417862174e-06,
+      "loss": 0.4216,
+      "step": 4315
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.295346606579539e-06,
+      "loss": 0.4442,
+      "step": 4316
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.291017380855249e-06,
+      "loss": 0.4134,
+      "step": 4317
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.286689741892573e-06,
+      "loss": 0.4071,
+      "step": 4318
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.282363690894336e-06,
+      "loss": 0.4701,
+      "step": 4319
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.278039229062927e-06,
+      "loss": 0.4208,
+      "step": 4320
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.273716357600294e-06,
+      "loss": 0.4162,
+      "step": 4321
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.269395077707944e-06,
+      "loss": 0.3911,
+      "step": 4322
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2650753905869345e-06,
+      "loss": 0.4438,
+      "step": 4323
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.260757297437888e-06,
+      "loss": 0.3597,
+      "step": 4324
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.256440799460977e-06,
+      "loss": 0.4029,
+      "step": 4325
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2521258978559324e-06,
+      "loss": 0.4502,
+      "step": 4326
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.247812593822047e-06,
+      "loss": 0.4221,
+      "step": 4327
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.243500888558168e-06,
+      "loss": 0.4219,
+      "step": 4328
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.239190783262695e-06,
+      "loss": 0.3995,
+      "step": 4329
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.234882279133581e-06,
+      "loss": 0.4291,
+      "step": 4330
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2305753773683345e-06,
+      "loss": 0.3552,
+      "step": 4331
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.226270079164029e-06,
+      "loss": 0.4336,
+      "step": 4332
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.221966385717277e-06,
+      "loss": 0.3939,
+      "step": 4333
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.217664298224258e-06,
+      "loss": 0.4507,
+      "step": 4334
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.213363817880697e-06,
+      "loss": 0.3873,
+      "step": 4335
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.209064945881871e-06,
+      "loss": 0.4526,
+      "step": 4336
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.204767683422622e-06,
+      "loss": 0.4118,
+      "step": 4337
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.200472031697331e-06,
+      "loss": 0.4085,
+      "step": 4338
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.196177991899934e-06,
+      "loss": 0.4401,
+      "step": 4339
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.191885565223925e-06,
+      "loss": 0.4163,
+      "step": 4340
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.18759475286235e-06,
+      "loss": 0.3851,
+      "step": 4341
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.183305556007798e-06,
+      "loss": 0.3816,
+      "step": 4342
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1790179758524165e-06,
+      "loss": 0.4782,
+      "step": 4343
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1747320135878985e-06,
+      "loss": 0.4191,
+      "step": 4344
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.170447670405487e-06,
+      "loss": 0.4388,
+      "step": 4345
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.16616494749598e-06,
+      "loss": 0.4301,
+      "step": 4346
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.161883846049727e-06,
+      "loss": 0.4285,
+      "step": 4347
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.157604367256619e-06,
+      "loss": 0.4377,
+      "step": 4348
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.153326512306099e-06,
+      "loss": 0.4158,
+      "step": 4349
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.149050282387157e-06,
+      "loss": 0.4878,
+      "step": 4350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.144775678688339e-06,
+      "loss": 0.39,
+      "step": 4351
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.14050270239773e-06,
+      "loss": 0.3977,
+      "step": 4352
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.136231354702969e-06,
+      "loss": 0.4049,
+      "step": 4353
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.131961636791238e-06,
+      "loss": 0.4633,
+      "step": 4354
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.127693549849263e-06,
+      "loss": 0.4004,
+      "step": 4355
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.12342709506333e-06,
+      "loss": 0.4354,
+      "step": 4356
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.119162273619255e-06,
+      "loss": 0.393,
+      "step": 4357
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.114899086702415e-06,
+      "loss": 0.4667,
+      "step": 4358
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.110637535497721e-06,
+      "loss": 0.4148,
+      "step": 4359
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.106377621189631e-06,
+      "loss": 0.4656,
+      "step": 4360
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.102119344962157e-06,
+      "loss": 0.3602,
+      "step": 4361
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.097862707998847e-06,
+      "loss": 0.5065,
+      "step": 4362
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0936077114827925e-06,
+      "loss": 0.4106,
+      "step": 4363
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0893543565966375e-06,
+      "loss": 0.4344,
+      "step": 4364
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.085102644522559e-06,
+      "loss": 0.3939,
+      "step": 4365
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.080852576442291e-06,
+      "loss": 0.3878,
+      "step": 4366
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076604153537098e-06,
+      "loss": 0.4315,
+      "step": 4367
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.072357376987793e-06,
+      "loss": 0.3745,
+      "step": 4368
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.068112247974723e-06,
+      "loss": 0.4079,
+      "step": 4369
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.063868767677793e-06,
+      "loss": 0.4044,
+      "step": 4370
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.059626937276442e-06,
+      "loss": 0.4275,
+      "step": 4371
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.055386757949646e-06,
+      "loss": 0.4039,
+      "step": 4372
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.051148230875927e-06,
+      "loss": 0.4332,
+      "step": 4373
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.046911357233343e-06,
+      "loss": 0.4505,
+      "step": 4374
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0426761381994965e-06,
+      "loss": 0.4513,
+      "step": 4375
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0384425749515295e-06,
+      "loss": 0.4331,
+      "step": 4376
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034210668666131e-06,
+      "loss": 0.4741,
+      "step": 4377
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.029980420519515e-06,
+      "loss": 0.4577,
+      "step": 4378
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.025751831687444e-06,
+      "loss": 0.3869,
+      "step": 4379
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.021524903345213e-06,
+      "loss": 0.3962,
+      "step": 4380
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.017299636667667e-06,
+      "loss": 0.4266,
+      "step": 4381
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.013076032829174e-06,
+      "loss": 0.3779,
+      "step": 4382
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.008854093003656e-06,
+      "loss": 0.4245,
+      "step": 4383
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.00463381836456e-06,
+      "loss": 0.4576,
+      "step": 4384
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.000415210084869e-06,
+      "loss": 0.4179,
+      "step": 4385
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.996198269337117e-06,
+      "loss": 0.4279,
+      "step": 4386
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.991982997293362e-06,
+      "loss": 0.3874,
+      "step": 4387
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9877693951251985e-06,
+      "loss": 0.4361,
+      "step": 4388
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.983557464003768e-06,
+      "loss": 0.4034,
+      "step": 4389
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.97934720509973e-06,
+      "loss": 0.4236,
+      "step": 4390
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.975138619583299e-06,
+      "loss": 0.3723,
+      "step": 4391
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.970931708624208e-06,
+      "loss": 0.3999,
+      "step": 4392
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.966726473391734e-06,
+      "loss": 0.439,
+      "step": 4393
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.962522915054681e-06,
+      "loss": 0.4486,
+      "step": 4394
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.958321034781393e-06,
+      "loss": 0.4275,
+      "step": 4395
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.95412083373975e-06,
+      "loss": 0.426,
+      "step": 4396
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.949922313097159e-06,
+      "loss": 0.4398,
+      "step": 4397
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.945725474020558e-06,
+      "loss": 0.412,
+      "step": 4398
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9415303176764245e-06,
+      "loss": 0.4221,
+      "step": 4399
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.937336845230762e-06,
+      "loss": 0.4823,
+      "step": 4400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.933145057849118e-06,
+      "loss": 0.3924,
+      "step": 4401
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928954956696558e-06,
+      "loss": 0.4097,
+      "step": 4402
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9247665429376825e-06,
+      "loss": 0.4749,
+      "step": 4403
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.920579817736625e-06,
+      "loss": 0.4016,
+      "step": 4404
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.916394782257046e-06,
+      "loss": 0.3679,
+      "step": 4405
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.912211437662142e-06,
+      "loss": 0.4218,
+      "step": 4406
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.908029785114642e-06,
+      "loss": 0.4411,
+      "step": 4407
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.903849825776796e-06,
+      "loss": 0.4127,
+      "step": 4408
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8996715608103855e-06,
+      "loss": 0.4206,
+      "step": 4409
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.895494991376719e-06,
+      "loss": 0.4244,
+      "step": 4410
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.891320118636646e-06,
+      "loss": 0.3925,
+      "step": 4411
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.887146943750527e-06,
+      "loss": 0.3872,
+      "step": 4412
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.882975467878267e-06,
+      "loss": 0.4362,
+      "step": 4413
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.87880569217929e-06,
+      "loss": 0.4375,
+      "step": 4414
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.87463761781254e-06,
+      "loss": 0.4194,
+      "step": 4415
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.870471245936508e-06,
+      "loss": 0.4565,
+      "step": 4416
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8663065777091955e-06,
+      "loss": 0.4356,
+      "step": 4417
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.862143614288132e-06,
+      "loss": 0.4451,
+      "step": 4418
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.857982356830385e-06,
+      "loss": 0.3926,
+      "step": 4419
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.853822806492529e-06,
+      "loss": 0.3808,
+      "step": 4420
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.849664964430684e-06,
+      "loss": 0.4171,
+      "step": 4421
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.845508831800482e-06,
+      "loss": 0.369,
+      "step": 4422
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8413544097570834e-06,
+      "loss": 0.4197,
+      "step": 4423
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.837201699455171e-06,
+      "loss": 0.3876,
+      "step": 4424
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.833050702048955e-06,
+      "loss": 0.4318,
+      "step": 4425
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.828901418692175e-06,
+      "loss": 0.4273,
+      "step": 4426
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.824753850538082e-06,
+      "loss": 0.4233,
+      "step": 4427
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.820607998739459e-06,
+      "loss": 0.3932,
+      "step": 4428
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8164638644486064e-06,
+      "loss": 0.4195,
+      "step": 4429
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8123214488173468e-06,
+      "loss": 0.4212,
+      "step": 4430
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.808180752997034e-06,
+      "loss": 0.4343,
+      "step": 4431
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8040417781385387e-06,
+      "loss": 0.3953,
+      "step": 4432
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.799904525392251e-06,
+      "loss": 0.4449,
+      "step": 4433
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7957689959080833e-06,
+      "loss": 0.478,
+      "step": 4434
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7916351908354665e-06,
+      "loss": 0.4361,
+      "step": 4435
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7875031113233583e-06,
+      "loss": 0.4404,
+      "step": 4436
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7833727585202397e-06,
+      "loss": 0.4114,
+      "step": 4437
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.779244133574099e-06,
+      "loss": 0.3951,
+      "step": 4438
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7751172376324497e-06,
+      "loss": 0.347,
+      "step": 4439
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.770992071842334e-06,
+      "loss": 0.436,
+      "step": 4440
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7668686373503006e-06,
+      "loss": 0.4295,
+      "step": 4441
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7627469353024192e-06,
+      "loss": 0.3739,
+      "step": 4442
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7586269668442865e-06,
+      "loss": 0.4165,
+      "step": 4443
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.754508733121007e-06,
+      "loss": 0.406,
+      "step": 4444
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7503922352772127e-06,
+      "loss": 0.4563,
+      "step": 4445
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7462774744570453e-06,
+      "loss": 0.4468,
+      "step": 4446
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7421644518041657e-06,
+      "loss": 0.3923,
+      "step": 4447
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.738053168461749e-06,
+      "loss": 0.3693,
+      "step": 4448
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.733943625572495e-06,
+      "loss": 0.3863,
+      "step": 4449
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.729835824278618e-06,
+      "loss": 0.4155,
+      "step": 4450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7257297657218407e-06,
+      "loss": 0.4089,
+      "step": 4451
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.721625451043408e-06,
+      "loss": 0.4339,
+      "step": 4452
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.717522881384077e-06,
+      "loss": 0.4634,
+      "step": 4453
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7134220578841162e-06,
+      "loss": 0.4888,
+      "step": 4454
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7093229816833197e-06,
+      "loss": 0.4585,
+      "step": 4455
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.705225653920992e-06,
+      "loss": 0.3882,
+      "step": 4456
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.701130075735946e-06,
+      "loss": 0.4241,
+      "step": 4457
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.69703624826651e-06,
+      "loss": 0.3862,
+      "step": 4458
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6929441726505268e-06,
+      "loss": 0.3737,
+      "step": 4459
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.688853850025358e-06,
+      "loss": 0.4138,
+      "step": 4460
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6847652815278667e-06,
+      "loss": 0.4341,
+      "step": 4461
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.680678468294441e-06,
+      "loss": 0.3671,
+      "step": 4462
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6765934114609714e-06,
+      "loss": 0.3854,
+      "step": 4463
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6725101121628605e-06,
+      "loss": 0.4211,
+      "step": 4464
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6684285715350333e-06,
+      "loss": 0.4608,
+      "step": 4465
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6643487907119125e-06,
+      "loss": 0.4174,
+      "step": 4466
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6602707708274355e-06,
+      "loss": 0.3661,
+      "step": 4467
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.656194513015058e-06,
+      "loss": 0.3579,
+      "step": 4468
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6521200184077343e-06,
+      "loss": 0.4257,
+      "step": 4469
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6480472881379415e-06,
+      "loss": 0.3905,
+      "step": 4470
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.643976323337655e-06,
+      "loss": 0.4339,
+      "step": 4471
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.639907125138359e-06,
+      "loss": 0.4291,
+      "step": 4472
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6358396946710608e-06,
+      "loss": 0.4363,
+      "step": 4473
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6317740330662586e-06,
+      "loss": 0.442,
+      "step": 4474
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.627710141453975e-06,
+      "loss": 0.4016,
+      "step": 4475
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6236480209637304e-06,
+      "loss": 0.3914,
+      "step": 4476
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6195876727245526e-06,
+      "loss": 0.4445,
+      "step": 4477
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6155290978649794e-06,
+      "loss": 0.4513,
+      "step": 4478
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6114722975130566e-06,
+      "loss": 0.4178,
+      "step": 4479
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.607417272796343e-06,
+      "loss": 0.4112,
+      "step": 4480
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.603364024841891e-06,
+      "loss": 0.4023,
+      "step": 4481
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5993125547762663e-06,
+      "loss": 0.4323,
+      "step": 4482
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5952628637255407e-06,
+      "loss": 0.4068,
+      "step": 4483
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.591214952815284e-06,
+      "loss": 0.459,
+      "step": 4484
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5871688231705825e-06,
+      "loss": 0.4814,
+      "step": 4485
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5831244759160267e-06,
+      "loss": 0.433,
+      "step": 4486
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5790819121757036e-06,
+      "loss": 0.396,
+      "step": 4487
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.575041133073207e-06,
+      "loss": 0.4363,
+      "step": 4488
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.571002139731634e-06,
+      "loss": 0.4401,
+      "step": 4489
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5669649332735934e-06,
+      "loss": 0.4246,
+      "step": 4490
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.562929514821184e-06,
+      "loss": 0.4312,
+      "step": 4491
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.558895885496023e-06,
+      "loss": 0.4311,
+      "step": 4492
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5548640464192185e-06,
+      "loss": 0.3974,
+      "step": 4493
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5508339987113817e-06,
+      "loss": 0.4419,
+      "step": 4494
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5468057434926338e-06,
+      "loss": 0.4003,
+      "step": 4495
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5427792818825926e-06,
+      "loss": 0.4058,
+      "step": 4496
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.538754615000373e-06,
+      "loss": 0.3869,
+      "step": 4497
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.534731743964601e-06,
+      "loss": 0.3742,
+      "step": 4498
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5307106698933936e-06,
+      "loss": 0.4157,
+      "step": 4499
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.526691393904379e-06,
+      "loss": 0.3776,
+      "step": 4500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.522673917114677e-06,
+      "loss": 0.4173,
+      "step": 4501
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5186582406409096e-06,
+      "loss": 0.4171,
+      "step": 4502
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5146443655991946e-06,
+      "loss": 0.3812,
+      "step": 4503
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5106322931051583e-06,
+      "loss": 0.3727,
+      "step": 4504
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5066220242739235e-06,
+      "loss": 0.456,
+      "step": 4505
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.502613560220106e-06,
+      "loss": 0.4538,
+      "step": 4506
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.498606902057823e-06,
+      "loss": 0.4305,
+      "step": 4507
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4946020509006908e-06,
+      "loss": 0.4188,
+      "step": 4508
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4905990078618166e-06,
+      "loss": 0.3865,
+      "step": 4509
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4865977740538183e-06,
+      "loss": 0.4206,
+      "step": 4510
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.482598350588805e-06,
+      "loss": 0.4833,
+      "step": 4511
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.478600738578377e-06,
+      "loss": 0.4162,
+      "step": 4512
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4746049391336376e-06,
+      "loss": 0.391,
+      "step": 4513
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.47061095336518e-06,
+      "loss": 0.4488,
+      "step": 4514
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.466618782383101e-06,
+      "loss": 0.4484,
+      "step": 4515
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.462628427296992e-06,
+      "loss": 0.3749,
+      "step": 4516
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4586398892159345e-06,
+      "loss": 0.407,
+      "step": 4517
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4546531692485084e-06,
+      "loss": 0.416,
+      "step": 4518
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4506682685027826e-06,
+      "loss": 0.4203,
+      "step": 4519
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4466851880863326e-06,
+      "loss": 0.4267,
+      "step": 4520
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4427039291062135e-06,
+      "loss": 0.3903,
+      "step": 4521
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.43872449266899e-06,
+      "loss": 0.4302,
+      "step": 4522
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.434746879880706e-06,
+      "loss": 0.413,
+      "step": 4523
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4307710918469018e-06,
+      "loss": 0.4161,
+      "step": 4524
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4267971296726187e-06,
+      "loss": 0.4513,
+      "step": 4525
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4228249944623816e-06,
+      "loss": 0.4082,
+      "step": 4526
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4188546873202077e-06,
+      "loss": 0.4105,
+      "step": 4527
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.414886209349615e-06,
+      "loss": 0.3714,
+      "step": 4528
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.410919561653602e-06,
+      "loss": 0.4376,
+      "step": 4529
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4069547453346676e-06,
+      "loss": 0.4226,
+      "step": 4530
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.402991761494796e-06,
+      "loss": 0.3547,
+      "step": 4531
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3990306112354643e-06,
+      "loss": 0.417,
+      "step": 4532
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.395071295657636e-06,
+      "loss": 0.4004,
+      "step": 4533
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3911138158617695e-06,
+      "loss": 0.4322,
+      "step": 4534
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3871581729478174e-06,
+      "loss": 0.4631,
+      "step": 4535
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3832043680152117e-06,
+      "loss": 0.4096,
+      "step": 4536
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.379252402162877e-06,
+      "loss": 0.3991,
+      "step": 4537
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.375302276489225e-06,
+      "loss": 0.3943,
+      "step": 4538
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3713539920921644e-06,
+      "loss": 0.3868,
+      "step": 4539
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3674075500690804e-06,
+      "loss": 0.409,
+      "step": 4540
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3634629515168583e-06,
+      "loss": 0.4612,
+      "step": 4541
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3595201975318614e-06,
+      "loss": 0.3853,
+      "step": 4542
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3555792892099403e-06,
+      "loss": 0.388,
+      "step": 4543
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.351640227646442e-06,
+      "loss": 0.4045,
+      "step": 4544
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.347703013936191e-06,
+      "loss": 0.4195,
+      "step": 4545
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.343767649173498e-06,
+      "loss": 0.4127,
+      "step": 4546
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.33983413445217e-06,
+      "loss": 0.4351,
+      "step": 4547
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.335902470865485e-06,
+      "loss": 0.4941,
+      "step": 4548
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3319726595062242e-06,
+      "loss": 0.3648,
+      "step": 4549
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3280447014666372e-06,
+      "loss": 0.4745,
+      "step": 4550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3241185978384636e-06,
+      "loss": 0.341,
+      "step": 4551
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3201943497129376e-06,
+      "loss": 0.4187,
+      "step": 4552
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3162719581807603e-06,
+      "loss": 0.4212,
+      "step": 4553
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3123514243321333e-06,
+      "loss": 0.3759,
+      "step": 4554
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3084327492567324e-06,
+      "loss": 0.4394,
+      "step": 4555
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.304515934043718e-06,
+      "loss": 0.4381,
+      "step": 4556
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.300600979781732e-06,
+      "loss": 0.3899,
+      "step": 4557
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2966878875589036e-06,
+      "loss": 0.3866,
+      "step": 4558
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2927766584628475e-06,
+      "loss": 0.411,
+      "step": 4559
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.288867293580651e-06,
+      "loss": 0.4137,
+      "step": 4560
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2849597939988874e-06,
+      "loss": 0.4361,
+      "step": 4561
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2810541608036127e-06,
+      "loss": 0.4009,
+      "step": 4562
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2771503950803596e-06,
+      "loss": 0.3789,
+      "step": 4563
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.27324849791415e-06,
+      "loss": 0.4009,
+      "step": 4564
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2693484703894827e-06,
+      "loss": 0.418,
+      "step": 4565
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2654503135903358e-06,
+      "loss": 0.4275,
+      "step": 4566
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2615540286001655e-06,
+      "loss": 0.4786,
+      "step": 4567
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.257659616501909e-06,
+      "loss": 0.4305,
+      "step": 4568
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2537670783779883e-06,
+      "loss": 0.4555,
+      "step": 4569
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2498764153102945e-06,
+      "loss": 0.3775,
+      "step": 4570
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2459876283802106e-06,
+      "loss": 0.4179,
+      "step": 4571
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.242100718668586e-06,
+      "loss": 0.3553,
+      "step": 4572
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.23821568725575e-06,
+      "loss": 0.4624,
+      "step": 4573
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2343325352215216e-06,
+      "loss": 0.4536,
+      "step": 4574
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2304512636451835e-06,
+      "loss": 0.3806,
+      "step": 4575
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2265718736054974e-06,
+      "loss": 0.4044,
+      "step": 4576
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2226943661807142e-06,
+      "loss": 0.4609,
+      "step": 4577
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2188187424485463e-06,
+      "loss": 0.4606,
+      "step": 4578
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.214945003486194e-06,
+      "loss": 0.4378,
+      "step": 4579
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2110731503703273e-06,
+      "loss": 0.3566,
+      "step": 4580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2072031841770935e-06,
+      "loss": 0.4047,
+      "step": 4581
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2033351059821115e-06,
+      "loss": 0.4362,
+      "step": 4582
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1994689168604855e-06,
+      "loss": 0.4928,
+      "step": 4583
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1956046178867894e-06,
+      "loss": 0.4062,
+      "step": 4584
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.191742210135068e-06,
+      "loss": 0.3596,
+      "step": 4585
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.187881694678846e-06,
+      "loss": 0.4408,
+      "step": 4586
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.184023072591115e-06,
+      "loss": 0.3788,
+      "step": 4587
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.180166344944351e-06,
+      "loss": 0.3807,
+      "step": 4588
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1763115128104926e-06,
+      "loss": 0.4325,
+      "step": 4589
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1724585772609616e-06,
+      "loss": 0.3877,
+      "step": 4590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.168607539366645e-06,
+      "loss": 0.4005,
+      "step": 4591
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.164758400197906e-06,
+      "loss": 0.4167,
+      "step": 4592
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1609111608245734e-06,
+      "loss": 0.3753,
+      "step": 4593
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.15706582231596e-06,
+      "loss": 0.4125,
+      "step": 4594
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1532223857408437e-06,
+      "loss": 0.4546,
+      "step": 4595
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.149380852167474e-06,
+      "loss": 0.3829,
+      "step": 4596
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1455412226635694e-06,
+      "loss": 0.4775,
+      "step": 4597
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.141703498296318e-06,
+      "loss": 0.3764,
+      "step": 4598
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1378676801323893e-06,
+      "loss": 0.4389,
+      "step": 4599
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1340337692379074e-06,
+      "loss": 0.3794,
+      "step": 4600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1302017666784822e-06,
+      "loss": 0.4148,
+      "step": 4601
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1263716735191794e-06,
+      "loss": 0.3986,
+      "step": 4602
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.12254349082454e-06,
+      "loss": 0.4118,
+      "step": 4603
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.118717219658578e-06,
+      "loss": 0.4245,
+      "step": 4604
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1148928610847696e-06,
+      "loss": 0.4469,
+      "step": 4605
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1110704161660566e-06,
+      "loss": 0.4274,
+      "step": 4606
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1072498859648625e-06,
+      "loss": 0.4843,
+      "step": 4607
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1034312715430636e-06,
+      "loss": 0.3836,
+      "step": 4608
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0996145739620166e-06,
+      "loss": 0.3887,
+      "step": 4609
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0957997942825337e-06,
+      "loss": 0.4411,
+      "step": 4610
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.091986933564902e-06,
+      "loss": 0.3728,
+      "step": 4611
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.088175992868869e-06,
+      "loss": 0.3867,
+      "step": 4612
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.084366973253655e-06,
+      "loss": 0.3833,
+      "step": 4613
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0805598757779464e-06,
+      "loss": 0.4849,
+      "step": 4614
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.076754701499891e-06,
+      "loss": 0.3976,
+      "step": 4615
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.072951451477101e-06,
+      "loss": 0.413,
+      "step": 4616
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.069150126766658e-06,
+      "loss": 0.481,
+      "step": 4617
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0653507284251037e-06,
+      "loss": 0.3569,
+      "step": 4618
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0615532575084504e-06,
+      "loss": 0.4681,
+      "step": 4619
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.057757715072175e-06,
+      "loss": 0.4372,
+      "step": 4620
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.053964102171212e-06,
+      "loss": 0.4025,
+      "step": 4621
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0501724198599625e-06,
+      "loss": 0.4072,
+      "step": 4622
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0463826691922882e-06,
+      "loss": 0.4365,
+      "step": 4623
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.042594851221524e-06,
+      "loss": 0.4365,
+      "step": 4624
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.038808967000454e-06,
+      "loss": 0.4168,
+      "step": 4625
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.035025017581338e-06,
+      "loss": 0.4524,
+      "step": 4626
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.031243004015888e-06,
+      "loss": 0.4116,
+      "step": 4627
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.027462927355278e-06,
+      "loss": 0.4006,
+      "step": 4628
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.023684788650154e-06,
+      "loss": 0.4648,
+      "step": 4629
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0199085889506106e-06,
+      "loss": 0.3844,
+      "step": 4630
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0161343293062153e-06,
+      "loss": 0.3723,
+      "step": 4631
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0123620107659835e-06,
+      "loss": 0.4682,
+      "step": 4632
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.008591634378405e-06,
+      "loss": 0.3602,
+      "step": 4633
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0048232011914203e-06,
+      "loss": 0.3945,
+      "step": 4634
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0010567122524303e-06,
+      "loss": 0.4166,
+      "step": 4635
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9972921686082968e-06,
+      "loss": 0.4032,
+      "step": 4636
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9935295713053427e-06,
+      "loss": 0.4835,
+      "step": 4637
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.989768921389353e-06,
+      "loss": 0.4249,
+      "step": 4638
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.986010219905564e-06,
+      "loss": 0.4551,
+      "step": 4639
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.982253467898675e-06,
+      "loss": 0.3431,
+      "step": 4640
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.97849866641284e-06,
+      "loss": 0.4258,
+      "step": 4641
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.974745816491672e-06,
+      "loss": 0.3947,
+      "step": 4642
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.970994919178245e-06,
+      "loss": 0.3619,
+      "step": 4643
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.967245975515092e-06,
+      "loss": 0.4301,
+      "step": 4644
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9634989865441933e-06,
+      "loss": 0.3975,
+      "step": 4645
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.959753953306994e-06,
+      "loss": 0.3779,
+      "step": 4646
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9560108768443896e-06,
+      "loss": 0.4081,
+      "step": 4647
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.952269758196742e-06,
+      "loss": 0.4031,
+      "step": 4648
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9485305984038548e-06,
+      "loss": 0.4253,
+      "step": 4649
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9447933985050016e-06,
+      "loss": 0.4245,
+      "step": 4650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.941058159538901e-06,
+      "loss": 0.4195,
+      "step": 4651
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.937324882543726e-06,
+      "loss": 0.4393,
+      "step": 4652
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.933593568557117e-06,
+      "loss": 0.4212,
+      "step": 4653
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.929864218616154e-06,
+      "loss": 0.4051,
+      "step": 4654
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9261368337573757e-06,
+      "loss": 0.4524,
+      "step": 4655
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9224114150167816e-06,
+      "loss": 0.4465,
+      "step": 4656
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9186879634298126e-06,
+      "loss": 0.3338,
+      "step": 4657
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.914966480031376e-06,
+      "loss": 0.4165,
+      "step": 4658
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.911246965855823e-06,
+      "loss": 0.4224,
+      "step": 4659
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9075294219369586e-06,
+      "loss": 0.4107,
+      "step": 4660
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9038138493080394e-06,
+      "loss": 0.4615,
+      "step": 4661
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.900100249001779e-06,
+      "loss": 0.335,
+      "step": 4662
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8963886220503433e-06,
+      "loss": 0.4383,
+      "step": 4663
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.892678969485342e-06,
+      "loss": 0.4019,
+      "step": 4664
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.888971292337843e-06,
+      "loss": 0.4051,
+      "step": 4665
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8852655916383574e-06,
+      "loss": 0.4111,
+      "step": 4666
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8815618684168556e-06,
+      "loss": 0.4259,
+      "step": 4667
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8778601237027594e-06,
+      "loss": 0.4236,
+      "step": 4668
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8741603585249312e-06,
+      "loss": 0.4422,
+      "step": 4669
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.87046257391169e-06,
+      "loss": 0.4476,
+      "step": 4670
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.866766770890802e-06,
+      "loss": 0.4469,
+      "step": 4671
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8630729504894793e-06,
+      "loss": 0.4182,
+      "step": 4672
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8593811137343908e-06,
+      "loss": 0.3895,
+      "step": 4673
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.855691261651653e-06,
+      "loss": 0.4494,
+      "step": 4674
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.852003395266825e-06,
+      "loss": 0.4123,
+      "step": 4675
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.848317515604916e-06,
+      "loss": 0.4481,
+      "step": 4676
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8446336236903826e-06,
+      "loss": 0.3908,
+      "step": 4677
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8409517205471348e-06,
+      "loss": 0.4216,
+      "step": 4678
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8372718071985194e-06,
+      "loss": 0.3676,
+      "step": 4679
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8335938846673426e-06,
+      "loss": 0.3613,
+      "step": 4680
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.829917953975848e-06,
+      "loss": 0.3619,
+      "step": 4681
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.826244016145725e-06,
+      "loss": 0.3816,
+      "step": 4682
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8225720721981166e-06,
+      "loss": 0.4329,
+      "step": 4683
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8189021231536073e-06,
+      "loss": 0.429,
+      "step": 4684
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.815234170032222e-06,
+      "loss": 0.4177,
+      "step": 4685
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.811568213853443e-06,
+      "loss": 0.3784,
+      "step": 4686
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8079042556361845e-06,
+      "loss": 0.3423,
+      "step": 4687
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.804242296398817e-06,
+      "loss": 0.4267,
+      "step": 4688
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.800582337159148e-06,
+      "loss": 0.4079,
+      "step": 4689
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.79692437893443e-06,
+      "loss": 0.4054,
+      "step": 4690
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.793268422741359e-06,
+      "loss": 0.4905,
+      "step": 4691
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.789614469596077e-06,
+      "loss": 0.5113,
+      "step": 4692
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.785962520514173e-06,
+      "loss": 0.4691,
+      "step": 4693
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7823125765106697e-06,
+      "loss": 0.4034,
+      "step": 4694
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7786646386000373e-06,
+      "loss": 0.4229,
+      "step": 4695
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7750187077961876e-06,
+      "loss": 0.4151,
+      "step": 4696
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.771374785112474e-06,
+      "loss": 0.4325,
+      "step": 4697
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7677328715616948e-06,
+      "loss": 0.3647,
+      "step": 4698
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7640929681560902e-06,
+      "loss": 0.4108,
+      "step": 4699
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.760455075907337e-06,
+      "loss": 0.3492,
+      "step": 4700
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7568191958265554e-06,
+      "loss": 0.4496,
+      "step": 4701
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.753185328924304e-06,
+      "loss": 0.4092,
+      "step": 4702
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7495534762105856e-06,
+      "loss": 0.4246,
+      "step": 4703
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7459236386948464e-06,
+      "loss": 0.3648,
+      "step": 4704
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7422958173859636e-06,
+      "loss": 0.4121,
+      "step": 4705
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7386700132922595e-06,
+      "loss": 0.3669,
+      "step": 4706
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7350462274214905e-06,
+      "loss": 0.3888,
+      "step": 4707
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7314244607808627e-06,
+      "loss": 0.4109,
+      "step": 4708
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.727804714377008e-06,
+      "loss": 0.4336,
+      "step": 4709
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.72418698921601e-06,
+      "loss": 0.4296,
+      "step": 4710
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7205712863033793e-06,
+      "loss": 0.3881,
+      "step": 4711
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.716957606644067e-06,
+      "loss": 0.391,
+      "step": 4712
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7133459512424685e-06,
+      "loss": 0.4113,
+      "step": 4713
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7097363211024097e-06,
+      "loss": 0.4188,
+      "step": 4714
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7061287172271523e-06,
+      "loss": 0.394,
+      "step": 4715
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7025231406194043e-06,
+      "loss": 0.4249,
+      "step": 4716
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.698919592281297e-06,
+      "loss": 0.381,
+      "step": 4717
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6953180732144125e-06,
+      "loss": 0.3896,
+      "step": 4718
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.691718584419758e-06,
+      "loss": 0.3854,
+      "step": 4719
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.688121126897779e-06,
+      "loss": 0.494,
+      "step": 4720
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.684525701648355e-06,
+      "loss": 0.4458,
+      "step": 4721
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6809323096708063e-06,
+      "loss": 0.4005,
+      "step": 4722
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6773409519638873e-06,
+      "loss": 0.4673,
+      "step": 4723
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6737516295257804e-06,
+      "loss": 0.4695,
+      "step": 4724
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.670164343354107e-06,
+      "loss": 0.4121,
+      "step": 4725
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6665790944459224e-06,
+      "loss": 0.4046,
+      "step": 4726
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.662995883797711e-06,
+      "loss": 0.3739,
+      "step": 4727
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6594147124053983e-06,
+      "loss": 0.431,
+      "step": 4728
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.655835581264343e-06,
+      "loss": 0.389,
+      "step": 4729
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6522584913693295e-06,
+      "loss": 0.4213,
+      "step": 4730
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6486834437145747e-06,
+      "loss": 0.4212,
+      "step": 4731
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6451104392937398e-06,
+      "loss": 0.3969,
+      "step": 4732
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6415394790999048e-06,
+      "loss": 0.3643,
+      "step": 4733
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.637970564125586e-06,
+      "loss": 0.4405,
+      "step": 4734
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.634403695362736e-06,
+      "loss": 0.4491,
+      "step": 4735
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6308388738027303e-06,
+      "loss": 0.4263,
+      "step": 4736
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6272761004363833e-06,
+      "loss": 0.3998,
+      "step": 4737
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.623715376253937e-06,
+      "loss": 0.3985,
+      "step": 4738
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6201567022450612e-06,
+      "loss": 0.3774,
+      "step": 4739
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6166000793988556e-06,
+      "loss": 0.374,
+      "step": 4740
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.613045508703854e-06,
+      "loss": 0.4495,
+      "step": 4741
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.609492991148024e-06,
+      "loss": 0.3981,
+      "step": 4742
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.605942527718751e-06,
+      "loss": 0.4905,
+      "step": 4743
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.602394119402858e-06,
+      "loss": 0.4154,
+      "step": 4744
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5988477671865875e-06,
+      "loss": 0.3975,
+      "step": 4745
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5953034720556224e-06,
+      "loss": 0.4221,
+      "step": 4746
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5917612349950707e-06,
+      "loss": 0.3893,
+      "step": 4747
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.588221056989463e-06,
+      "loss": 0.3484,
+      "step": 4748
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5846829390227603e-06,
+      "loss": 0.4251,
+      "step": 4749
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5811468820783524e-06,
+      "loss": 0.4501,
+      "step": 4750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5776128871390514e-06,
+      "loss": 0.4278,
+      "step": 4751
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5740809551871038e-06,
+      "loss": 0.4195,
+      "step": 4752
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5705510872041817e-06,
+      "loss": 0.4766,
+      "step": 4753
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5670232841713784e-06,
+      "loss": 0.4197,
+      "step": 4754
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.563497547069216e-06,
+      "loss": 0.3889,
+      "step": 4755
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5599738768776373e-06,
+      "loss": 0.4143,
+      "step": 4756
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5564522745760246e-06,
+      "loss": 0.4473,
+      "step": 4757
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.552932741143167e-06,
+      "loss": 0.4411,
+      "step": 4758
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.549415277557297e-06,
+      "loss": 0.4232,
+      "step": 4759
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.545899884796057e-06,
+      "loss": 0.442,
+      "step": 4760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.542386563836519e-06,
+      "loss": 0.3749,
+      "step": 4761
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5388753156551837e-06,
+      "loss": 0.3686,
+      "step": 4762
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5353661412279694e-06,
+      "loss": 0.3715,
+      "step": 4763
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.531859041530219e-06,
+      "loss": 0.4037,
+      "step": 4764
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5283540175367027e-06,
+      "loss": 0.4226,
+      "step": 4765
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5248510702216087e-06,
+      "loss": 0.3735,
+      "step": 4766
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5213502005585546e-06,
+      "loss": 0.403,
+      "step": 4767
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.517851409520573e-06,
+      "loss": 0.4009,
+      "step": 4768
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5143546980801228e-06,
+      "loss": 0.4517,
+      "step": 4769
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5108600672090812e-06,
+      "loss": 0.3859,
+      "step": 4770
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5073675178787536e-06,
+      "loss": 0.3994,
+      "step": 4771
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5038770510598655e-06,
+      "loss": 0.3721,
+      "step": 4772
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5003886677225596e-06,
+      "loss": 0.395,
+      "step": 4773
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4969023688364004e-06,
+      "loss": 0.4214,
+      "step": 4774
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.493418155370375e-06,
+      "loss": 0.4296,
+      "step": 4775
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4899360282928865e-06,
+      "loss": 0.3743,
+      "step": 4776
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4864559885717644e-06,
+      "loss": 0.422,
+      "step": 4777
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4829780371742584e-06,
+      "loss": 0.3695,
+      "step": 4778
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.479502175067032e-06,
+      "loss": 0.4038,
+      "step": 4779
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4760284032161695e-06,
+      "loss": 0.3916,
+      "step": 4780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.472556722587174e-06,
+      "loss": 0.3922,
+      "step": 4781
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4690871341449696e-06,
+      "loss": 0.4145,
+      "step": 4782
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4656196388539024e-06,
+      "loss": 0.382,
+      "step": 4783
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4621542376777274e-06,
+      "loss": 0.4447,
+      "step": 4784
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.458690931579626e-06,
+      "loss": 0.3655,
+      "step": 4785
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4552297215221864e-06,
+      "loss": 0.4379,
+      "step": 4786
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.451770608467432e-06,
+      "loss": 0.4345,
+      "step": 4787
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.448313593376783e-06,
+      "loss": 0.433,
+      "step": 4788
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.444858677211095e-06,
+      "loss": 0.433,
+      "step": 4789
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4414058609306267e-06,
+      "loss": 0.4252,
+      "step": 4790
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.437955145495057e-06,
+      "loss": 0.3465,
+      "step": 4791
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4345065318634864e-06,
+      "loss": 0.5306,
+      "step": 4792
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4310600209944244e-06,
+      "loss": 0.4795,
+      "step": 4793
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4276156138457963e-06,
+      "loss": 0.4187,
+      "step": 4794
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.42417331137495e-06,
+      "loss": 0.4077,
+      "step": 4795
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.420733114538636e-06,
+      "loss": 0.4821,
+      "step": 4796
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4172950242930347e-06,
+      "loss": 0.4286,
+      "step": 4797
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.41385904159373e-06,
+      "loss": 0.3872,
+      "step": 4798
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4104251673957225e-06,
+      "loss": 0.4463,
+      "step": 4799
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4069934026534237e-06,
+      "loss": 0.4408,
+      "step": 4800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4035637483206676e-06,
+      "loss": 0.4037,
+      "step": 4801
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.400136205350697e-06,
+      "loss": 0.3768,
+      "step": 4802
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.396710774696166e-06,
+      "loss": 0.4403,
+      "step": 4803
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3932874573091435e-06,
+      "loss": 0.4301,
+      "step": 4804
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.389866254141109e-06,
+      "loss": 0.3962,
+      "step": 4805
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3864471661429532e-06,
+      "loss": 0.3643,
+      "step": 4806
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.383030194264985e-06,
+      "loss": 0.3782,
+      "step": 4807
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.379615339456924e-06,
+      "loss": 0.4104,
+      "step": 4808
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3762026026678987e-06,
+      "loss": 0.3736,
+      "step": 4809
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3727919848464452e-06,
+      "loss": 0.4398,
+      "step": 4810
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3693834869405163e-06,
+      "loss": 0.4111,
+      "step": 4811
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.365977109897476e-06,
+      "loss": 0.3554,
+      "step": 4812
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3625728546640938e-06,
+      "loss": 0.4158,
+      "step": 4813
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3591707221865566e-06,
+      "loss": 0.424,
+      "step": 4814
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3557707134104556e-06,
+      "loss": 0.3718,
+      "step": 4815
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3523728292807903e-06,
+      "loss": 0.4453,
+      "step": 4816
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3489770707419766e-06,
+      "loss": 0.4279,
+      "step": 4817
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.345583438737834e-06,
+      "loss": 0.4775,
+      "step": 4818
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.342191934211595e-06,
+      "loss": 0.4403,
+      "step": 4819
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.338802558105897e-06,
+      "loss": 0.4185,
+      "step": 4820
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.335415311362784e-06,
+      "loss": 0.4011,
+      "step": 4821
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3320301949237177e-06,
+      "loss": 0.4015,
+      "step": 4822
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.328647209729559e-06,
+      "loss": 0.4439,
+      "step": 4823
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3252663567205747e-06,
+      "loss": 0.3784,
+      "step": 4824
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3218876368364495e-06,
+      "loss": 0.4131,
+      "step": 4825
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3185110510162646e-06,
+      "loss": 0.4261,
+      "step": 4826
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3151366001985164e-06,
+      "loss": 0.3396,
+      "step": 4827
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.311764285321102e-06,
+      "loss": 0.403,
+      "step": 4828
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3083941073213255e-06,
+      "loss": 0.4156,
+      "step": 4829
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.305026067135898e-06,
+      "loss": 0.426,
+      "step": 4830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3016601657009364e-06,
+      "loss": 0.4372,
+      "step": 4831
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2982964039519683e-06,
+      "loss": 0.4656,
+      "step": 4832
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2949347828239187e-06,
+      "loss": 0.4221,
+      "step": 4833
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.29157530325112e-06,
+      "loss": 0.4312,
+      "step": 4834
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2882179661673078e-06,
+      "loss": 0.4109,
+      "step": 4835
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2848627725056295e-06,
+      "loss": 0.419,
+      "step": 4836
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.281509723198626e-06,
+      "loss": 0.4064,
+      "step": 4837
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2781588191782554e-06,
+      "loss": 0.4177,
+      "step": 4838
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2748100613758673e-06,
+      "loss": 0.4719,
+      "step": 4839
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.271463450722218e-06,
+      "loss": 0.4103,
+      "step": 4840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.268118988147473e-06,
+      "loss": 0.4058,
+      "step": 4841
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.264776674581194e-06,
+      "loss": 0.3843,
+      "step": 4842
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.261436510952345e-06,
+      "loss": 0.498,
+      "step": 4843
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2580984981893007e-06,
+      "loss": 0.389,
+      "step": 4844
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2547626372198274e-06,
+      "loss": 0.4134,
+      "step": 4845
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.251428928971102e-06,
+      "loss": 0.4466,
+      "step": 4846
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2480973743697e-06,
+      "loss": 0.3164,
+      "step": 4847
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.244767974341594e-06,
+      "loss": 0.4251,
+      "step": 4848
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.241440729812162e-06,
+      "loss": 0.4323,
+      "step": 4849
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2381156417061833e-06,
+      "loss": 0.354,
+      "step": 4850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2347927109478407e-06,
+      "loss": 0.4359,
+      "step": 4851
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.23147193846071e-06,
+      "loss": 0.3731,
+      "step": 4852
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.228153325167771e-06,
+      "loss": 0.4254,
+      "step": 4853
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2248368719914037e-06,
+      "loss": 0.4393,
+      "step": 4854
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2215225798533837e-06,
+      "loss": 0.3718,
+      "step": 4855
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2182104496748924e-06,
+      "loss": 0.4388,
+      "step": 4856
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.214900482376511e-06,
+      "loss": 0.4207,
+      "step": 4857
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2115926788782114e-06,
+      "loss": 0.3933,
+      "step": 4858
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2082870400993693e-06,
+      "loss": 0.4233,
+      "step": 4859
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.204983566958756e-06,
+      "loss": 0.4066,
+      "step": 4860
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.201682260374545e-06,
+      "loss": 0.437,
+      "step": 4861
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.198383121264307e-06,
+      "loss": 0.3748,
+      "step": 4862
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1950861505450084e-06,
+      "loss": 0.3896,
+      "step": 4863
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.191791349133011e-06,
+      "loss": 0.4312,
+      "step": 4864
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.188498717944075e-06,
+      "loss": 0.3885,
+      "step": 4865
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.185208257893363e-06,
+      "loss": 0.4353,
+      "step": 4866
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.181919969895424e-06,
+      "loss": 0.3981,
+      "step": 4867
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1786338548642137e-06,
+      "loss": 0.4193,
+      "step": 4868
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1753499137130783e-06,
+      "loss": 0.4392,
+      "step": 4869
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1720681473547557e-06,
+      "loss": 0.4795,
+      "step": 4870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.16878855670139e-06,
+      "loss": 0.4628,
+      "step": 4871
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1655111426645116e-06,
+      "loss": 0.4216,
+      "step": 4872
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1622359061550467e-06,
+      "loss": 0.3725,
+      "step": 4873
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1589628480833234e-06,
+      "loss": 0.3957,
+      "step": 4874
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.155691969359056e-06,
+      "loss": 0.3972,
+      "step": 4875
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1524232708913593e-06,
+      "loss": 0.3915,
+      "step": 4876
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.149156753588738e-06,
+      "loss": 0.4152,
+      "step": 4877
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1458924183590923e-06,
+      "loss": 0.4001,
+      "step": 4878
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.142630266109712e-06,
+      "loss": 0.3814,
+      "step": 4879
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1393702977472863e-06,
+      "loss": 0.3939,
+      "step": 4880
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.136112514177898e-06,
+      "loss": 0.4116,
+      "step": 4881
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1328569163070168e-06,
+      "loss": 0.4103,
+      "step": 4882
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.129603505039507e-06,
+      "loss": 0.3998,
+      "step": 4883
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.126352281279627e-06,
+      "loss": 0.38,
+      "step": 4884
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1231032459310208e-06,
+      "loss": 0.384,
+      "step": 4885
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1198563998967336e-06,
+      "loss": 0.4476,
+      "step": 4886
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.116611744079201e-06,
+      "loss": 0.3985,
+      "step": 4887
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.113369279380243e-06,
+      "loss": 0.4119,
+      "step": 4888
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.110129006701075e-06,
+      "loss": 0.421,
+      "step": 4889
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1068909269422988e-06,
+      "loss": 0.4492,
+      "step": 4890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1036550410039147e-06,
+      "loss": 0.4026,
+      "step": 4891
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1004213497853055e-06,
+      "loss": 0.4541,
+      "step": 4892
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.097189854185252e-06,
+      "loss": 0.369,
+      "step": 4893
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.093960555101917e-06,
+      "loss": 0.414,
+      "step": 4894
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0907334534328527e-06,
+      "loss": 0.4213,
+      "step": 4895
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0875085500750105e-06,
+      "loss": 0.385,
+      "step": 4896
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.084285845924717e-06,
+      "loss": 0.4147,
+      "step": 4897
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0810653418777017e-06,
+      "loss": 0.4314,
+      "step": 4898
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0778470388290708e-06,
+      "loss": 0.4315,
+      "step": 4899
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.074630937673322e-06,
+      "loss": 0.4239,
+      "step": 4900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0714170393043464e-06,
+      "loss": 0.4163,
+      "step": 4901
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0682053446154184e-06,
+      "loss": 0.4119,
+      "step": 4902
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0649958544991956e-06,
+      "loss": 0.4364,
+      "step": 4903
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0617885698477337e-06,
+      "loss": 0.3832,
+      "step": 4904
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.058583491552465e-06,
+      "loss": 0.4016,
+      "step": 4905
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.055380620504216e-06,
+      "loss": 0.4331,
+      "step": 4906
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0521799575931954e-06,
+      "loss": 0.4274,
+      "step": 4907
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.048981503709e-06,
+      "loss": 0.3804,
+      "step": 4908
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.045785259740608e-06,
+      "loss": 0.3769,
+      "step": 4909
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.04259122657639e-06,
+      "loss": 0.4285,
+      "step": 4910
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.039399405104103e-06,
+      "loss": 0.3996,
+      "step": 4911
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.036209796210883e-06,
+      "loss": 0.3762,
+      "step": 4912
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0330224007832533e-06,
+      "loss": 0.4368,
+      "step": 4913
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0298372197071235e-06,
+      "loss": 0.4191,
+      "step": 4914
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0266542538677823e-06,
+      "loss": 0.4309,
+      "step": 4915
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.023473504149911e-06,
+      "loss": 0.4062,
+      "step": 4916
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.020294971437573e-06,
+      "loss": 0.4348,
+      "step": 4917
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.017118656614212e-06,
+      "loss": 0.4488,
+      "step": 4918
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.013944560562656e-06,
+      "loss": 0.3546,
+      "step": 4919
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0107726841651154e-06,
+      "loss": 0.4328,
+      "step": 4920
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0076030283031887e-06,
+      "loss": 0.407,
+      "step": 4921
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.004435593857851e-06,
+      "loss": 0.3732,
+      "step": 4922
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.001270381709467e-06,
+      "loss": 0.3767,
+      "step": 4923
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.998107392737774e-06,
+      "loss": 0.4477,
+      "step": 4924
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.994946627821903e-06,
+      "loss": 0.41,
+      "step": 4925
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9917880878403586e-06,
+      "loss": 0.403,
+      "step": 4926
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9886317736710282e-06,
+      "loss": 0.4284,
+      "step": 4927
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9854776861911783e-06,
+      "loss": 0.421,
+      "step": 4928
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.982325826277466e-06,
+      "loss": 0.4034,
+      "step": 4929
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9791761948059217e-06,
+      "loss": 0.4014,
+      "step": 4930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.976028792651957e-06,
+      "loss": 0.378,
+      "step": 4931
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.972883620690366e-06,
+      "loss": 0.4382,
+      "step": 4932
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9697406797953155e-06,
+      "loss": 0.4351,
+      "step": 4933
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9665999708403673e-06,
+      "loss": 0.4207,
+      "step": 4934
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9634614946984454e-06,
+      "loss": 0.466,
+      "step": 4935
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9603252522418705e-06,
+      "loss": 0.3987,
+      "step": 4936
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957191244342328e-06,
+      "loss": 0.4422,
+      "step": 4937
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9540594718708885e-06,
+      "loss": 0.4049,
+      "step": 4938
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9509299356979993e-06,
+      "loss": 0.386,
+      "step": 4939
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.94780263669349e-06,
+      "loss": 0.3752,
+      "step": 4940
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.944677575726567e-06,
+      "loss": 0.4008,
+      "step": 4941
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9415547536658118e-06,
+      "loss": 0.4366,
+      "step": 4942
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9384341713791856e-06,
+      "loss": 0.4244,
+      "step": 4943
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9353158297340236e-06,
+      "loss": 0.4311,
+      "step": 4944
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9321997295970486e-06,
+      "loss": 0.4128,
+      "step": 4945
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9290858718343452e-06,
+      "loss": 0.4424,
+      "step": 4946
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9259742573113905e-06,
+      "loss": 0.3515,
+      "step": 4947
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9228648868930266e-06,
+      "loss": 0.4022,
+      "step": 4948
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919757761443473e-06,
+      "loss": 0.3955,
+      "step": 4949
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.916652881826333e-06,
+      "loss": 0.3723,
+      "step": 4950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9135502489045797e-06,
+      "loss": 0.3725,
+      "step": 4951
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9104498635405587e-06,
+      "loss": 0.3943,
+      "step": 4952
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9073517265960005e-06,
+      "loss": 0.3361,
+      "step": 4953
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9042558389319986e-06,
+      "loss": 0.3991,
+      "step": 4954
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.901162201409036e-06,
+      "loss": 0.3834,
+      "step": 4955
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8980708148869565e-06,
+      "loss": 0.4393,
+      "step": 4956
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8949816802249843e-06,
+      "loss": 0.3925,
+      "step": 4957
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.891894798281716e-06,
+      "loss": 0.436,
+      "step": 4958
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.888810169915125e-06,
+      "loss": 0.4269,
+      "step": 4959
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8857277959825605e-06,
+      "loss": 0.4607,
+      "step": 4960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8826476773407364e-06,
+      "loss": 0.4015,
+      "step": 4961
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8795698148457475e-06,
+      "loss": 0.3807,
+      "step": 4962
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8764942093530558e-06,
+      "loss": 0.4544,
+      "step": 4963
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8734208617174986e-06,
+      "loss": 0.4693,
+      "step": 4964
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8703497727932873e-06,
+      "loss": 0.4535,
+      "step": 4965
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8672809434340067e-06,
+      "loss": 0.3797,
+      "step": 4966
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8642143744926078e-06,
+      "loss": 0.3739,
+      "step": 4967
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8611500668214189e-06,
+      "loss": 0.3918,
+      "step": 4968
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8580880212721319e-06,
+      "loss": 0.368,
+      "step": 4969
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.855028238695823e-06,
+      "loss": 0.39,
+      "step": 4970
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.851970719942925e-06,
+      "loss": 0.3927,
+      "step": 4971
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8489154658632546e-06,
+      "loss": 0.3903,
+      "step": 4972
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8458624773059908e-06,
+      "loss": 0.4107,
+      "step": 4973
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.84281175511968e-06,
+      "loss": 0.3746,
+      "step": 4974
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8397633001522519e-06,
+      "loss": 0.4191,
+      "step": 4975
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8367171132509897e-06,
+      "loss": 0.4797,
+      "step": 4976
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.833673195262562e-06,
+      "loss": 0.5044,
+      "step": 4977
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.830631547032996e-06,
+      "loss": 0.3891,
+      "step": 4978
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.827592169407687e-06,
+      "loss": 0.4676,
+      "step": 4979
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.824555063231409e-06,
+      "loss": 0.409,
+      "step": 4980
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.821520229348298e-06,
+      "loss": 0.3842,
+      "step": 4981
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.818487668601856e-06,
+      "loss": 0.3921,
+      "step": 4982
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8154573818349607e-06,
+      "loss": 0.4016,
+      "step": 4983
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.81242936988985e-06,
+      "loss": 0.4007,
+      "step": 4984
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8094036336081366e-06,
+      "loss": 0.4604,
+      "step": 4985
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8063801738307974e-06,
+      "loss": 0.4348,
+      "step": 4986
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8033589913981743e-06,
+      "loss": 0.452,
+      "step": 4987
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8003400871499765e-06,
+      "loss": 0.4009,
+      "step": 4988
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7973234619252855e-06,
+      "loss": 0.3819,
+      "step": 4989
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.794309116562547e-06,
+      "loss": 0.4163,
+      "step": 4990
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.791297051899571e-06,
+      "loss": 0.453,
+      "step": 4991
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.788287268773532e-06,
+      "loss": 0.4247,
+      "step": 4992
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7852797680209744e-06,
+      "loss": 0.4188,
+      "step": 4993
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7822745504778037e-06,
+      "loss": 0.3955,
+      "step": 4994
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7792716169792957e-06,
+      "loss": 0.4029,
+      "step": 4995
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7762709683600932e-06,
+      "loss": 0.3689,
+      "step": 4996
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7732726054541971e-06,
+      "loss": 0.438,
+      "step": 4997
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7702765290949764e-06,
+      "loss": 0.3613,
+      "step": 4998
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7672827401151603e-06,
+      "loss": 0.5039,
+      "step": 4999
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7642912393468524e-06,
+      "loss": 0.3539,
+      "step": 5000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7613020276215098e-06,
+      "loss": 0.3996,
+      "step": 5001
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.75831510576996e-06,
+      "loss": 0.3511,
+      "step": 5002
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.755330474622392e-06,
+      "loss": 0.4137,
+      "step": 5003
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7523481350083549e-06,
+      "loss": 0.4,
+      "step": 5004
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7493680877567664e-06,
+      "loss": 0.4392,
+      "step": 5005
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7463903336959054e-06,
+      "loss": 0.3772,
+      "step": 5006
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7434148736534073e-06,
+      "loss": 0.4272,
+      "step": 5007
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7404417084562818e-06,
+      "loss": 0.4261,
+      "step": 5008
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7374708389308881e-06,
+      "loss": 0.3608,
+      "step": 5009
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7345022659029587e-06,
+      "loss": 0.4654,
+      "step": 5010
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7315359901975793e-06,
+      "loss": 0.3871,
+      "step": 5011
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7285720126391992e-06,
+      "loss": 0.3966,
+      "step": 5012
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7256103340516329e-06,
+      "loss": 0.3541,
+      "step": 5013
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7226509552580495e-06,
+      "loss": 0.4139,
+      "step": 5014
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.719693877080988e-06,
+      "loss": 0.3948,
+      "step": 5015
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7167391003423372e-06,
+      "loss": 0.4807,
+      "step": 5016
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7137866258633539e-06,
+      "loss": 0.3709,
+      "step": 5017
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7108364544646483e-06,
+      "loss": 0.4228,
+      "step": 5018
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7078885869661987e-06,
+      "loss": 0.4233,
+      "step": 5019
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.70494302418734e-06,
+      "loss": 0.418,
+      "step": 5020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7019997669467636e-06,
+      "loss": 0.3529,
+      "step": 5021
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6990588160625221e-06,
+      "loss": 0.383,
+      "step": 5022
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6961201723520248e-06,
+      "loss": 0.4959,
+      "step": 5023
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6931838366320464e-06,
+      "loss": 0.3524,
+      "step": 5024
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6902498097187103e-06,
+      "loss": 0.3804,
+      "step": 5025
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6873180924275101e-06,
+      "loss": 0.3714,
+      "step": 5026
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6843886855732872e-06,
+      "loss": 0.3637,
+      "step": 5027
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6814615899702413e-06,
+      "loss": 0.3668,
+      "step": 5028
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6785368064319396e-06,
+      "loss": 0.4049,
+      "step": 5029
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6756143357712962e-06,
+      "loss": 0.4501,
+      "step": 5030
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6726941788005846e-06,
+      "loss": 0.4266,
+      "step": 5031
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6697763363314412e-06,
+      "loss": 0.3669,
+      "step": 5032
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6668608091748495e-06,
+      "loss": 0.444,
+      "step": 5033
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.66394759814116e-06,
+      "loss": 0.3771,
+      "step": 5034
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.661036704040072e-06,
+      "loss": 0.3732,
+      "step": 5035
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6581281276806438e-06,
+      "loss": 0.4297,
+      "step": 5036
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.655221869871284e-06,
+      "loss": 0.3946,
+      "step": 5037
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6523179314197645e-06,
+      "loss": 0.4574,
+      "step": 5038
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6494163131332131e-06,
+      "loss": 0.3597,
+      "step": 5039
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6465170158181066e-06,
+      "loss": 0.4197,
+      "step": 5040
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6436200402802772e-06,
+      "loss": 0.4248,
+      "step": 5041
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6407253873249162e-06,
+      "loss": 0.4189,
+      "step": 5042
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6378330577565637e-06,
+      "loss": 0.4415,
+      "step": 5043
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6349430523791198e-06,
+      "loss": 0.3601,
+      "step": 5044
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6320553719958386e-06,
+      "loss": 0.386,
+      "step": 5045
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6291700174093227e-06,
+      "loss": 0.3572,
+      "step": 5046
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6262869894215305e-06,
+      "loss": 0.4336,
+      "step": 5047
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6234062888337743e-06,
+      "loss": 0.4169,
+      "step": 5048
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.620527916446719e-06,
+      "loss": 0.3774,
+      "step": 5049
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6176518730603873e-06,
+      "loss": 0.439,
+      "step": 5050
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6147781594741475e-06,
+      "loss": 0.3737,
+      "step": 5051
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6119067764867225e-06,
+      "loss": 0.3979,
+      "step": 5052
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6090377248961863e-06,
+      "loss": 0.4344,
+      "step": 5053
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6061710054999713e-06,
+      "loss": 0.4473,
+      "step": 5054
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.603306619094851e-06,
+      "loss": 0.3791,
+      "step": 5055
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.600444566476963e-06,
+      "loss": 0.3993,
+      "step": 5056
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5975848484417866e-06,
+      "loss": 0.487,
+      "step": 5057
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5947274657841528e-06,
+      "loss": 0.3732,
+      "step": 5058
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5918724192982504e-06,
+      "loss": 0.4301,
+      "step": 5059
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5890197097776129e-06,
+      "loss": 0.4472,
+      "step": 5060
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5861693380151222e-06,
+      "loss": 0.4632,
+      "step": 5061
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5833213048030204e-06,
+      "loss": 0.4503,
+      "step": 5062
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5804756109328888e-06,
+      "loss": 0.4245,
+      "step": 5063
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5776322571956671e-06,
+      "loss": 0.4069,
+      "step": 5064
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5747912443816372e-06,
+      "loss": 0.4152,
+      "step": 5065
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5719525732804341e-06,
+      "loss": 0.455,
+      "step": 5066
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5691162446810403e-06,
+      "loss": 0.4863,
+      "step": 5067
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5662822593717908e-06,
+      "loss": 0.3814,
+      "step": 5068
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5634506181403675e-06,
+      "loss": 0.3663,
+      "step": 5069
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5606213217737986e-06,
+      "loss": 0.3913,
+      "step": 5070
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5577943710584643e-06,
+      "loss": 0.4134,
+      "step": 5071
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5549697667800879e-06,
+      "loss": 0.4129,
+      "step": 5072
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5521475097237416e-06,
+      "loss": 0.4016,
+      "step": 5073
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.54932760067385e-06,
+      "loss": 0.3815,
+      "step": 5074
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5465100404141853e-06,
+      "loss": 0.3814,
+      "step": 5075
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5436948297278598e-06,
+      "loss": 0.4181,
+      "step": 5076
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5408819693973364e-06,
+      "loss": 0.4014,
+      "step": 5077
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5380714602044233e-06,
+      "loss": 0.3884,
+      "step": 5078
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5352633029302821e-06,
+      "loss": 0.4321,
+      "step": 5079
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5324574983554107e-06,
+      "loss": 0.4123,
+      "step": 5080
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5296540472596611e-06,
+      "loss": 0.3765,
+      "step": 5081
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5268529504222262e-06,
+      "loss": 0.4238,
+      "step": 5082
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5240542086216447e-06,
+      "loss": 0.3819,
+      "step": 5083
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5212578226358065e-06,
+      "loss": 0.3887,
+      "step": 5084
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.518463793241941e-06,
+      "loss": 0.5022,
+      "step": 5085
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5156721212166203e-06,
+      "loss": 0.3948,
+      "step": 5086
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5128828073357704e-06,
+      "loss": 0.4283,
+      "step": 5087
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5100958523746534e-06,
+      "loss": 0.3687,
+      "step": 5088
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5073112571078807e-06,
+      "loss": 0.4426,
+      "step": 5089
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5045290223094067e-06,
+      "loss": 0.4045,
+      "step": 5090
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5017491487525248e-06,
+      "loss": 0.3964,
+      "step": 5091
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4989716372098817e-06,
+      "loss": 0.4307,
+      "step": 5092
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4961964884534586e-06,
+      "loss": 0.5016,
+      "step": 5093
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4934237032545873e-06,
+      "loss": 0.3275,
+      "step": 5094
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.490653282383937e-06,
+      "loss": 0.4257,
+      "step": 5095
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4878852266115217e-06,
+      "loss": 0.4664,
+      "step": 5096
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4851195367066973e-06,
+      "loss": 0.4217,
+      "step": 5097
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.482356213438163e-06,
+      "loss": 0.4157,
+      "step": 5098
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4795952575739648e-06,
+      "loss": 0.4528,
+      "step": 5099
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4768366698814817e-06,
+      "loss": 0.4064,
+      "step": 5100
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.474080451127441e-06,
+      "loss": 0.3913,
+      "step": 5101
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4713266020779083e-06,
+      "loss": 0.3582,
+      "step": 5102
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4685751234982903e-06,
+      "loss": 0.3902,
+      "step": 5103
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4658260161533377e-06,
+      "loss": 0.4007,
+      "step": 5104
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4630792808071426e-06,
+      "loss": 0.4014,
+      "step": 5105
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4603349182231363e-06,
+      "loss": 0.426,
+      "step": 5106
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4575929291640867e-06,
+      "loss": 0.4652,
+      "step": 5107
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4548533143921063e-06,
+      "loss": 0.4443,
+      "step": 5108
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4521160746686503e-06,
+      "loss": 0.3303,
+      "step": 5109
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4493812107545057e-06,
+      "loss": 0.4063,
+      "step": 5110
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4466487234098082e-06,
+      "loss": 0.4247,
+      "step": 5111
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4439186133940276e-06,
+      "loss": 0.4331,
+      "step": 5112
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4411908814659714e-06,
+      "loss": 0.3815,
+      "step": 5113
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4384655283837923e-06,
+      "loss": 0.3543,
+      "step": 5114
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4357425549049775e-06,
+      "loss": 0.4464,
+      "step": 5115
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4330219617863494e-06,
+      "loss": 0.4468,
+      "step": 5116
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.43030374978408e-06,
+      "loss": 0.3625,
+      "step": 5117
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4275879196536657e-06,
+      "loss": 0.444,
+      "step": 5118
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.424874472149953e-06,
+      "loss": 0.416,
+      "step": 5119
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4221634080271186e-06,
+      "loss": 0.408,
+      "step": 5120
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4194547280386794e-06,
+      "loss": 0.3488,
+      "step": 5121
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.416748432937486e-06,
+      "loss": 0.3619,
+      "step": 5122
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4140445234757326e-06,
+      "loss": 0.4431,
+      "step": 5123
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.411343000404949e-06,
+      "loss": 0.4857,
+      "step": 5124
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.408643864475997e-06,
+      "loss": 0.4246,
+      "step": 5125
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4059471164390781e-06,
+      "loss": 0.3678,
+      "step": 5126
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4032527570437283e-06,
+      "loss": 0.4354,
+      "step": 5127
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4005607870388216e-06,
+      "loss": 0.4787,
+      "step": 5128
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.397871207172572e-06,
+      "loss": 0.4176,
+      "step": 5129
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3951840181925203e-06,
+      "loss": 0.458,
+      "step": 5130
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.392499220845549e-06,
+      "loss": 0.3878,
+      "step": 5131
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3898168158778702e-06,
+      "loss": 0.3915,
+      "step": 5132
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3871368040350398e-06,
+      "loss": 0.3837,
+      "step": 5133
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3844591860619382e-06,
+      "loss": 0.4424,
+      "step": 5134
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3817839627027918e-06,
+      "loss": 0.3909,
+      "step": 5135
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3791111347011531e-06,
+      "loss": 0.4233,
+      "step": 5136
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3764407027999082e-06,
+      "loss": 0.4138,
+      "step": 5137
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3737726677412844e-06,
+      "loss": 0.3767,
+      "step": 5138
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3711070302668372e-06,
+      "loss": 0.408,
+      "step": 5139
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3684437911174553e-06,
+      "loss": 0.4093,
+      "step": 5140
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3657829510333653e-06,
+      "loss": 0.3869,
+      "step": 5141
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3631245107541214e-06,
+      "loss": 0.3742,
+      "step": 5142
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3604684710186189e-06,
+      "loss": 0.3815,
+      "step": 5143
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3578148325650764e-06,
+      "loss": 0.365,
+      "step": 5144
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3551635961310504e-06,
+      "loss": 0.4534,
+      "step": 5145
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3525147624534262e-06,
+      "loss": 0.3862,
+      "step": 5146
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.349868332268427e-06,
+      "loss": 0.36,
+      "step": 5147
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3472243063116063e-06,
+      "loss": 0.4416,
+      "step": 5148
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3445826853178468e-06,
+      "loss": 0.4188,
+      "step": 5149
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.341943470021363e-06,
+      "loss": 0.4217,
+      "step": 5150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3393066611557005e-06,
+      "loss": 0.4067,
+      "step": 5151
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3366722594537375e-06,
+      "loss": 0.3988,
+      "step": 5152
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3340402656476826e-06,
+      "loss": 0.4337,
+      "step": 5153
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3314106804690784e-06,
+      "loss": 0.4265,
+      "step": 5154
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.328783504648794e-06,
+      "loss": 0.3945,
+      "step": 5155
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3261587389170283e-06,
+      "loss": 0.4138,
+      "step": 5156
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.323536384003311e-06,
+      "loss": 0.4646,
+      "step": 5157
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3209164406365061e-06,
+      "loss": 0.383,
+      "step": 5158
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3182989095447996e-06,
+      "loss": 0.3996,
+      "step": 5159
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3156837914557153e-06,
+      "loss": 0.3952,
+      "step": 5160
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3130710870961016e-06,
+      "loss": 0.3787,
+      "step": 5161
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3104607971921325e-06,
+      "loss": 0.4681,
+      "step": 5162
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3078529224693216e-06,
+      "loss": 0.428,
+      "step": 5163
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3052474636524993e-06,
+      "loss": 0.4094,
+      "step": 5164
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.302644421465834e-06,
+      "loss": 0.3744,
+      "step": 5165
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3000437966328171e-06,
+      "loss": 0.3875,
+      "step": 5166
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2974455898762672e-06,
+      "loss": 0.428,
+      "step": 5167
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2948498019183364e-06,
+      "loss": 0.4596,
+      "step": 5168
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.292256433480501e-06,
+      "loss": 0.3964,
+      "step": 5169
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.289665485283561e-06,
+      "loss": 0.437,
+      "step": 5170
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2870769580476528e-06,
+      "loss": 0.43,
+      "step": 5171
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2844908524922295e-06,
+      "loss": 0.3565,
+      "step": 5172
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2819071693360817e-06,
+      "loss": 0.3906,
+      "step": 5173
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2793259092973198e-06,
+      "loss": 0.39,
+      "step": 5174
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2767470730933817e-06,
+      "loss": 0.3862,
+      "step": 5175
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.274170661441031e-06,
+      "loss": 0.3792,
+      "step": 5176
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2715966750563592e-06,
+      "loss": 0.4554,
+      "step": 5177
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.269025114654786e-06,
+      "loss": 0.3904,
+      "step": 5178
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2664559809510525e-06,
+      "loss": 0.386,
+      "step": 5179
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2638892746592268e-06,
+      "loss": 0.3931,
+      "step": 5180
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.261324996492701e-06,
+      "loss": 0.4152,
+      "step": 5181
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2587631471641937e-06,
+      "loss": 0.3858,
+      "step": 5182
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2562037273857496e-06,
+      "loss": 0.4051,
+      "step": 5183
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.253646737868739e-06,
+      "loss": 0.39,
+      "step": 5184
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2510921793238517e-06,
+      "loss": 0.4397,
+      "step": 5185
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.248540052461107e-06,
+      "loss": 0.4006,
+      "step": 5186
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2459903579898413e-06,
+      "loss": 0.4159,
+      "step": 5187
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.243443096618726e-06,
+      "loss": 0.4395,
+      "step": 5188
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2408982690557458e-06,
+      "loss": 0.4555,
+      "step": 5189
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.238355876008217e-06,
+      "loss": 0.4044,
+      "step": 5190
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2358159181827734e-06,
+      "loss": 0.3938,
+      "step": 5191
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2332783962853712e-06,
+      "loss": 0.455,
+      "step": 5192
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2307433110212963e-06,
+      "loss": 0.4091,
+      "step": 5193
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.228210663095153e-06,
+      "loss": 0.4031,
+      "step": 5194
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2256804532108658e-06,
+      "loss": 0.4257,
+      "step": 5195
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2231526820716877e-06,
+      "loss": 0.4293,
+      "step": 5196
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2206273503801868e-06,
+      "loss": 0.3679,
+      "step": 5197
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2181044588382618e-06,
+      "loss": 0.4017,
+      "step": 5198
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2155840081471248e-06,
+      "loss": 0.4251,
+      "step": 5199
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2130659990073146e-06,
+      "loss": 0.376,
+      "step": 5200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2105504321186857e-06,
+      "loss": 0.4228,
+      "step": 5201
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.208037308180421e-06,
+      "loss": 0.456,
+      "step": 5202
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2055266278910239e-06,
+      "loss": 0.4117,
+      "step": 5203
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.203018391948313e-06,
+      "loss": 0.4494,
+      "step": 5204
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2005126010494306e-06,
+      "loss": 0.4042,
+      "step": 5205
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1980092558908375e-06,
+      "loss": 0.4045,
+      "step": 5206
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.195508357168318e-06,
+      "loss": 0.3892,
+      "step": 5207
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1930099055769773e-06,
+      "loss": 0.4166,
+      "step": 5208
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1905139018112367e-06,
+      "loss": 0.3552,
+      "step": 5209
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1880203465648365e-06,
+      "loss": 0.4043,
+      "step": 5210
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1855292405308416e-06,
+      "loss": 0.4448,
+      "step": 5211
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1830405844016291e-06,
+      "loss": 0.438,
+      "step": 5212
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.180554378868901e-06,
+      "loss": 0.4214,
+      "step": 5213
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1780706246236794e-06,
+      "loss": 0.3955,
+      "step": 5214
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1755893223562987e-06,
+      "loss": 0.3327,
+      "step": 5215
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1731104727564135e-06,
+      "loss": 0.4341,
+      "step": 5216
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1706340765130032e-06,
+      "loss": 0.4554,
+      "step": 5217
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1681601343143578e-06,
+      "loss": 0.383,
+      "step": 5218
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1656886468480854e-06,
+      "loss": 0.3737,
+      "step": 5219
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1632196148011188e-06,
+      "loss": 0.4117,
+      "step": 5220
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1607530388597e-06,
+      "loss": 0.3217,
+      "step": 5221
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1582889197093949e-06,
+      "loss": 0.394,
+      "step": 5222
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1558272580350826e-06,
+      "loss": 0.3875,
+      "step": 5223
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1533680545209613e-06,
+      "loss": 0.4661,
+      "step": 5224
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1509113098505408e-06,
+      "loss": 0.4601,
+      "step": 5225
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1484570247066561e-06,
+      "loss": 0.4187,
+      "step": 5226
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.146005199771455e-06,
+      "loss": 0.3639,
+      "step": 5227
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1435558357263987e-06,
+      "loss": 0.3895,
+      "step": 5228
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1411089332522662e-06,
+      "loss": 0.3839,
+      "step": 5229
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1386644930291534e-06,
+      "loss": 0.4083,
+      "step": 5230
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1362225157364692e-06,
+      "loss": 0.4716,
+      "step": 5231
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1337830020529406e-06,
+      "loss": 0.4005,
+      "step": 5232
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1313459526566105e-06,
+      "loss": 0.3964,
+      "step": 5233
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1289113682248354e-06,
+      "loss": 0.4283,
+      "step": 5234
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1264792494342858e-06,
+      "loss": 0.4128,
+      "step": 5235
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1240495969609454e-06,
+      "loss": 0.4229,
+      "step": 5236
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1216224114801189e-06,
+      "loss": 0.3477,
+      "step": 5237
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1191976936664184e-06,
+      "loss": 0.4684,
+      "step": 5238
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.116775444193775e-06,
+      "loss": 0.4211,
+      "step": 5239
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1143556637354303e-06,
+      "loss": 0.4093,
+      "step": 5240
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1119383529639395e-06,
+      "loss": 0.3877,
+      "step": 5241
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1095235125511771e-06,
+      "loss": 0.423,
+      "step": 5242
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1071111431683224e-06,
+      "loss": 0.4385,
+      "step": 5243
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1047012454858763e-06,
+      "loss": 0.3553,
+      "step": 5244
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1022938201736467e-06,
+      "loss": 0.3569,
+      "step": 5245
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0998888679007547e-06,
+      "loss": 0.4335,
+      "step": 5246
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.097486389335638e-06,
+      "loss": 0.4172,
+      "step": 5247
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0950863851460447e-06,
+      "loss": 0.407,
+      "step": 5248
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0926888559990322e-06,
+      "loss": 0.4242,
+      "step": 5249
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0902938025609765e-06,
+      "loss": 0.4347,
+      "step": 5250
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0879012254975574e-06,
+      "loss": 0.4612,
+      "step": 5251
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.085511125473775e-06,
+      "loss": 0.47,
+      "step": 5252
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0831235031539343e-06,
+      "loss": 0.4054,
+      "step": 5253
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0807383592016553e-06,
+      "loss": 0.4123,
+      "step": 5254
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0783556942798634e-06,
+      "loss": 0.4167,
+      "step": 5255
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0759755090508029e-06,
+      "loss": 0.4099,
+      "step": 5256
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0735978041760275e-06,
+      "loss": 0.427,
+      "step": 5257
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.071222580316398e-06,
+      "loss": 0.4903,
+      "step": 5258
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0688498381320855e-06,
+      "loss": 0.3754,
+      "step": 5259
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0664795782825733e-06,
+      "loss": 0.4423,
+      "step": 5260
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0641118014266527e-06,
+      "loss": 0.3987,
+      "step": 5261
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0617465082224288e-06,
+      "loss": 0.4578,
+      "step": 5262
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0593836993273154e-06,
+      "loss": 0.3905,
+      "step": 5263
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.057023375398033e-06,
+      "loss": 0.4218,
+      "step": 5264
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0546655370906123e-06,
+      "loss": 0.3779,
+      "step": 5265
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0523101850603922e-06,
+      "loss": 0.3995,
+      "step": 5266
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0499573199620261e-06,
+      "loss": 0.3753,
+      "step": 5267
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0476069424494674e-06,
+      "loss": 0.4259,
+      "step": 5268
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0452590531759887e-06,
+      "loss": 0.4203,
+      "step": 5269
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.042913652794162e-06,
+      "loss": 0.3733,
+      "step": 5270
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0405707419558685e-06,
+      "loss": 0.4651,
+      "step": 5271
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0382303213123047e-06,
+      "loss": 0.4844,
+      "step": 5272
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0358923915139662e-06,
+      "loss": 0.3961,
+      "step": 5273
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0335569532106604e-06,
+      "loss": 0.4449,
+      "step": 5274
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0312240070515044e-06,
+      "loss": 0.4382,
+      "step": 5275
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0288935536849176e-06,
+      "loss": 0.3597,
+      "step": 5276
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0265655937586305e-06,
+      "loss": 0.3742,
+      "step": 5277
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.024240127919679e-06,
+      "loss": 0.4177,
+      "step": 5278
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0219171568144036e-06,
+      "loss": 0.4405,
+      "step": 5279
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0195966810884572e-06,
+      "loss": 0.3901,
+      "step": 5280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0172787013867924e-06,
+      "loss": 0.4434,
+      "step": 5281
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.014963218353675e-06,
+      "loss": 0.3788,
+      "step": 5282
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0126502326326705e-06,
+      "loss": 0.445,
+      "step": 5283
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.010339744866654e-06,
+      "loss": 0.4134,
+      "step": 5284
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0080317556978036e-06,
+      "loss": 0.3199,
+      "step": 5285
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0057262657676059e-06,
+      "loss": 0.3918,
+      "step": 5286
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0034232757168527e-06,
+      "loss": 0.4125,
+      "step": 5287
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0011227861856399e-06,
+      "loss": 0.432,
+      "step": 5288
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.98824797813367e-07,
+      "loss": 0.3685,
+      "step": 5289
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.965293112387397e-07,
+      "loss": 0.4575,
+      "step": 5290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.942363270997668e-07,
+      "loss": 0.3782,
+      "step": 5291
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.91945846033765e-07,
+      "loss": 0.4313,
+      "step": 5292
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.89657868677355e-07,
+      "loss": 0.4058,
+      "step": 5293
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.873723956664593e-07,
+      "loss": 0.4411,
+      "step": 5294
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.850894276363032e-07,
+      "loss": 0.4697,
+      "step": 5295
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.828089652214167e-07,
+      "loss": 0.4313,
+      "step": 5296
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.805310090556385e-07,
+      "loss": 0.3912,
+      "step": 5297
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.782555597721032e-07,
+      "loss": 0.3812,
+      "step": 5298
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.75982618003254e-07,
+      "loss": 0.4103,
+      "step": 5299
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.737121843808339e-07,
+      "loss": 0.4177,
+      "step": 5300
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.714442595358886e-07,
+      "loss": 0.4189,
+      "step": 5301
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.691788440987715e-07,
+      "loss": 0.4378,
+      "step": 5302
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.669159386991323e-07,
+      "loss": 0.3704,
+      "step": 5303
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.646555439659255e-07,
+      "loss": 0.4506,
+      "step": 5304
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.623976605274098e-07,
+      "loss": 0.4131,
+      "step": 5305
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.601422890111411e-07,
+      "loss": 0.4528,
+      "step": 5306
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.578894300439855e-07,
+      "loss": 0.4981,
+      "step": 5307
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.556390842521002e-07,
+      "loss": 0.3911,
+      "step": 5308
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.533912522609523e-07,
+      "loss": 0.4023,
+      "step": 5309
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.511459346953023e-07,
+      "loss": 0.471,
+      "step": 5310
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.489031321792208e-07,
+      "loss": 0.4248,
+      "step": 5311
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.46662845336076e-07,
+      "loss": 0.4032,
+      "step": 5312
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.444250747885331e-07,
+      "loss": 0.4386,
+      "step": 5313
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.421898211585623e-07,
+      "loss": 0.4066,
+      "step": 5314
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.3995708506743e-07,
+      "loss": 0.4376,
+      "step": 5315
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.377268671357098e-07,
+      "loss": 0.4041,
+      "step": 5316
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.35499167983267e-07,
+      "loss": 0.3764,
+      "step": 5317
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.332739882292752e-07,
+      "loss": 0.4205,
+      "step": 5318
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.310513284922007e-07,
+      "loss": 0.4473,
+      "step": 5319
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.288311893898116e-07,
+      "loss": 0.3972,
+      "step": 5320
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.266135715391799e-07,
+      "loss": 0.4354,
+      "step": 5321
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.243984755566693e-07,
+      "loss": 0.4046,
+      "step": 5322
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.221859020579482e-07,
+      "loss": 0.4038,
+      "step": 5323
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.199758516579827e-07,
+      "loss": 0.4058,
+      "step": 5324
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.177683249710323e-07,
+      "loss": 0.3955,
+      "step": 5325
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.155633226106664e-07,
+      "loss": 0.4299,
+      "step": 5326
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.133608451897414e-07,
+      "loss": 0.4233,
+      "step": 5327
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.111608933204163e-07,
+      "loss": 0.4499,
+      "step": 5328
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.089634676141523e-07,
+      "loss": 0.4004,
+      "step": 5329
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.067685686817008e-07,
+      "loss": 0.421,
+      "step": 5330
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.045761971331169e-07,
+      "loss": 0.4372,
+      "step": 5331
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.023863535777511e-07,
+      "loss": 0.3816,
+      "step": 5332
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.001990386242498e-07,
+      "loss": 0.3888,
+      "step": 5333
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.980142528805569e-07,
+      "loss": 0.4237,
+      "step": 5334
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.958319969539164e-07,
+      "loss": 0.4593,
+      "step": 5335
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.936522714508678e-07,
+      "loss": 0.4336,
+      "step": 5336
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.914750769772462e-07,
+      "loss": 0.3887,
+      "step": 5337
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.893004141381834e-07,
+      "loss": 0.3898,
+      "step": 5338
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.871282835381079e-07,
+      "loss": 0.4135,
+      "step": 5339
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.849586857807424e-07,
+      "loss": 0.3717,
+      "step": 5340
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.82791621469108e-07,
+      "loss": 0.4933,
+      "step": 5341
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.806270912055259e-07,
+      "loss": 0.3799,
+      "step": 5342
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.784650955916041e-07,
+      "loss": 0.3843,
+      "step": 5343
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.763056352282506e-07,
+      "loss": 0.3985,
+      "step": 5344
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.741487107156677e-07,
+      "loss": 0.3455,
+      "step": 5345
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.719943226533567e-07,
+      "loss": 0.3807,
+      "step": 5346
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.698424716401077e-07,
+      "loss": 0.4042,
+      "step": 5347
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.676931582740112e-07,
+      "loss": 0.4777,
+      "step": 5348
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.655463831524491e-07,
+      "loss": 0.3495,
+      "step": 5349
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.634021468720966e-07,
+      "loss": 0.4831,
+      "step": 5350
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.612604500289289e-07,
+      "loss": 0.4123,
+      "step": 5351
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.591212932182102e-07,
+      "loss": 0.409,
+      "step": 5352
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.569846770344991e-07,
+      "loss": 0.4454,
+      "step": 5353
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.548506020716518e-07,
+      "loss": 0.422,
+      "step": 5354
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.527190689228127e-07,
+      "loss": 0.3445,
+      "step": 5355
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.505900781804266e-07,
+      "loss": 0.368,
+      "step": 5356
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.484636304362259e-07,
+      "loss": 0.3596,
+      "step": 5357
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.463397262812356e-07,
+      "loss": 0.4621,
+      "step": 5358
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.442183663057801e-07,
+      "loss": 0.3942,
+      "step": 5359
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.420995510994701e-07,
+      "loss": 0.4504,
+      "step": 5360
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.399832812512143e-07,
+      "loss": 0.4028,
+      "step": 5361
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.378695573492101e-07,
+      "loss": 0.3833,
+      "step": 5362
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.357583799809477e-07,
+      "loss": 0.3968,
+      "step": 5363
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.336497497332097e-07,
+      "loss": 0.4334,
+      "step": 5364
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.315436671920718e-07,
+      "loss": 0.4266,
+      "step": 5365
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.294401329429047e-07,
+      "loss": 0.439,
+      "step": 5366
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.273391475703629e-07,
+      "loss": 0.3707,
+      "step": 5367
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.252407116583994e-07,
+      "loss": 0.453,
+      "step": 5368
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.231448257902552e-07,
+      "loss": 0.4212,
+      "step": 5369
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.210514905484612e-07,
+      "loss": 0.4831,
+      "step": 5370
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.18960706514843e-07,
+      "loss": 0.3763,
+      "step": 5371
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.168724742705192e-07,
+      "loss": 0.3494,
+      "step": 5372
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.14786794395892e-07,
+      "loss": 0.4223,
+      "step": 5373
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.12703667470659e-07,
+      "loss": 0.3873,
+      "step": 5374
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.106230940738036e-07,
+      "loss": 0.4453,
+      "step": 5375
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.085450747836088e-07,
+      "loss": 0.3894,
+      "step": 5376
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.06469610177636e-07,
+      "loss": 0.3774,
+      "step": 5377
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.04396700832748e-07,
+      "loss": 0.3812,
+      "step": 5378
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.023263473250886e-07,
+      "loss": 0.4506,
+      "step": 5379
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.00258550230093e-07,
+      "loss": 0.3899,
+      "step": 5380
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.981933101224915e-07,
+      "loss": 0.4324,
+      "step": 5381
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.961306275762959e-07,
+      "loss": 0.3864,
+      "step": 5382
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.94070503164811e-07,
+      "loss": 0.4404,
+      "step": 5383
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.920129374606322e-07,
+      "loss": 0.4697,
+      "step": 5384
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.899579310356398e-07,
+      "loss": 0.3534,
+      "step": 5385
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.879054844610068e-07,
+      "loss": 0.3824,
+      "step": 5386
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.858555983071914e-07,
+      "loss": 0.4073,
+      "step": 5387
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.838082731439423e-07,
+      "loss": 0.4576,
+      "step": 5388
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.817635095402932e-07,
+      "loss": 0.4459,
+      "step": 5389
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.797213080645694e-07,
+      "loss": 0.4014,
+      "step": 5390
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.776816692843858e-07,
+      "loss": 0.3984,
+      "step": 5391
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.756445937666401e-07,
+      "loss": 0.4753,
+      "step": 5392
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.736100820775183e-07,
+      "loss": 0.4628,
+      "step": 5393
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.715781347824947e-07,
+      "loss": 0.3385,
+      "step": 5394
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.695487524463341e-07,
+      "loss": 0.434,
+      "step": 5395
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.675219356330821e-07,
+      "loss": 0.4699,
+      "step": 5396
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.654976849060791e-07,
+      "loss": 0.4004,
+      "step": 5397
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.634760008279441e-07,
+      "loss": 0.443,
+      "step": 5398
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.614568839605874e-07,
+      "loss": 0.4724,
+      "step": 5399
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.594403348652046e-07,
+      "loss": 0.415,
+      "step": 5400
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.574263541022764e-07,
+      "loss": 0.3999,
+      "step": 5401
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.554149422315759e-07,
+      "loss": 0.4253,
+      "step": 5402
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.53406099812154e-07,
+      "loss": 0.3898,
+      "step": 5403
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.513998274023504e-07,
+      "loss": 0.4278,
+      "step": 5404
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.493961255597903e-07,
+      "loss": 0.4254,
+      "step": 5405
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.473949948413883e-07,
+      "loss": 0.3818,
+      "step": 5406
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.453964358033383e-07,
+      "loss": 0.4058,
+      "step": 5407
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.434004490011238e-07,
+      "loss": 0.4304,
+      "step": 5408
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.414070349895109e-07,
+      "loss": 0.4059,
+      "step": 5409
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.394161943225508e-07,
+      "loss": 0.3646,
+      "step": 5410
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.374279275535823e-07,
+      "loss": 0.437,
+      "step": 5411
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.354422352352264e-07,
+      "loss": 0.4093,
+      "step": 5412
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.334591179193851e-07,
+      "loss": 0.4246,
+      "step": 5413
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.314785761572529e-07,
+      "loss": 0.4237,
+      "step": 5414
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.29500610499303e-07,
+      "loss": 0.4156,
+      "step": 5415
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.275252214952932e-07,
+      "loss": 0.4015,
+      "step": 5416
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.255524096942656e-07,
+      "loss": 0.3997,
+      "step": 5417
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.23582175644546e-07,
+      "loss": 0.4305,
+      "step": 5418
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.216145198937407e-07,
+      "loss": 0.4035,
+      "step": 5419
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.196494429887468e-07,
+      "loss": 0.4155,
+      "step": 5420
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.176869454757385e-07,
+      "loss": 0.4081,
+      "step": 5421
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.157270279001749e-07,
+      "loss": 0.3878,
+      "step": 5422
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.137696908067981e-07,
+      "loss": 0.3733,
+      "step": 5423
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.118149347396308e-07,
+      "loss": 0.4078,
+      "step": 5424
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.098627602419839e-07,
+      "loss": 0.428,
+      "step": 5425
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.079131678564422e-07,
+      "loss": 0.4036,
+      "step": 5426
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.059661581248833e-07,
+      "loss": 0.3955,
+      "step": 5427
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.040217315884578e-07,
+      "loss": 0.4053,
+      "step": 5428
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.020798887876024e-07,
+      "loss": 0.4584,
+      "step": 5429
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.001406302620372e-07,
+      "loss": 0.4999,
+      "step": 5430
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.982039565507615e-07,
+      "loss": 0.3976,
+      "step": 5431
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.962698681920533e-07,
+      "loss": 0.4599,
+      "step": 5432
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.943383657234804e-07,
+      "loss": 0.4335,
+      "step": 5433
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.924094496818834e-07,
+      "loss": 0.4441,
+      "step": 5434
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.904831206033901e-07,
+      "loss": 0.3901,
+      "step": 5435
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.885593790234057e-07,
+      "loss": 0.4362,
+      "step": 5436
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.866382254766158e-07,
+      "loss": 0.3965,
+      "step": 5437
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.847196604969908e-07,
+      "loss": 0.3767,
+      "step": 5438
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.828036846177766e-07,
+      "loss": 0.3905,
+      "step": 5439
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.808902983715049e-07,
+      "loss": 0.344,
+      "step": 5440
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.789795022899825e-07,
+      "loss": 0.3856,
+      "step": 5441
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.770712969042981e-07,
+      "loss": 0.4177,
+      "step": 5442
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.751656827448206e-07,
+      "loss": 0.37,
+      "step": 5443
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.732626603411985e-07,
+      "loss": 0.4723,
+      "step": 5444
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.713622302223632e-07,
+      "loss": 0.4179,
+      "step": 5445
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.694643929165202e-07,
+      "loss": 0.4365,
+      "step": 5446
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.675691489511571e-07,
+      "loss": 0.4651,
+      "step": 5447
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.656764988530395e-07,
+      "loss": 0.338,
+      "step": 5448
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.637864431482122e-07,
+      "loss": 0.4486,
+      "step": 5449
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.618989823620014e-07,
+      "loss": 0.3544,
+      "step": 5450
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.600141170190111e-07,
+      "loss": 0.3931,
+      "step": 5451
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.58131847643122e-07,
+      "loss": 0.4153,
+      "step": 5452
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.562521747574934e-07,
+      "loss": 0.3584,
+      "step": 5453
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.543750988845643e-07,
+      "loss": 0.4034,
+      "step": 5454
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.525006205460526e-07,
+      "loss": 0.4045,
+      "step": 5455
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.506287402629507e-07,
+      "loss": 0.3599,
+      "step": 5456
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.487594585555335e-07,
+      "loss": 0.4526,
+      "step": 5457
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.468927759433519e-07,
+      "loss": 0.4041,
+      "step": 5458
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.450286929452309e-07,
+      "loss": 0.4285,
+      "step": 5459
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.431672100792796e-07,
+      "loss": 0.4177,
+      "step": 5460
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.413083278628796e-07,
+      "loss": 0.3429,
+      "step": 5461
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.394520468126886e-07,
+      "loss": 0.4229,
+      "step": 5462
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.375983674446473e-07,
+      "loss": 0.4359,
+      "step": 5463
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.357472902739658e-07,
+      "loss": 0.4242,
+      "step": 5464
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.338988158151383e-07,
+      "loss": 0.4232,
+      "step": 5465
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.320529445819313e-07,
+      "loss": 0.3735,
+      "step": 5466
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.302096770873867e-07,
+      "loss": 0.3668,
+      "step": 5467
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.283690138438247e-07,
+      "loss": 0.3943,
+      "step": 5468
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.265309553628418e-07,
+      "loss": 0.3949,
+      "step": 5469
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.246955021553136e-07,
+      "loss": 0.3774,
+      "step": 5470
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.228626547313843e-07,
+      "loss": 0.4209,
+      "step": 5471
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.210324136004786e-07,
+      "loss": 0.3509,
+      "step": 5472
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.192047792712951e-07,
+      "loss": 0.3634,
+      "step": 5473
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.173797522518099e-07,
+      "loss": 0.3873,
+      "step": 5474
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.155573330492748e-07,
+      "loss": 0.4001,
+      "step": 5475
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.137375221702135e-07,
+      "loss": 0.3789,
+      "step": 5476
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.11920320120426e-07,
+      "loss": 0.4469,
+      "step": 5477
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.101057274049882e-07,
+      "loss": 0.4154,
+      "step": 5478
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.082937445282488e-07,
+      "loss": 0.4718,
+      "step": 5479
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.064843719938352e-07,
+      "loss": 0.4081,
+      "step": 5480
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.046776103046459e-07,
+      "loss": 0.3681,
+      "step": 5481
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.028734599628538e-07,
+      "loss": 0.4147,
+      "step": 5482
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.010719214699068e-07,
+      "loss": 0.3833,
+      "step": 5483
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.992729953265253e-07,
+      "loss": 0.4598,
+      "step": 5484
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.974766820327083e-07,
+      "loss": 0.3725,
+      "step": 5485
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.956829820877208e-07,
+      "loss": 0.4386,
+      "step": 5486
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.938918959901108e-07,
+      "loss": 0.4048,
+      "step": 5487
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.921034242376933e-07,
+      "loss": 0.4246,
+      "step": 5488
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.903175673275552e-07,
+      "loss": 0.3989,
+      "step": 5489
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.88534325756065e-07,
+      "loss": 0.4366,
+      "step": 5490
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.867537000188561e-07,
+      "loss": 0.3925,
+      "step": 5491
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.849756906108361e-07,
+      "loss": 0.4483,
+      "step": 5492
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.832002980261908e-07,
+      "loss": 0.433,
+      "step": 5493
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.81427522758371e-07,
+      "loss": 0.3795,
+      "step": 5494
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.796573653001091e-07,
+      "loss": 0.4111,
+      "step": 5495
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.778898261434019e-07,
+      "loss": 0.3843,
+      "step": 5496
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.761249057795204e-07,
+      "loss": 0.417,
+      "step": 5497
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.743626046990092e-07,
+      "loss": 0.4926,
+      "step": 5498
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.726029233916852e-07,
+      "loss": 0.3773,
+      "step": 5499
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.708458623466373e-07,
+      "loss": 0.4079,
+      "step": 5500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.690914220522237e-07,
+      "loss": 0.4731,
+      "step": 5501
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.673396029960776e-07,
+      "loss": 0.3592,
+      "step": 5502
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.655904056650985e-07,
+      "loss": 0.338,
+      "step": 5503
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.638438305454619e-07,
+      "loss": 0.4049,
+      "step": 5504
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.620998781226127e-07,
+      "loss": 0.3798,
+      "step": 5505
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.603585488812702e-07,
+      "loss": 0.3975,
+      "step": 5506
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.58619843305418e-07,
+      "loss": 0.3807,
+      "step": 5507
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.568837618783163e-07,
+      "loss": 0.3672,
+      "step": 5508
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.551503050824902e-07,
+      "loss": 0.3932,
+      "step": 5509
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.534194733997423e-07,
+      "loss": 0.4222,
+      "step": 5510
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.516912673111407e-07,
+      "loss": 0.3934,
+      "step": 5511
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.499656872970271e-07,
+      "loss": 0.4358,
+      "step": 5512
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.482427338370078e-07,
+      "loss": 0.4119,
+      "step": 5513
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.465224074099651e-07,
+      "loss": 0.3832,
+      "step": 5514
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.448047084940489e-07,
+      "loss": 0.4406,
+      "step": 5515
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.43089637566675e-07,
+      "loss": 0.3891,
+      "step": 5516
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.413771951045377e-07,
+      "loss": 0.4401,
+      "step": 5517
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.396673815835907e-07,
+      "loss": 0.4406,
+      "step": 5518
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.379601974790649e-07,
+      "loss": 0.3749,
+      "step": 5519
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.362556432654553e-07,
+      "loss": 0.4272,
+      "step": 5520
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.345537194165284e-07,
+      "loss": 0.4421,
+      "step": 5521
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.32854426405317e-07,
+      "loss": 0.4308,
+      "step": 5522
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.311577647041277e-07,
+      "loss": 0.4249,
+      "step": 5523
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.294637347845322e-07,
+      "loss": 0.4707,
+      "step": 5524
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.277723371173704e-07,
+      "loss": 0.4134,
+      "step": 5525
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.26083572172752e-07,
+      "loss": 0.3811,
+      "step": 5526
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.243974404200547e-07,
+      "loss": 0.4155,
+      "step": 5527
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.227139423279204e-07,
+      "loss": 0.4285,
+      "step": 5528
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.210330783642669e-07,
+      "loss": 0.3954,
+      "step": 5529
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.193548489962752e-07,
+      "loss": 0.3714,
+      "step": 5530
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.17679254690393e-07,
+      "loss": 0.4425,
+      "step": 5531
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.16006295912338e-07,
+      "loss": 0.3931,
+      "step": 5532
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.143359731270925e-07,
+      "loss": 0.4301,
+      "step": 5533
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.126682867989108e-07,
+      "loss": 0.3849,
+      "step": 5534
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.110032373913087e-07,
+      "loss": 0.4113,
+      "step": 5535
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.093408253670751e-07,
+      "loss": 0.4224,
+      "step": 5536
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.076810511882624e-07,
+      "loss": 0.4245,
+      "step": 5537
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.060239153161872e-07,
+      "loss": 0.3612,
+      "step": 5538
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.043694182114389e-07,
+      "loss": 0.4103,
+      "step": 5539
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.027175603338708e-07,
+      "loss": 0.491,
+      "step": 5540
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.010683421425988e-07,
+      "loss": 0.3991,
+      "step": 5541
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.994217640960119e-07,
+      "loss": 0.3853,
+      "step": 5542
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.977778266517596e-07,
+      "loss": 0.3968,
+      "step": 5543
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.961365302667631e-07,
+      "loss": 0.4231,
+      "step": 5544
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.944978753972052e-07,
+      "loss": 0.3644,
+      "step": 5545
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.928618624985337e-07,
+      "loss": 0.3961,
+      "step": 5546
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.91228492025464e-07,
+      "loss": 0.4807,
+      "step": 5547
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.895977644319783e-07,
+      "loss": 0.4157,
+      "step": 5548
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.879696801713252e-07,
+      "loss": 0.3796,
+      "step": 5549
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.863442396960138e-07,
+      "loss": 0.36,
+      "step": 5550
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.847214434578218e-07,
+      "loss": 0.4698,
+      "step": 5551
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.831012919077894e-07,
+      "loss": 0.3575,
+      "step": 5552
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.81483785496224e-07,
+      "loss": 0.4186,
+      "step": 5553
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.798689246727006e-07,
+      "loss": 0.3829,
+      "step": 5554
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.782567098860536e-07,
+      "loss": 0.4637,
+      "step": 5555
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.766471415843832e-07,
+      "loss": 0.4196,
+      "step": 5556
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7504022021505505e-07,
+      "loss": 0.44,
+      "step": 5557
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7343594622469735e-07,
+      "loss": 0.4395,
+      "step": 5558
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.718343200592068e-07,
+      "loss": 0.4327,
+      "step": 5559
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.702353421637407e-07,
+      "loss": 0.4246,
+      "step": 5560
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6863901298272006e-07,
+      "loss": 0.4025,
+      "step": 5561
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.670453329598312e-07,
+      "loss": 0.3961,
+      "step": 5562
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6545430253802206e-07,
+      "loss": 0.4605,
+      "step": 5563
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6386592215950765e-07,
+      "loss": 0.42,
+      "step": 5564
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.622801922657627e-07,
+      "loss": 0.4787,
+      "step": 5565
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.60697113297528e-07,
+      "loss": 0.4067,
+      "step": 5566
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5911668569480703e-07,
+      "loss": 0.4285,
+      "step": 5567
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.57538909896863e-07,
+      "loss": 0.402,
+      "step": 5568
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.559637863422284e-07,
+      "loss": 0.3925,
+      "step": 5569
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5439131546869296e-07,
+      "loss": 0.4156,
+      "step": 5570
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.528214977133094e-07,
+      "loss": 0.3641,
+      "step": 5571
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.512543335123998e-07,
+      "loss": 0.4501,
+      "step": 5572
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.496898233015379e-07,
+      "loss": 0.3964,
+      "step": 5573
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.481279675155703e-07,
+      "loss": 0.4225,
+      "step": 5574
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4656876658859984e-07,
+      "loss": 0.3758,
+      "step": 5575
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4501222095399086e-07,
+      "loss": 0.4147,
+      "step": 5576
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.43458331044373e-07,
+      "loss": 0.4601,
+      "step": 5577
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4190709729163505e-07,
+      "loss": 0.3352,
+      "step": 5578
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4035852012693227e-07,
+      "loss": 0.3936,
+      "step": 5579
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3881259998067474e-07,
+      "loss": 0.3694,
+      "step": 5580
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.372693372825387e-07,
+      "loss": 0.4028,
+      "step": 5581
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3572873246146006e-07,
+      "loss": 0.3648,
+      "step": 5582
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3419078594563514e-07,
+      "loss": 0.4179,
+      "step": 5583
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3265549816252417e-07,
+      "loss": 0.4,
+      "step": 5584
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.311228695388481e-07,
+      "loss": 0.4049,
+      "step": 5585
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.295929005005861e-07,
+      "loss": 0.4448,
+      "step": 5586
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2806559147298033e-07,
+      "loss": 0.3851,
+      "step": 5587
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.265409428805312e-07,
+      "loss": 0.4696,
+      "step": 5588
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.250189551470019e-07,
+      "loss": 0.4119,
+      "step": 5589
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2349962869541847e-07,
+      "loss": 0.3474,
+      "step": 5590
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2198296394806103e-07,
+      "loss": 0.3803,
+      "step": 5591
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.204689613264756e-07,
+      "loss": 0.4534,
+      "step": 5592
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.189576212514623e-07,
+      "loss": 0.4,
+      "step": 5593
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1744894414308954e-07,
+      "loss": 0.3547,
+      "step": 5594
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.159429304206775e-07,
+      "loss": 0.4538,
+      "step": 5595
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.144395805028112e-07,
+      "loss": 0.4077,
+      "step": 5596
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.129388948073332e-07,
+      "loss": 0.451,
+      "step": 5597
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1144087375134424e-07,
+      "loss": 0.3926,
+      "step": 5598
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0994551775120793e-07,
+      "loss": 0.3907,
+      "step": 5599
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0845282722254633e-07,
+      "loss": 0.3869,
+      "step": 5600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0696280258023635e-07,
+      "loss": 0.413,
+      "step": 5601
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.054754442384201e-07,
+      "loss": 0.3482,
+      "step": 5602
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.039907526104947e-07,
+      "loss": 0.3765,
+      "step": 5603
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0250872810911776e-07,
+      "loss": 0.4121,
+      "step": 5604
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.010293711462043e-07,
+      "loss": 0.4695,
+      "step": 5605
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9955268213292985e-07,
+      "loss": 0.4464,
+      "step": 5606
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.980786614797239e-07,
+      "loss": 0.4327,
+      "step": 5607
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.966073095962808e-07,
+      "loss": 0.3895,
+      "step": 5608
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9513862689155027e-07,
+      "loss": 0.3847,
+      "step": 5609
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9367261377373676e-07,
+      "loss": 0.4155,
+      "step": 5610
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.922092706503089e-07,
+      "loss": 0.3651,
+      "step": 5611
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.907485979279868e-07,
+      "loss": 0.433,
+      "step": 5612
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8929059601275463e-07,
+      "loss": 0.4143,
+      "step": 5613
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.878352653098494e-07,
+      "loss": 0.3717,
+      "step": 5614
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8638260622376855e-07,
+      "loss": 0.4525,
+      "step": 5615
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8493261915826585e-07,
+      "loss": 0.4063,
+      "step": 5616
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8348530451635114e-07,
+      "loss": 0.4711,
+      "step": 5617
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8204066270029483e-07,
+      "loss": 0.3723,
+      "step": 5618
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.805986941116213e-07,
+      "loss": 0.4287,
+      "step": 5619
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.791593991511111e-07,
+      "loss": 0.3929,
+      "step": 5620
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7772277821880866e-07,
+      "loss": 0.3705,
+      "step": 5621
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7628883171400455e-07,
+      "loss": 0.3912,
+      "step": 5622
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.748575600352555e-07,
+      "loss": 0.4312,
+      "step": 5623
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7342896358037006e-07,
+      "loss": 0.3771,
+      "step": 5624
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7200304274641276e-07,
+      "loss": 0.4307,
+      "step": 5625
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.705797979297077e-07,
+      "loss": 0.4255,
+      "step": 5626
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.691592295258317e-07,
+      "loss": 0.4383,
+      "step": 5627
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.677413379296202e-07,
+      "loss": 0.3633,
+      "step": 5628
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.663261235351634e-07,
+      "loss": 0.4222,
+      "step": 5629
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6491358673580777e-07,
+      "loss": 0.4086,
+      "step": 5630
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.635037279241549e-07,
+      "loss": 0.3531,
+      "step": 5631
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6209654749206236e-07,
+      "loss": 0.4133,
+      "step": 5632
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.606920458306462e-07,
+      "loss": 0.3885,
+      "step": 5633
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5929022333027307e-07,
+      "loss": 0.3543,
+      "step": 5634
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5789108038056797e-07,
+      "loss": 0.4083,
+      "step": 5635
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.564946173704109e-07,
+      "loss": 0.4197,
+      "step": 5636
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5510083468793364e-07,
+      "loss": 0.3833,
+      "step": 5637
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5370973272052857e-07,
+      "loss": 0.3896,
+      "step": 5638
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5232131185484075e-07,
+      "loss": 0.4025,
+      "step": 5639
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5093557247676824e-07,
+      "loss": 0.4641,
+      "step": 5640
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4955251497146623e-07,
+      "loss": 0.3816,
+      "step": 5641
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.481721397233406e-07,
+      "loss": 0.3847,
+      "step": 5642
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.467944471160578e-07,
+      "loss": 0.3489,
+      "step": 5643
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.454194375325337e-07,
+      "loss": 0.4111,
+      "step": 5644
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.440471113549404e-07,
+      "loss": 0.4197,
+      "step": 5645
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.42677468964705e-07,
+      "loss": 0.4074,
+      "step": 5646
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.413105107425052e-07,
+      "loss": 0.3802,
+      "step": 5647
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3994623706827825e-07,
+      "loss": 0.3837,
+      "step": 5648
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3858464832121075e-07,
+      "loss": 0.3826,
+      "step": 5649
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3722574487974115e-07,
+      "loss": 0.4106,
+      "step": 5650
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.358695271215695e-07,
+      "loss": 0.4606,
+      "step": 5651
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.345159954236421e-07,
+      "loss": 0.4249,
+      "step": 5652
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3316515016216244e-07,
+      "loss": 0.3865,
+      "step": 5653
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3181699171258463e-07,
+      "loss": 0.4029,
+      "step": 5654
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3047152044961893e-07,
+      "loss": 0.3858,
+      "step": 5655
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.291287367472251e-07,
+      "loss": 0.3626,
+      "step": 5656
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.277886409786202e-07,
+      "loss": 0.4052,
+      "step": 5657
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.264512335162728e-07,
+      "loss": 0.3641,
+      "step": 5658
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.251165147319013e-07,
+      "loss": 0.4079,
+      "step": 5659
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2378448499647996e-07,
+      "loss": 0.4605,
+      "step": 5660
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2245514468023485e-07,
+      "loss": 0.3692,
+      "step": 5661
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.211284941526438e-07,
+      "loss": 0.3994,
+      "step": 5662
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1980453378243845e-07,
+      "loss": 0.4057,
+      "step": 5663
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.184832639376023e-07,
+      "loss": 0.4078,
+      "step": 5664
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1716468498537043e-07,
+      "loss": 0.4142,
+      "step": 5665
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1584879729222974e-07,
+      "loss": 0.4626,
+      "step": 5666
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.145356012239187e-07,
+      "loss": 0.373,
+      "step": 5667
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.132250971454309e-07,
+      "loss": 0.3525,
+      "step": 5668
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.119172854210084e-07,
+      "loss": 0.4347,
+      "step": 5669
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1061216641414704e-07,
+      "loss": 0.347,
+      "step": 5670
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.093097404875911e-07,
+      "loss": 0.3799,
+      "step": 5671
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.080100080033388e-07,
+      "loss": 0.3966,
+      "step": 5672
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0671296932264114e-07,
+      "loss": 0.3505,
+      "step": 5673
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.054186248059976e-07,
+      "loss": 0.3961,
+      "step": 5674
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0412697481315925e-07,
+      "loss": 0.3308,
+      "step": 5675
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.028380197031311e-07,
+      "loss": 0.3914,
+      "step": 5676
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.015517598341633e-07,
+      "loss": 0.389,
+      "step": 5677
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.002681955637632e-07,
+      "loss": 0.3816,
+      "step": 5678
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.989873272486865e-07,
+      "loss": 0.4236,
+      "step": 5679
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9770915524493624e-07,
+      "loss": 0.383,
+      "step": 5680
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9643367990777275e-07,
+      "loss": 0.4173,
+      "step": 5681
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.951609015917012e-07,
+      "loss": 0.4241,
+      "step": 5682
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.938908206504798e-07,
+      "loss": 0.3813,
+      "step": 5683
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.926234374371162e-07,
+      "loss": 0.3708,
+      "step": 5684
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.913587523038686e-07,
+      "loss": 0.3306,
+      "step": 5685
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9009676560224355e-07,
+      "loss": 0.4575,
+      "step": 5686
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.888374776830016e-07,
+      "loss": 0.4067,
+      "step": 5687
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8758088889615067e-07,
+      "loss": 0.4285,
+      "step": 5688
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.863269995909468e-07,
+      "loss": 0.3707,
+      "step": 5689
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8507581011589905e-07,
+      "loss": 0.4725,
+      "step": 5690
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8382732081876363e-07,
+      "loss": 0.3907,
+      "step": 5691
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8258153204654634e-07,
+      "loss": 0.4132,
+      "step": 5692
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.813384441455047e-07,
+      "loss": 0.4422,
+      "step": 5693
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.800980574611445e-07,
+      "loss": 0.3769,
+      "step": 5694
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7886037233821996e-07,
+      "loss": 0.4075,
+      "step": 5695
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.776253891207337e-07,
+      "loss": 0.3667,
+      "step": 5696
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7639310815193886e-07,
+      "loss": 0.4057,
+      "step": 5697
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.751635297743382e-07,
+      "loss": 0.3605,
+      "step": 5698
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7393665432968064e-07,
+      "loss": 0.3626,
+      "step": 5699
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.727124821589666e-07,
+      "loss": 0.3838,
+      "step": 5700
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.71491013602444e-07,
+      "loss": 0.3721,
+      "step": 5701
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7027224899960793e-07,
+      "loss": 0.4059,
+      "step": 5702
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6905618868920624e-07,
+      "loss": 0.4477,
+      "step": 5703
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6784283300922864e-07,
+      "loss": 0.3718,
+      "step": 5704
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.666321822969187e-07,
+      "loss": 0.4201,
+      "step": 5705
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6542423688876607e-07,
+      "loss": 0.391,
+      "step": 5706
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.642189971205089e-07,
+      "loss": 0.4049,
+      "step": 5707
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.630164633271315e-07,
+      "loss": 0.3977,
+      "step": 5708
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6181663584286975e-07,
+      "loss": 0.3233,
+      "step": 5709
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.606195150012025e-07,
+      "loss": 0.4263,
+      "step": 5710
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.594251011348603e-07,
+      "loss": 0.4367,
+      "step": 5711
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.58233394575822e-07,
+      "loss": 0.4506,
+      "step": 5712
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5704439565530925e-07,
+      "loss": 0.4244,
+      "step": 5713
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5585810470379446e-07,
+      "loss": 0.4066,
+      "step": 5714
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.546745220509972e-07,
+      "loss": 0.4528,
+      "step": 5715
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.534936480258832e-07,
+      "loss": 0.4231,
+      "step": 5716
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5231548295666653e-07,
+      "loss": 0.4393,
+      "step": 5717
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5114002717080864e-07,
+      "loss": 0.4776,
+      "step": 5718
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.499672809950171e-07,
+      "loss": 0.3618,
+      "step": 5719
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4879724475524445e-07,
+      "loss": 0.4022,
+      "step": 5720
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.47629918776694e-07,
+      "loss": 0.4064,
+      "step": 5721
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4646530338381267e-07,
+      "loss": 0.4327,
+      "step": 5722
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4530339890029395e-07,
+      "loss": 0.3992,
+      "step": 5723
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.441442056490828e-07,
+      "loss": 0.375,
+      "step": 5724
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.429877239523637e-07,
+      "loss": 0.4073,
+      "step": 5725
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.418339541315706e-07,
+      "loss": 0.3945,
+      "step": 5726
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4068289650738484e-07,
+      "loss": 0.4575,
+      "step": 5727
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.395345513997327e-07,
+      "loss": 0.4405,
+      "step": 5728
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.383889191277855e-07,
+      "loss": 0.3939,
+      "step": 5729
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.372460000099641e-07,
+      "loss": 0.4091,
+      "step": 5730
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3610579436392999e-07,
+      "loss": 0.3671,
+      "step": 5731
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3496830250659519e-07,
+      "loss": 0.4154,
+      "step": 5732
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.338335247541146e-07,
+      "loss": 0.431,
+      "step": 5733
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3270146142189032e-07,
+      "loss": 0.4292,
+      "step": 5734
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3157211282456848e-07,
+      "loss": 0.3873,
+      "step": 5735
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3044547927604244e-07,
+      "loss": 0.3779,
+      "step": 5736
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2932156108945059e-07,
+      "loss": 0.4099,
+      "step": 5737
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2820035857717526e-07,
+      "loss": 0.3954,
+      "step": 5738
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.270818720508461e-07,
+      "loss": 0.3612,
+      "step": 5739
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2596610182133328e-07,
+      "loss": 0.385,
+      "step": 5740
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2485304819875985e-07,
+      "loss": 0.4487,
+      "step": 5741
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2374271149248616e-07,
+      "loss": 0.4626,
+      "step": 5742
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.22635092011122e-07,
+      "loss": 0.3823,
+      "step": 5743
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2153019006251998e-07,
+      "loss": 0.4376,
+      "step": 5744
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2042800595377777e-07,
+      "loss": 0.4435,
+      "step": 5745
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1932853999123704e-07,
+      "loss": 0.4556,
+      "step": 5746
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1823179248048553e-07,
+      "loss": 0.4228,
+      "step": 5747
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.171377637263561e-07,
+      "loss": 0.3742,
+      "step": 5748
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1604645403292322e-07,
+      "loss": 0.3902,
+      "step": 5749
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1495786370350546e-07,
+      "loss": 0.3656,
+      "step": 5750
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.138719930406685e-07,
+      "loss": 0.428,
+      "step": 5751
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1278884234621987e-07,
+      "loss": 0.3916,
+      "step": 5752
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1170841192121206e-07,
+      "loss": 0.428,
+      "step": 5753
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.106307020659426e-07,
+      "loss": 0.437,
+      "step": 5754
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.095557130799486e-07,
+      "loss": 0.3802,
+      "step": 5755
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0848344526201548e-07,
+      "loss": 0.362,
+      "step": 5756
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.074138989101715e-07,
+      "loss": 0.4135,
+      "step": 5757
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0634707432168667e-07,
+      "loss": 0.4492,
+      "step": 5758
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0528297179307377e-07,
+      "loss": 0.427,
+      "step": 5759
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0422159162009404e-07,
+      "loss": 0.4367,
+      "step": 5760
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0316293409774702e-07,
+      "loss": 0.4142,
+      "step": 5761
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0210699952027847e-07,
+      "loss": 0.3458,
+      "step": 5762
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.010537881811736e-07,
+      "loss": 0.5382,
+      "step": 5763
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0000330037316605e-07,
+      "loss": 0.3633,
+      "step": 5764
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9895553638822785e-07,
+      "loss": 0.4496,
+      "step": 5765
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.97910496517576e-07,
+      "loss": 0.4205,
+      "step": 5766
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.968681810516726e-07,
+      "loss": 0.4022,
+      "step": 5767
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9582859028021705e-07,
+      "loss": 0.4316,
+      "step": 5768
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.94791724492156e-07,
+      "loss": 0.4145,
+      "step": 5769
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.937575839756767e-07,
+      "loss": 0.4156,
+      "step": 5770
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9272616901820806e-07,
+      "loss": 0.3896,
+      "step": 5771
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.916974799064242e-07,
+      "loss": 0.421,
+      "step": 5772
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9067151692624185e-07,
+      "loss": 0.4163,
+      "step": 5773
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8964828036281636e-07,
+      "loss": 0.4147,
+      "step": 5774
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8862777050054793e-07,
+      "loss": 0.409,
+      "step": 5775
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8760998762307747e-07,
+      "loss": 0.3813,
+      "step": 5776
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8659493201329093e-07,
+      "loss": 0.4179,
+      "step": 5777
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.855826039533104e-07,
+      "loss": 0.4421,
+      "step": 5778
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.845730037245075e-07,
+      "loss": 0.4158,
+      "step": 5779
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8356613160749104e-07,
+      "loss": 0.3474,
+      "step": 5780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8256198788210942e-07,
+      "loss": 0.3693,
+      "step": 5781
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8156057282745942e-07,
+      "loss": 0.3906,
+      "step": 5782
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.805618867218717e-07,
+      "loss": 0.3739,
+      "step": 5783
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.795659298429253e-07,
+      "loss": 0.3865,
+      "step": 5784
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7857270246743775e-07,
+      "loss": 0.4088,
+      "step": 5785
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7758220487146483e-07,
+      "loss": 0.3863,
+      "step": 5786
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7659443733030968e-07,
+      "loss": 0.3839,
+      "step": 5787
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7560940011851268e-07,
+      "loss": 0.3841,
+      "step": 5788
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7462709350985596e-07,
+      "loss": 0.4215,
+      "step": 5789
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7364751777736334e-07,
+      "loss": 0.3817,
+      "step": 5790
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7267067319329922e-07,
+      "loss": 0.4016,
+      "step": 5791
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.716965600291687e-07,
+      "loss": 0.3712,
+      "step": 5792
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7072517855571846e-07,
+      "loss": 0.442,
+      "step": 5793
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6975652904293593e-07,
+      "loss": 0.4764,
+      "step": 5794
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6879061176004686e-07,
+      "loss": 0.4032,
+      "step": 5795
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.678274269755209e-07,
+      "loss": 0.4316,
+      "step": 5796
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6686697495706617e-07,
+      "loss": 0.3613,
+      "step": 5797
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6590925597163354e-07,
+      "loss": 0.3756,
+      "step": 5798
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6495427028541123e-07,
+      "loss": 0.4488,
+      "step": 5799
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.640020181638291e-07,
+      "loss": 0.4483,
+      "step": 5800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6305249987155657e-07,
+      "loss": 0.4335,
+      "step": 5801
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6210571567250478e-07,
+      "loss": 0.4292,
+      "step": 5802
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6116166582982428e-07,
+      "loss": 0.379,
+      "step": 5803
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6022035060590634e-07,
+      "loss": 0.3466,
+      "step": 5804
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5928177026237835e-07,
+      "loss": 0.3546,
+      "step": 5805
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5834592506011382e-07,
+      "loss": 0.4171,
+      "step": 5806
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5741281525922024e-07,
+      "loss": 0.3376,
+      "step": 5807
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.56482441119048e-07,
+      "loss": 0.3776,
+      "step": 5808
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5555480289818793e-07,
+      "loss": 0.3854,
+      "step": 5809
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5462990085446604e-07,
+      "loss": 0.3596,
+      "step": 5810
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.537077352449534e-07,
+      "loss": 0.3941,
+      "step": 5811
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.527883063259572e-07,
+      "loss": 0.3523,
+      "step": 5812
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.51871614353023e-07,
+      "loss": 0.3977,
+      "step": 5813
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5095765958093922e-07,
+      "loss": 0.4063,
+      "step": 5814
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5004644226373044e-07,
+      "loss": 0.4226,
+      "step": 5815
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4913796265466185e-07,
+      "loss": 0.4468,
+      "step": 5816
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4823222100623814e-07,
+      "loss": 0.4233,
+      "step": 5817
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4732921757020236e-07,
+      "loss": 0.4109,
+      "step": 5818
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4642895259753376e-07,
+      "loss": 0.382,
+      "step": 5819
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.455314263384544e-07,
+      "loss": 0.4067,
+      "step": 5820
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4463663904242697e-07,
+      "loss": 0.3802,
+      "step": 5821
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4374459095814585e-07,
+      "loss": 0.4004,
+      "step": 5822
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.428552823335494e-07,
+      "loss": 0.4111,
+      "step": 5823
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.419687134158132e-07,
+      "loss": 0.418,
+      "step": 5824
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4108488445135016e-07,
+      "loss": 0.4057,
+      "step": 5825
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4020379568581378e-07,
+      "loss": 0.3919,
+      "step": 5826
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3932544736409591e-07,
+      "loss": 0.4468,
+      "step": 5827
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3844983973032467e-07,
+      "loss": 0.4071,
+      "step": 5828
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3757697302786754e-07,
+      "loss": 0.4393,
+      "step": 5829
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.367068474993294e-07,
+      "loss": 0.351,
+      "step": 5830
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3583946338655562e-07,
+      "loss": 0.3932,
+      "step": 5831
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3497482093062675e-07,
+      "loss": 0.3836,
+      "step": 5832
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3411292037186274e-07,
+      "loss": 0.3596,
+      "step": 5833
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3325376194982197e-07,
+      "loss": 0.4357,
+      "step": 5834
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.323973459033001e-07,
+      "loss": 0.4045,
+      "step": 5835
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3154367247032895e-07,
+      "loss": 0.4248,
+      "step": 5836
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3069274188818205e-07,
+      "loss": 0.4017,
+      "step": 5837
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2984455439336462e-07,
+      "loss": 0.4452,
+      "step": 5838
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.289991102216259e-07,
+      "loss": 0.4339,
+      "step": 5839
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2815640960794906e-07,
+      "loss": 0.4116,
+      "step": 5840
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2731645278655448e-07,
+      "loss": 0.4008,
+      "step": 5841
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2647923999090096e-07,
+      "loss": 0.3686,
+      "step": 5842
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.256447714536857e-07,
+      "loss": 0.4187,
+      "step": 5843
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2481304740683986e-07,
+      "loss": 0.356,
+      "step": 5844
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2398406808153517e-07,
+      "loss": 0.444,
+      "step": 5845
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2315783370817958e-07,
+      "loss": 0.4091,
+      "step": 5846
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2233434451641713e-07,
+      "loss": 0.4011,
+      "step": 5847
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2151360073512808e-07,
+      "loss": 0.4229,
+      "step": 5848
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.206956025924333e-07,
+      "loss": 0.3989,
+      "step": 5849
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1988035031568646e-07,
+      "loss": 0.4489,
+      "step": 5850
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1906784413147965e-07,
+      "loss": 0.4467,
+      "step": 5851
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1825808426564445e-07,
+      "loss": 0.4616,
+      "step": 5852
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1745107094324415e-07,
+      "loss": 0.4075,
+      "step": 5853
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1664680438858044e-07,
+      "loss": 0.4157,
+      "step": 5854
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1584528482519342e-07,
+      "loss": 0.4176,
+      "step": 5855
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.150465124758593e-07,
+      "loss": 0.3903,
+      "step": 5856
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1425048756258716e-07,
+      "loss": 0.4024,
+      "step": 5857
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1345721030662782e-07,
+      "loss": 0.3666,
+      "step": 5858
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1266668092846488e-07,
+      "loss": 0.4133,
+      "step": 5859
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1187889964781818e-07,
+      "loss": 0.3734,
+      "step": 5860
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1109386668364586e-07,
+      "loss": 0.3864,
+      "step": 5861
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1031158225413896e-07,
+      "loss": 0.4042,
+      "step": 5862
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0953204657672912e-07,
+      "loss": 0.4243,
+      "step": 5863
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0875525986807966e-07,
+      "loss": 0.4221,
+      "step": 5864
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0798122234409237e-07,
+      "loss": 0.4225,
+      "step": 5865
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0720993421990511e-07,
+      "loss": 0.4054,
+      "step": 5866
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0644139570988865e-07,
+      "loss": 0.3398,
+      "step": 5867
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0567560702765212e-07,
+      "loss": 0.4071,
+      "step": 5868
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.049125683860408e-07,
+      "loss": 0.3823,
+      "step": 5869
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0415227999713285e-07,
+      "loss": 0.4328,
+      "step": 5870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0339474207224587e-07,
+      "loss": 0.3886,
+      "step": 5871
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0263995482192924e-07,
+      "loss": 0.3939,
+      "step": 5872
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0188791845596957e-07,
+      "loss": 0.4715,
+      "step": 5873
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0113863318338968e-07,
+      "loss": 0.3692,
+      "step": 5874
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.003920992124463e-07,
+      "loss": 0.3888,
+      "step": 5875
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.964831675063235e-08,
+      "loss": 0.4195,
+      "step": 5876
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.89072860046747e-08,
+      "loss": 0.427,
+      "step": 5877
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.816900718053746e-08,
+      "loss": 0.376,
+      "step": 5878
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.743348048341871e-08,
+      "loss": 0.4537,
+      "step": 5879
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.67007061177505e-08,
+      "loss": 0.4226,
+      "step": 5880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.597068428720102e-08,
+      "loss": 0.3784,
+      "step": 5881
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.524341519467572e-08,
+      "loss": 0.3942,
+      "step": 5882
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.451889904230959e-08,
+      "loss": 0.4174,
+      "step": 5883
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.379713603147823e-08,
+      "loss": 0.3942,
+      "step": 5884
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.307812636278779e-08,
+      "loss": 0.383,
+      "step": 5885
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.236187023608068e-08,
+      "loss": 0.3963,
+      "step": 5886
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.16483678504343e-08,
+      "loss": 0.3525,
+      "step": 5887
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.093761940416112e-08,
+      "loss": 0.4477,
+      "step": 5888
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.022962509480649e-08,
+      "loss": 0.3357,
+      "step": 5889
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.952438511915073e-08,
+      "loss": 0.3361,
+      "step": 5890
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.882189967321042e-08,
+      "loss": 0.4193,
+      "step": 5891
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.812216895223492e-08,
+      "loss": 0.3691,
+      "step": 5892
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.742519315070752e-08,
+      "loss": 0.3264,
+      "step": 5893
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.673097246234664e-08,
+      "loss": 0.3289,
+      "step": 5894
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.603950708010567e-08,
+      "loss": 0.3725,
+      "step": 5895
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.53507971961709e-08,
+      "loss": 0.3425,
+      "step": 5896
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.466484300196364e-08,
+      "loss": 0.3571,
+      "step": 5897
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.398164468813696e-08,
+      "loss": 0.4261,
+      "step": 5898
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.330120244458228e-08,
+      "loss": 0.3791,
+      "step": 5899
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.262351646042055e-08,
+      "loss": 0.391,
+      "step": 5900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.194858692400997e-08,
+      "loss": 0.4324,
+      "step": 5901
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.127641402294162e-08,
+      "loss": 0.5148,
+      "step": 5902
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.060699794403937e-08,
+      "loss": 0.4259,
+      "step": 5903
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.994033887335994e-08,
+      "loss": 0.4298,
+      "step": 5904
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.927643699619736e-08,
+      "loss": 0.3931,
+      "step": 5905
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.861529249707733e-08,
+      "loss": 0.379,
+      "step": 5906
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.795690555975954e-08,
+      "loss": 0.3546,
+      "step": 5907
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.730127636723539e-08,
+      "loss": 0.4248,
+      "step": 5908
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.664840510173133e-08,
+      "loss": 0.3496,
+      "step": 5909
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.599829194470887e-08,
+      "loss": 0.3935,
+      "step": 5910
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.535093707685904e-08,
+      "loss": 0.4338,
+      "step": 5911
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.470634067811012e-08,
+      "loss": 0.3868,
+      "step": 5912
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.406450292762101e-08,
+      "loss": 0.4004,
+      "step": 5913
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.342542400378572e-08,
+      "loss": 0.3529,
+      "step": 5914
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.27891040842299e-08,
+      "loss": 0.4,
+      "step": 5915
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.215554334581431e-08,
+      "loss": 0.4522,
+      "step": 5916
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.152474196463033e-08,
+      "loss": 0.3909,
+      "step": 5917
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.089670011600325e-08,
+      "loss": 0.4494,
+      "step": 5918
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.027141797449344e-08,
+      "loss": 0.4252,
+      "step": 5919
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.964889571389188e-08,
+      "loss": 0.387,
+      "step": 5920
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.90291335072224e-08,
+      "loss": 0.436,
+      "step": 5921
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.841213152674387e-08,
+      "loss": 0.4186,
+      "step": 5922
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.779788994394465e-08,
+      "loss": 0.38,
+      "step": 5923
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.718640892954931e-08,
+      "loss": 0.3918,
+      "step": 5924
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.65776886535141e-08,
+      "loss": 0.4036,
+      "step": 5925
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.597172928502594e-08,
+      "loss": 0.4719,
+      "step": 5926
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.536853099250672e-08,
+      "loss": 0.4189,
+      "step": 5927
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.476809394361017e-08,
+      "loss": 0.4572,
+      "step": 5928
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.417041830522163e-08,
+      "loss": 0.3647,
+      "step": 5929
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.35755042434616e-08,
+      "loss": 0.3823,
+      "step": 5930
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.298335192368e-08,
+      "loss": 0.3574,
+      "step": 5931
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.239396151045963e-08,
+      "loss": 0.4886,
+      "step": 5932
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.180733316761833e-08,
+      "loss": 0.3962,
+      "step": 5933
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.122346705820237e-08,
+      "loss": 0.4329,
+      "step": 5934
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.064236334449303e-08,
+      "loss": 0.3875,
+      "step": 5935
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.006402218800333e-08,
+      "loss": 0.4181,
+      "step": 5936
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.948844374947804e-08,
+      "loss": 0.3886,
+      "step": 5937
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8915628188893625e-08,
+      "loss": 0.4377,
+      "step": 5938
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8345575665459397e-08,
+      "loss": 0.461,
+      "step": 5939
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7778286337616396e-08,
+      "loss": 0.3866,
+      "step": 5940
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.721376036303738e-08,
+      "loss": 0.3807,
+      "step": 5941
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.665199789862907e-08,
+      "loss": 0.4466,
+      "step": 5942
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6092999100526566e-08,
+      "loss": 0.4135,
+      "step": 5943
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.553676412409892e-08,
+      "loss": 0.4437,
+      "step": 5944
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.49832931239469e-08,
+      "loss": 0.4112,
+      "step": 5945
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.443258625390413e-08,
+      "loss": 0.3344,
+      "step": 5946
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.388464366703261e-08,
+      "loss": 0.3605,
+      "step": 5947
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3339465515629405e-08,
+      "loss": 0.3699,
+      "step": 5948
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.27970519512222e-08,
+      "loss": 0.4271,
+      "step": 5949
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.225740312457039e-08,
+      "loss": 0.3914,
+      "step": 5950
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.17205191856629e-08,
+      "loss": 0.38,
+      "step": 5951
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.118640028372368e-08,
+      "loss": 0.3502,
+      "step": 5952
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.065504656720621e-08,
+      "loss": 0.4127,
+      "step": 5953
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.012645818379347e-08,
+      "loss": 0.427,
+      "step": 5954
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9600635280405706e-08,
+      "loss": 0.3639,
+      "step": 5955
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.907757800318713e-08,
+      "loss": 0.4402,
+      "step": 5956
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.855728649751923e-08,
+      "loss": 0.4378,
+      "step": 5957
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8039760908011876e-08,
+      "loss": 0.3677,
+      "step": 5958
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7525001378505574e-08,
+      "loss": 0.3797,
+      "step": 5959
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.701300805207476e-08,
+      "loss": 0.4268,
+      "step": 5960
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.650378107102338e-08,
+      "loss": 0.3946,
+      "step": 5961
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5997320576884886e-08,
+      "loss": 0.3954,
+      "step": 5962
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5493626710428897e-08,
+      "loss": 0.4431,
+      "step": 5963
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.499269961164898e-08,
+      "loss": 0.3924,
+      "step": 5964
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.449453941977488e-08,
+      "loss": 0.4597,
+      "step": 5965
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.399914627326696e-08,
+      "loss": 0.372,
+      "step": 5966
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.350652030981395e-08,
+      "loss": 0.3586,
+      "step": 5967
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3016661666338554e-08,
+      "loss": 0.4071,
+      "step": 5968
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.252957047898965e-08,
+      "loss": 0.3856,
+      "step": 5969
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.2045246883153366e-08,
+      "loss": 0.4453,
+      "step": 5970
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1563691013440935e-08,
+      "loss": 0.3944,
+      "step": 5971
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.108490300369861e-08,
+      "loss": 0.4509,
+      "step": 5972
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0608882986998834e-08,
+      "loss": 0.4103,
+      "step": 5973
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0135631095649105e-08,
+      "loss": 0.4273,
+      "step": 5974
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9665147461184216e-08,
+      "loss": 0.3847,
+      "step": 5975
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.919743221437178e-08,
+      "loss": 0.3686,
+      "step": 5976
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8732485485210024e-08,
+      "loss": 0.4524,
+      "step": 5977
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.827030740292448e-08,
+      "loss": 0.4067,
+      "step": 5978
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.781089809597571e-08,
+      "loss": 0.4156,
+      "step": 5979
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7354257692051584e-08,
+      "loss": 0.4195,
+      "step": 5980
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.690038631807169e-08,
+      "loss": 0.441,
+      "step": 5981
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6449284100185114e-08,
+      "loss": 0.4208,
+      "step": 5982
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.600095116377156e-08,
+      "loss": 0.3687,
+      "step": 5983
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.555538763344135e-08,
+      "loss": 0.3416,
+      "step": 5984
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.511259363303543e-08,
+      "loss": 0.441,
+      "step": 5985
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4672569285624235e-08,
+      "loss": 0.4292,
+      "step": 5986
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.423531471350883e-08,
+      "loss": 0.4617,
+      "step": 5987
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.380083003822088e-08,
+      "loss": 0.3624,
+      "step": 5988
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.336911538051935e-08,
+      "loss": 0.4737,
+      "step": 5989
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2940170860398244e-08,
+      "loss": 0.463,
+      "step": 5990
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2513996597076656e-08,
+      "loss": 0.3536,
+      "step": 5991
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.209059270900872e-08,
+      "loss": 0.4127,
+      "step": 5992
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1669959313872514e-08,
+      "loss": 0.4395,
+      "step": 5993
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.125209652858119e-08,
+      "loss": 0.4123,
+      "step": 5994
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.08370044692774e-08,
+      "loss": 0.4401,
+      "step": 5995
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0424683251328854e-08,
+      "loss": 0.4054,
+      "step": 5996
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0015132989340554e-08,
+      "loss": 0.4118,
+      "step": 5997
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9608353797139222e-08,
+      "loss": 0.4135,
+      "step": 5998
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9204345787789966e-08,
+      "loss": 0.3677,
+      "step": 5999
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.880310907357964e-08,
+      "loss": 0.3918,
+      "step": 6000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8404643766031247e-08,
+      "loss": 0.4281,
+      "step": 6001
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8008949975891765e-08,
+      "loss": 0.3696,
+      "step": 6002
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7616027813143208e-08,
+      "loss": 0.3599,
+      "step": 6003
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7225877386994893e-08,
+      "loss": 0.3902,
+      "step": 6004
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6838498805883407e-08,
+      "loss": 0.3328,
+      "step": 6005
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.64538921774804e-08,
+      "loss": 0.3846,
+      "step": 6006
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6072057608681478e-08,
+      "loss": 0.3898,
+      "step": 6007
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5692995205613968e-08,
+      "loss": 0.4167,
+      "step": 6008
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5316705073636928e-08,
+      "loss": 0.3999,
+      "step": 6009
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4943187317335584e-08,
+      "loss": 0.4677,
+      "step": 6010
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.457244204052578e-08,
+      "loss": 0.3797,
+      "step": 6011
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.420446934625398e-08,
+      "loss": 0.426,
+      "step": 6012
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3839269336793924e-08,
+      "loss": 0.4295,
+      "step": 6013
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3476842113651087e-08,
+      "loss": 0.396,
+      "step": 6014
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3117187777557116e-08,
+      "loss": 0.4483,
+      "step": 6015
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2760306428476485e-08,
+      "loss": 0.4349,
+      "step": 6016
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2406198165599857e-08,
+      "loss": 0.3904,
+      "step": 6017
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.20548630873485e-08,
+      "loss": 0.385,
+      "step": 6018
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1706301291374297e-08,
+      "loss": 0.4051,
+      "step": 6019
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1360512874555316e-08,
+      "loss": 0.4014,
+      "step": 6020
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1017497933002452e-08,
+      "loss": 0.3688,
+      "step": 6021
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0677256562051663e-08,
+      "loss": 0.4025,
+      "step": 6022
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0339788856270636e-08,
+      "loss": 0.4004,
+      "step": 6023
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.000509490945657e-08,
+      "loss": 0.3887,
+      "step": 6024
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9673174814633933e-08,
+      "loss": 0.392,
+      "step": 6025
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9344028664056715e-08,
+      "loss": 0.397,
+      "step": 6026
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9017656549208396e-08,
+      "loss": 0.3866,
+      "step": 6027
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8694058560800864e-08,
+      "loss": 0.3759,
+      "step": 6028
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.837323478877662e-08,
+      "loss": 0.3333,
+      "step": 6029
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8055185322304326e-08,
+      "loss": 0.446,
+      "step": 6030
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7739910249784388e-08,
+      "loss": 0.3361,
+      "step": 6031
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7427409658843375e-08,
+      "loss": 0.342,
+      "step": 6032
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.711768363633959e-08,
+      "loss": 0.4116,
+      "step": 6033
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.68107322683575e-08,
+      "loss": 0.3941,
+      "step": 6034
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6506555640212195e-08,
+      "loss": 0.3829,
+      "step": 6035
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6205153836446052e-08,
+      "loss": 0.3976,
+      "step": 6036
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5906526940832058e-08,
+      "loss": 0.4023,
+      "step": 6037
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.561067503636937e-08,
+      "loss": 0.4106,
+      "step": 6038
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5317598205288887e-08,
+      "loss": 0.4713,
+      "step": 6039
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5027296529049885e-08,
+      "loss": 0.3684,
+      "step": 6040
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4739770088336713e-08,
+      "loss": 0.3634,
+      "step": 6041
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4455018963065448e-08,
+      "loss": 0.3681,
+      "step": 6042
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4173043232380557e-08,
+      "loss": 0.3837,
+      "step": 6043
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3893842974654903e-08,
+      "loss": 0.3575,
+      "step": 6044
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3617418267489745e-08,
+      "loss": 0.4061,
+      "step": 6045
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.334376918771474e-08,
+      "loss": 0.4373,
+      "step": 6046
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3072895811387932e-08,
+      "loss": 0.4151,
+      "step": 6047
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2804798213795766e-08,
+      "loss": 0.4306,
+      "step": 6048
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2539476469455302e-08,
+      "loss": 0.3654,
+      "step": 6049
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2276930652108665e-08,
+      "loss": 0.432,
+      "step": 6050
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2017160834729702e-08,
+      "loss": 0.422,
+      "step": 6051
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1760167089517326e-08,
+      "loss": 0.3607,
+      "step": 6052
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.150594948790218e-08,
+      "loss": 0.3991,
+      "step": 6053
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1254508100541073e-08,
+      "loss": 0.3948,
+      "step": 6054
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1005842997321436e-08,
+      "loss": 0.4191,
+      "step": 6055
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0759954247354654e-08,
+      "loss": 0.3536,
+      "step": 6056
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0516841918986053e-08,
+      "loss": 0.4509,
+      "step": 6057
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.027650607978492e-08,
+      "loss": 0.4022,
+      "step": 6058
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0038946796551153e-08,
+      "loss": 0.3587,
+      "step": 6059
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.804164135313044e-09,
+      "loss": 0.3269,
+      "step": 6060
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.572158161325062e-09,
+      "loss": 0.4089,
+      "step": 6061
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.342928939071183e-09,
+      "loss": 0.4398,
+      "step": 6062
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.116476532263775e-09,
+      "loss": 0.3588,
+      "step": 6063
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.892801003844709e-09,
+      "loss": 0.4244,
+      "step": 6064
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.671902415982036e-09,
+      "loss": 0.4856,
+      "step": 6065
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.45378083007109e-09,
+      "loss": 0.4004,
+      "step": 6066
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.238436306738929e-09,
+      "loss": 0.3932,
+      "step": 6067
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.025868905836565e-09,
+      "loss": 0.405,
+      "step": 6068
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.816078686446737e-09,
+      "loss": 0.5022,
+      "step": 6069
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.60906570687836e-09,
+      "loss": 0.4364,
+      "step": 6070
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.4048300246698515e-09,
+      "loss": 0.4733,
+      "step": 6071
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.203371696584693e-09,
+      "loss": 0.4151,
+      "step": 6072
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.0046907786180954e-09,
+      "loss": 0.4421,
+      "step": 6073
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.808787325991439e-09,
+      "loss": 0.4498,
+      "step": 6074
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.6156613931533945e-09,
+      "loss": 0.4692,
+      "step": 6075
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.4253130337821325e-09,
+      "loss": 0.3839,
+      "step": 6076
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.237742300784222e-09,
+      "loss": 0.3721,
+      "step": 6077
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.052949246292405e-09,
+      "loss": 0.41,
+      "step": 6078
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.8709339216689306e-09,
+      "loss": 0.3397,
+      "step": 6079
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.691696377502221e-09,
+      "loss": 0.4395,
+      "step": 6080
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.515236663610202e-09,
+      "loss": 0.454,
+      "step": 6081
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.341554829038087e-09,
+      "loss": 0.4117,
+      "step": 6082
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.1706509220605934e-09,
+      "loss": 0.4136,
+      "step": 6083
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.002524990178614e-09,
+      "loss": 0.3736,
+      "step": 6084
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.837177080119215e-09,
+      "loss": 0.3519,
+      "step": 6085
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.674607237841189e-09,
+      "loss": 0.3992,
+      "step": 6086
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.514815508529502e-09,
+      "loss": 0.4029,
+      "step": 6087
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3578019365964065e-09,
+      "loss": 0.4072,
+      "step": 6088
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.203566565682549e-09,
+      "loss": 0.4312,
+      "step": 6089
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0521094386569705e-09,
+      "loss": 0.4076,
+      "step": 6090
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.903430597613778e-09,
+      "loss": 0.4313,
+      "step": 6091
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.757530083878802e-09,
+      "loss": 0.4261,
+      "step": 6092
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6144079380029396e-09,
+      "loss": 0.3686,
+      "step": 6093
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4740641997665915e-09,
+      "loss": 0.4208,
+      "step": 6094
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3364989081763332e-09,
+      "loss": 0.3747,
+      "step": 6095
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2017121014682464e-09,
+      "loss": 0.3967,
+      "step": 6096
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0697038171034756e-09,
+      "loss": 0.4428,
+      "step": 6097
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.940474091773782e-09,
+      "loss": 0.4167,
+      "step": 6098
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8140229613971003e-09,
+      "loss": 0.4386,
+      "step": 6099
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6903504611186517e-09,
+      "loss": 0.4187,
+      "step": 6100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5694566253142704e-09,
+      "loss": 0.367,
+      "step": 6101
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4513414875826368e-09,
+      "loss": 0.4799,
+      "step": 6102
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3360050807541555e-09,
+      "loss": 0.4263,
+      "step": 6103
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2234474368865166e-09,
+      "loss": 0.4215,
+      "step": 6104
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1136685872624742e-09,
+      "loss": 0.3702,
+      "step": 6105
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0066685623953974e-09,
+      "loss": 0.3851,
+      "step": 6106
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.902447392023721e-09,
+      "loss": 0.376,
+      "step": 6107
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8010051051153832e-09,
+      "loss": 0.3886,
+      "step": 6108
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7023417298656086e-09,
+      "loss": 0.4372,
+      "step": 6109
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6064572936980162e-09,
+      "loss": 0.3975,
+      "step": 6110
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5133518232601785e-09,
+      "loss": 0.3631,
+      "step": 6111
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4230253444313946e-09,
+      "loss": 0.4417,
+      "step": 6112
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.335477882318248e-09,
+      "loss": 0.374,
+      "step": 6113
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2507094612523862e-09,
+      "loss": 0.4314,
+      "step": 6114
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1687201047949626e-09,
+      "loss": 0.4169,
+      "step": 6115
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0895098357333046e-09,
+      "loss": 0.3755,
+      "step": 6116
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0130786760842447e-09,
+      "loss": 0.401,
+      "step": 6117
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.39426647090791e-10,
+      "loss": 0.3898,
+      "step": 6118
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.685537692243451e-10,
+      "loss": 0.4245,
+      "step": 6119
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.004600621824843e-10,
+      "loss": 0.3838,
+      "step": 6120
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.351455448922906e-10,
+      "loss": 0.4201,
+      "step": 6121
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.726102355059105e-10,
+      "loss": 0.3603,
+      "step": 6122
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.128541514061059e-10,
+      "loss": 0.4055,
+      "step": 6123
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.558773092007031e-10,
+      "loss": 0.428,
+      "step": 6124
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.016797247259231e-10,
+      "loss": 0.3964,
+      "step": 6125
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.5026141304638225e-10,
+      "loss": 0.4172,
+      "step": 6126
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0162238845176114e-10,
+      "loss": 0.4181,
+      "step": 6127
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5576266446124553e-10,
+      "loss": 0.3603,
+      "step": 6128
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1268225382241613e-10,
+      "loss": 0.3656,
+      "step": 6129
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.7238116850791805e-10,
+      "loss": 0.369,
+      "step": 6130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3485941971990166e-10,
+      "loss": 0.3765,
+      "step": 6131
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.001170178866918e-10,
+      "loss": 0.407,
+      "step": 6132
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6815397266500831e-10,
+      "loss": 0.41,
+      "step": 6133
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3897029293774566e-10,
+      "loss": 0.4326,
+      "step": 6134
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.1256598681619324e-10,
+      "loss": 0.3877,
+      "step": 6135
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.894106164114569e-11,
+      "loss": 0.3809,
+      "step": 6136
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.809552397735175e-11,
+      "loss": 0.385,
+      "step": 6137
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.00293796190654e-11,
+      "loss": 0.3566,
+      "step": 6138
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.4742633586715146e-11,
+      "loss": 0.4257,
+      "step": 6139
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2235290130234732e-11,
+      "loss": 0.366,
+      "step": 6140
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2507352726842669e-11,
+      "loss": 0.3548,
+      "step": 6141
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.5588240777115775e-12,
+      "loss": 0.3716,
+      "step": 6142
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3897061157397418e-12,
+      "loss": 0.409,
+      "step": 6143
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.3675,
+      "step": 6144
+    },
+    {
+      "epoch": 3.0,
+      "step": 6144,
+      "total_flos": 3.991335235844309e+18,
+      "train_loss": 0.009399520199319037,
+      "train_runtime": 686.4589,
+      "train_samples_per_second": 1145.181,
+      "train_steps_per_second": 8.95
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 6144,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 3.991335235844309e+18,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}